{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 9106,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 3.5627039800772393,
      "learning_rate": 1.4598540145985402e-07,
      "loss": 1.7115,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.13746608939205,
      "learning_rate": 2.9197080291970804e-07,
      "loss": 1.9855,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.461789986681792,
      "learning_rate": 4.379562043795621e-07,
      "loss": 1.9055,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.249688529774583,
      "learning_rate": 5.839416058394161e-07,
      "loss": 2.1319,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.894729598061038,
      "learning_rate": 7.299270072992701e-07,
      "loss": 2.2653,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.128280029158089,
      "learning_rate": 8.759124087591242e-07,
      "loss": 1.7717,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.857974325866826,
      "learning_rate": 1.0218978102189781e-06,
      "loss": 2.2352,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.279510590210478,
      "learning_rate": 1.1678832116788322e-06,
      "loss": 2.1748,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.143207248683319,
      "learning_rate": 1.3138686131386864e-06,
      "loss": 2.0977,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.836348439753132,
      "learning_rate": 1.4598540145985402e-06,
      "loss": 2.1316,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.169681786772062,
      "learning_rate": 1.6058394160583942e-06,
      "loss": 1.8064,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.988943388141476,
      "learning_rate": 1.7518248175182485e-06,
      "loss": 1.8728,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3929806450185804,
      "learning_rate": 1.8978102189781023e-06,
      "loss": 2.2436,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8432284704849127,
      "learning_rate": 2.0437956204379563e-06,
      "loss": 1.9455,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.803494425471958,
      "learning_rate": 2.1897810218978103e-06,
      "loss": 1.8886,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1817557242072168,
      "learning_rate": 2.3357664233576643e-06,
      "loss": 2.1747,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1726437092871147,
      "learning_rate": 2.4817518248175183e-06,
      "loss": 2.1334,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.924586053232777,
      "learning_rate": 2.627737226277373e-06,
      "loss": 1.8445,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.763798514655918,
      "learning_rate": 2.7737226277372264e-06,
      "loss": 1.9356,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6064992808317755,
      "learning_rate": 2.9197080291970804e-06,
      "loss": 1.7732,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1626567213707286,
      "learning_rate": 3.065693430656935e-06,
      "loss": 1.9922,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8071379516264727,
      "learning_rate": 3.2116788321167884e-06,
      "loss": 1.8957,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2009291290769637,
      "learning_rate": 3.3576642335766425e-06,
      "loss": 1.8077,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.493585750362153,
      "learning_rate": 3.503649635036497e-06,
      "loss": 1.5587,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0208173010540813,
      "learning_rate": 3.6496350364963505e-06,
      "loss": 1.7271,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9588523812671719,
      "learning_rate": 3.7956204379562045e-06,
      "loss": 1.5694,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4878919197331266,
      "learning_rate": 3.9416058394160585e-06,
      "loss": 1.4266,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.556986426949064,
      "learning_rate": 4.0875912408759126e-06,
      "loss": 1.5129,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3363605297969032,
      "learning_rate": 4.233576642335767e-06,
      "loss": 1.4622,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3536518683402357,
      "learning_rate": 4.379562043795621e-06,
      "loss": 1.3519,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6851289834481316,
      "learning_rate": 4.525547445255475e-06,
      "loss": 1.6255,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1408008805631367,
      "learning_rate": 4.671532846715329e-06,
      "loss": 1.2072,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.366020314767896,
      "learning_rate": 4.8175182481751835e-06,
      "loss": 1.3705,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1839994232620947,
      "learning_rate": 4.963503649635037e-06,
      "loss": 1.2993,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1036145731357472,
      "learning_rate": 5.1094890510948916e-06,
      "loss": 1.2042,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1420864514532507,
      "learning_rate": 5.255474452554746e-06,
      "loss": 1.419,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0969643363545438,
      "learning_rate": 5.401459854014599e-06,
      "loss": 1.1267,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0775411932521246,
      "learning_rate": 5.547445255474453e-06,
      "loss": 1.2421,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.060841443934055,
      "learning_rate": 5.693430656934307e-06,
      "loss": 1.1793,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1547507799124865,
      "learning_rate": 5.839416058394161e-06,
      "loss": 1.3276,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9456819175034582,
      "learning_rate": 5.985401459854016e-06,
      "loss": 1.1627,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4670604159208278,
      "learning_rate": 6.13138686131387e-06,
      "loss": 1.4,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0901826721199976,
      "learning_rate": 6.277372262773723e-06,
      "loss": 1.28,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.034560582779727,
      "learning_rate": 6.423357664233577e-06,
      "loss": 1.3776,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9942941003831013,
      "learning_rate": 6.569343065693431e-06,
      "loss": 1.2454,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1051572219195003,
      "learning_rate": 6.715328467153285e-06,
      "loss": 1.3865,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0297208488285288,
      "learning_rate": 6.86131386861314e-06,
      "loss": 1.3139,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9557758739121303,
      "learning_rate": 7.007299270072994e-06,
      "loss": 1.3105,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9800237428455421,
      "learning_rate": 7.153284671532848e-06,
      "loss": 1.2289,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7469314346380025,
      "learning_rate": 7.299270072992701e-06,
      "loss": 0.9904,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0700614801908555,
      "learning_rate": 7.445255474452555e-06,
      "loss": 1.1376,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9785466903049703,
      "learning_rate": 7.591240875912409e-06,
      "loss": 1.1916,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9258408668090249,
      "learning_rate": 7.737226277372264e-06,
      "loss": 1.2119,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0861081325482576,
      "learning_rate": 7.883211678832117e-06,
      "loss": 1.1729,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9760518684519011,
      "learning_rate": 8.029197080291972e-06,
      "loss": 1.2895,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7278011797118368,
      "learning_rate": 8.175182481751825e-06,
      "loss": 0.9904,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9927374449574603,
      "learning_rate": 8.32116788321168e-06,
      "loss": 1.0993,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8716651171663807,
      "learning_rate": 8.467153284671533e-06,
      "loss": 1.0723,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.054762321926421,
      "learning_rate": 8.613138686131386e-06,
      "loss": 1.2407,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9782172869450889,
      "learning_rate": 8.759124087591241e-06,
      "loss": 1.1321,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9319051371049969,
      "learning_rate": 8.905109489051096e-06,
      "loss": 1.1106,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6960247121156266,
      "learning_rate": 9.05109489051095e-06,
      "loss": 0.8907,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9113706419404604,
      "learning_rate": 9.197080291970804e-06,
      "loss": 1.1704,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.799832260904814,
      "learning_rate": 9.343065693430657e-06,
      "loss": 1.1016,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9128577288869439,
      "learning_rate": 9.48905109489051e-06,
      "loss": 1.1215,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0567268717147842,
      "learning_rate": 9.635036496350367e-06,
      "loss": 1.352,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9094517809777068,
      "learning_rate": 9.78102189781022e-06,
      "loss": 1.1847,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7867342525951125,
      "learning_rate": 9.927007299270073e-06,
      "loss": 0.9282,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7933746952472809,
      "learning_rate": 1.0072992700729928e-05,
      "loss": 1.0709,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7806742263067411,
      "learning_rate": 1.0218978102189783e-05,
      "loss": 0.9287,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8109743025409091,
      "learning_rate": 1.0364963503649636e-05,
      "loss": 0.9879,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0589410979160223,
      "learning_rate": 1.0510948905109491e-05,
      "loss": 1.1894,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8271750633881442,
      "learning_rate": 1.0656934306569344e-05,
      "loss": 1.0289,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8404662907754663,
      "learning_rate": 1.0802919708029198e-05,
      "loss": 1.0454,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7471944070111499,
      "learning_rate": 1.0948905109489052e-05,
      "loss": 0.8529,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8895217353047934,
      "learning_rate": 1.1094890510948906e-05,
      "loss": 0.9981,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8869781114829942,
      "learning_rate": 1.124087591240876e-05,
      "loss": 0.9505,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.110690237178331,
      "learning_rate": 1.1386861313868614e-05,
      "loss": 1.1931,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9142916791478641,
      "learning_rate": 1.1532846715328467e-05,
      "loss": 0.9926,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0123559425445243,
      "learning_rate": 1.1678832116788322e-05,
      "loss": 1.26,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.937345396645472,
      "learning_rate": 1.1824817518248176e-05,
      "loss": 1.0732,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9726262222005855,
      "learning_rate": 1.1970802919708031e-05,
      "loss": 1.2389,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8794528194175176,
      "learning_rate": 1.2116788321167885e-05,
      "loss": 1.0843,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.936638563476656,
      "learning_rate": 1.226277372262774e-05,
      "loss": 1.1586,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8416181405560003,
      "learning_rate": 1.2408759124087593e-05,
      "loss": 0.9543,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.959753035329643,
      "learning_rate": 1.2554744525547446e-05,
      "loss": 1.0536,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8612870879139615,
      "learning_rate": 1.27007299270073e-05,
      "loss": 1.004,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.792060369338959,
      "learning_rate": 1.2846715328467154e-05,
      "loss": 0.9973,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8253692985267265,
      "learning_rate": 1.2992700729927009e-05,
      "loss": 0.93,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9603177685132234,
      "learning_rate": 1.3138686131386862e-05,
      "loss": 1.0306,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8960913168384407,
      "learning_rate": 1.3284671532846715e-05,
      "loss": 1.0169,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9697662068817549,
      "learning_rate": 1.343065693430657e-05,
      "loss": 1.1232,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.908702360371072,
      "learning_rate": 1.3576642335766423e-05,
      "loss": 0.9398,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9545076690973209,
      "learning_rate": 1.372262773722628e-05,
      "loss": 1.035,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9967858397214344,
      "learning_rate": 1.3868613138686133e-05,
      "loss": 1.0483,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9594865094220973,
      "learning_rate": 1.4014598540145988e-05,
      "loss": 1.1163,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8901218197528311,
      "learning_rate": 1.416058394160584e-05,
      "loss": 1.0035,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2195698841336509,
      "learning_rate": 1.4306569343065696e-05,
      "loss": 1.0776,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8629587183722728,
      "learning_rate": 1.4452554744525549e-05,
      "loss": 0.899,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8603954845219532,
      "learning_rate": 1.4598540145985402e-05,
      "loss": 0.937,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9607025107271405,
      "learning_rate": 1.4744525547445257e-05,
      "loss": 1.012,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8232046460407965,
      "learning_rate": 1.489051094890511e-05,
      "loss": 0.9778,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9969261730297488,
      "learning_rate": 1.5036496350364965e-05,
      "loss": 0.9778,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9876610841830437,
      "learning_rate": 1.5182481751824818e-05,
      "loss": 1.0761,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8006092538111557,
      "learning_rate": 1.5328467153284673e-05,
      "loss": 0.8488,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9416159241994874,
      "learning_rate": 1.5474452554744528e-05,
      "loss": 1.0294,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.927252329021459,
      "learning_rate": 1.5620437956204383e-05,
      "loss": 0.9767,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8985186332889099,
      "learning_rate": 1.5766423357664234e-05,
      "loss": 1.0582,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8019326840567439,
      "learning_rate": 1.591240875912409e-05,
      "loss": 1.0131,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8459260924140503,
      "learning_rate": 1.6058394160583944e-05,
      "loss": 0.9237,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.119833633464446,
      "learning_rate": 1.62043795620438e-05,
      "loss": 0.9056,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.835529049348371,
      "learning_rate": 1.635036496350365e-05,
      "loss": 0.8131,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6535515026908469,
      "learning_rate": 1.6496350364963505e-05,
      "loss": 0.8077,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8304368982818897,
      "learning_rate": 1.664233576642336e-05,
      "loss": 0.9227,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7453451345404556,
      "learning_rate": 1.678832116788321e-05,
      "loss": 0.7761,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.017988535878321,
      "learning_rate": 1.6934306569343066e-05,
      "loss": 1.0007,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7649825767951453,
      "learning_rate": 1.708029197080292e-05,
      "loss": 0.7844,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.898719080046926,
      "learning_rate": 1.7226277372262773e-05,
      "loss": 0.949,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8779647192155593,
      "learning_rate": 1.737226277372263e-05,
      "loss": 1.0252,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.779365632124019,
      "learning_rate": 1.7518248175182482e-05,
      "loss": 0.8929,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8128048251624574,
      "learning_rate": 1.7664233576642337e-05,
      "loss": 0.953,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8732728601647503,
      "learning_rate": 1.7810218978102192e-05,
      "loss": 0.9142,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8101234525039789,
      "learning_rate": 1.7956204379562047e-05,
      "loss": 0.8924,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8200518966162574,
      "learning_rate": 1.81021897810219e-05,
      "loss": 0.9443,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8558389568113152,
      "learning_rate": 1.8248175182481753e-05,
      "loss": 0.8735,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.812249952133708,
      "learning_rate": 1.8394160583941608e-05,
      "loss": 0.7464,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7711916597580152,
      "learning_rate": 1.854014598540146e-05,
      "loss": 0.7847,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.913375400620912,
      "learning_rate": 1.8686131386861315e-05,
      "loss": 1.0185,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9374111769242447,
      "learning_rate": 1.883211678832117e-05,
      "loss": 1.1294,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7809460620937397,
      "learning_rate": 1.897810218978102e-05,
      "loss": 0.8164,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7555568592404668,
      "learning_rate": 1.912408759124088e-05,
      "loss": 0.8667,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7688544148192044,
      "learning_rate": 1.9270072992700734e-05,
      "loss": 0.785,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8813496174882957,
      "learning_rate": 1.9416058394160586e-05,
      "loss": 0.9051,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8481248887356465,
      "learning_rate": 1.956204379562044e-05,
      "loss": 0.9196,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8583997915211697,
      "learning_rate": 1.9708029197080295e-05,
      "loss": 0.9492,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8923614038515579,
      "learning_rate": 1.9854014598540147e-05,
      "loss": 0.9376,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7630606544354205,
      "learning_rate": 2e-05,
      "loss": 0.921,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7884064392551693,
      "learning_rate": 2.0145985401459857e-05,
      "loss": 0.9123,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7908940728125956,
      "learning_rate": 2.0291970802919708e-05,
      "loss": 0.8448,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7893454922365724,
      "learning_rate": 2.0437956204379566e-05,
      "loss": 0.8967,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9704833981304126,
      "learning_rate": 2.0583941605839418e-05,
      "loss": 0.8912,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.004801428996121,
      "learning_rate": 2.0729927007299273e-05,
      "loss": 0.9759,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9369647087323253,
      "learning_rate": 2.0875912408759124e-05,
      "loss": 0.9831,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8921678130053814,
      "learning_rate": 2.1021897810218982e-05,
      "loss": 0.9521,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9372303256953518,
      "learning_rate": 2.1167883211678834e-05,
      "loss": 1.0517,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8552449948393822,
      "learning_rate": 2.131386861313869e-05,
      "loss": 0.9103,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7676121750284994,
      "learning_rate": 2.145985401459854e-05,
      "loss": 0.8621,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.821653595898437,
      "learning_rate": 2.1605839416058395e-05,
      "loss": 0.8404,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8035734668584738,
      "learning_rate": 2.1751824817518246e-05,
      "loss": 0.8717,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.867084875134465,
      "learning_rate": 2.1897810218978105e-05,
      "loss": 1.0253,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.81662699804174,
      "learning_rate": 2.204379562043796e-05,
      "loss": 0.9117,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8013922616530269,
      "learning_rate": 2.218978102189781e-05,
      "loss": 0.9208,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7479029185767977,
      "learning_rate": 2.2335766423357666e-05,
      "loss": 0.8442,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8770647551380415,
      "learning_rate": 2.248175182481752e-05,
      "loss": 0.9463,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8179934464217773,
      "learning_rate": 2.2627737226277376e-05,
      "loss": 0.8536,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8333939689192803,
      "learning_rate": 2.2773722627737227e-05,
      "loss": 0.8545,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9536900955725867,
      "learning_rate": 2.2919708029197082e-05,
      "loss": 1.0469,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7800980658506356,
      "learning_rate": 2.3065693430656934e-05,
      "loss": 0.7782,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8175695427062762,
      "learning_rate": 2.3211678832116792e-05,
      "loss": 0.8163,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8499995946883189,
      "learning_rate": 2.3357664233576643e-05,
      "loss": 0.7826,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9409476781805804,
      "learning_rate": 2.3503649635036498e-05,
      "loss": 1.0353,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8039945563184248,
      "learning_rate": 2.3649635036496353e-05,
      "loss": 0.7398,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9730280889277279,
      "learning_rate": 2.3795620437956204e-05,
      "loss": 0.9063,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9262831088183515,
      "learning_rate": 2.3941605839416063e-05,
      "loss": 0.9266,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8749799044890016,
      "learning_rate": 2.4087591240875914e-05,
      "loss": 0.8567,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8162508364873224,
      "learning_rate": 2.423357664233577e-05,
      "loss": 0.8099,
      "step": 166
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8779531779132826,
      "learning_rate": 2.437956204379562e-05,
      "loss": 0.9388,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.875780915150506,
      "learning_rate": 2.452554744525548e-05,
      "loss": 1.0335,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8484109077009614,
      "learning_rate": 2.467153284671533e-05,
      "loss": 0.7684,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9240908743069353,
      "learning_rate": 2.4817518248175185e-05,
      "loss": 0.911,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.776400161138908,
      "learning_rate": 2.4963503649635037e-05,
      "loss": 0.7548,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8982001281300744,
      "learning_rate": 2.510948905109489e-05,
      "loss": 0.8434,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9290136892749372,
      "learning_rate": 2.5255474452554743e-05,
      "loss": 0.9355,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.864578067043459,
      "learning_rate": 2.54014598540146e-05,
      "loss": 0.9516,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9777270583131618,
      "learning_rate": 2.5547445255474456e-05,
      "loss": 0.9818,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.872350086132436,
      "learning_rate": 2.5693430656934308e-05,
      "loss": 0.9193,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8453406775215188,
      "learning_rate": 2.5839416058394166e-05,
      "loss": 0.9481,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9523180062348314,
      "learning_rate": 2.5985401459854017e-05,
      "loss": 0.9359,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.837892118866129,
      "learning_rate": 2.6131386861313872e-05,
      "loss": 0.7993,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8423906254732197,
      "learning_rate": 2.6277372262773724e-05,
      "loss": 0.778,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8427808988910546,
      "learning_rate": 2.642335766423358e-05,
      "loss": 0.8411,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9137995415028294,
      "learning_rate": 2.656934306569343e-05,
      "loss": 0.804,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9027583173137855,
      "learning_rate": 2.6715328467153288e-05,
      "loss": 0.8251,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8641413895807349,
      "learning_rate": 2.686131386861314e-05,
      "loss": 0.876,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8854594426013437,
      "learning_rate": 2.7007299270072995e-05,
      "loss": 0.9454,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7840237777227991,
      "learning_rate": 2.7153284671532846e-05,
      "loss": 0.8771,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8778571076566607,
      "learning_rate": 2.7299270072992704e-05,
      "loss": 0.8599,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7821884622236454,
      "learning_rate": 2.744525547445256e-05,
      "loss": 0.8459,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7743019128627272,
      "learning_rate": 2.759124087591241e-05,
      "loss": 0.7925,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7866143878103941,
      "learning_rate": 2.7737226277372266e-05,
      "loss": 0.7913,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7935828100427934,
      "learning_rate": 2.7883211678832117e-05,
      "loss": 0.8149,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8604826637862417,
      "learning_rate": 2.8029197080291975e-05,
      "loss": 0.8143,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8508467159532684,
      "learning_rate": 2.8175182481751827e-05,
      "loss": 0.8817,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7435161142687337,
      "learning_rate": 2.832116788321168e-05,
      "loss": 0.7964,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8553910241895625,
      "learning_rate": 2.8467153284671533e-05,
      "loss": 0.9158,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8562460725234794,
      "learning_rate": 2.861313868613139e-05,
      "loss": 0.9361,
      "step": 196
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8265027765572417,
      "learning_rate": 2.8759124087591243e-05,
      "loss": 0.86,
      "step": 197
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8122909350284694,
      "learning_rate": 2.8905109489051098e-05,
      "loss": 0.8024,
      "step": 198
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8604388501230203,
      "learning_rate": 2.905109489051095e-05,
      "loss": 0.9568,
      "step": 199
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7296128588122619,
      "learning_rate": 2.9197080291970804e-05,
      "loss": 0.6976,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9666471731478535,
      "learning_rate": 2.9343065693430662e-05,
      "loss": 0.86,
      "step": 201
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9643032706897797,
      "learning_rate": 2.9489051094890514e-05,
      "loss": 0.9043,
      "step": 202
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8788740364105233,
      "learning_rate": 2.963503649635037e-05,
      "loss": 0.7955,
      "step": 203
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8494308108645364,
      "learning_rate": 2.978102189781022e-05,
      "loss": 0.7778,
      "step": 204
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7798279981602546,
      "learning_rate": 2.9927007299270075e-05,
      "loss": 0.7827,
      "step": 205
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6336256618906441,
      "learning_rate": 3.007299270072993e-05,
      "loss": 0.6995,
      "step": 206
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7463806958662407,
      "learning_rate": 3.0218978102189785e-05,
      "loss": 0.7585,
      "step": 207
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.759099005947736,
      "learning_rate": 3.0364963503649636e-05,
      "loss": 0.7035,
      "step": 208
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9245880421029641,
      "learning_rate": 3.051094890510949e-05,
      "loss": 0.8373,
      "step": 209
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.893976430953491,
      "learning_rate": 3.0656934306569346e-05,
      "loss": 0.8918,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7408089936344885,
      "learning_rate": 3.08029197080292e-05,
      "loss": 0.7725,
      "step": 211
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7932888568771177,
      "learning_rate": 3.0948905109489056e-05,
      "loss": 0.7768,
      "step": 212
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7333727551470285,
      "learning_rate": 3.109489051094891e-05,
      "loss": 0.6986,
      "step": 213
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8203950386302806,
      "learning_rate": 3.1240875912408765e-05,
      "loss": 0.8117,
      "step": 214
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8596703455332148,
      "learning_rate": 3.138686131386862e-05,
      "loss": 0.8495,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8846367411603081,
      "learning_rate": 3.153284671532847e-05,
      "loss": 0.8416,
      "step": 216
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7913297806424117,
      "learning_rate": 3.167883211678832e-05,
      "loss": 0.7883,
      "step": 217
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8864001067090056,
      "learning_rate": 3.182481751824818e-05,
      "loss": 0.8664,
      "step": 218
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7702646004600188,
      "learning_rate": 3.197080291970803e-05,
      "loss": 0.7783,
      "step": 219
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7858313699281773,
      "learning_rate": 3.211678832116789e-05,
      "loss": 0.7565,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7465709338663865,
      "learning_rate": 3.226277372262774e-05,
      "loss": 0.6761,
      "step": 221
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9191272972946686,
      "learning_rate": 3.24087591240876e-05,
      "loss": 0.8988,
      "step": 222
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8094731216162403,
      "learning_rate": 3.255474452554745e-05,
      "loss": 0.7862,
      "step": 223
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8658699818241734,
      "learning_rate": 3.27007299270073e-05,
      "loss": 0.8778,
      "step": 224
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8461867572813003,
      "learning_rate": 3.284671532846716e-05,
      "loss": 0.7938,
      "step": 225
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8709833733803517,
      "learning_rate": 3.299270072992701e-05,
      "loss": 0.7898,
      "step": 226
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.878491655686796,
      "learning_rate": 3.313868613138687e-05,
      "loss": 0.7788,
      "step": 227
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9325073497063265,
      "learning_rate": 3.328467153284672e-05,
      "loss": 0.7702,
      "step": 228
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7923231832565352,
      "learning_rate": 3.343065693430657e-05,
      "loss": 0.6818,
      "step": 229
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.942743767903117,
      "learning_rate": 3.357664233576642e-05,
      "loss": 0.8443,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8681702415152016,
      "learning_rate": 3.372262773722628e-05,
      "loss": 0.6942,
      "step": 231
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.006703853730836,
      "learning_rate": 3.386861313868613e-05,
      "loss": 0.9115,
      "step": 232
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.757843882117406,
      "learning_rate": 3.401459854014599e-05,
      "loss": 0.766,
      "step": 233
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.757128062322179,
      "learning_rate": 3.416058394160584e-05,
      "loss": 0.8539,
      "step": 234
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8875125239388543,
      "learning_rate": 3.4306569343065694e-05,
      "loss": 0.814,
      "step": 235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.808341055928206,
      "learning_rate": 3.4452554744525545e-05,
      "loss": 0.7324,
      "step": 236
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7680783182955603,
      "learning_rate": 3.4598540145985404e-05,
      "loss": 0.7022,
      "step": 237
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.773625784410827,
      "learning_rate": 3.474452554744526e-05,
      "loss": 0.7348,
      "step": 238
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8452745016586605,
      "learning_rate": 3.4890510948905113e-05,
      "loss": 0.7333,
      "step": 239
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8796795643097877,
      "learning_rate": 3.5036496350364965e-05,
      "loss": 0.8311,
      "step": 240
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6911046291584321,
      "learning_rate": 3.5182481751824816e-05,
      "loss": 0.6326,
      "step": 241
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8840908146595398,
      "learning_rate": 3.5328467153284675e-05,
      "loss": 0.788,
      "step": 242
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8462049303773562,
      "learning_rate": 3.5474452554744526e-05,
      "loss": 0.7607,
      "step": 243
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8676272471544045,
      "learning_rate": 3.5620437956204384e-05,
      "loss": 0.7835,
      "step": 244
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.90018214131091,
      "learning_rate": 3.5766423357664236e-05,
      "loss": 0.9244,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.796911575375581,
      "learning_rate": 3.5912408759124094e-05,
      "loss": 0.8376,
      "step": 246
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9128495996778878,
      "learning_rate": 3.6058394160583946e-05,
      "loss": 0.9573,
      "step": 247
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7764065714509375,
      "learning_rate": 3.62043795620438e-05,
      "loss": 0.7752,
      "step": 248
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8635953098902135,
      "learning_rate": 3.635036496350365e-05,
      "loss": 0.828,
      "step": 249
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7776935499618617,
      "learning_rate": 3.649635036496351e-05,
      "loss": 0.7998,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7325430807913631,
      "learning_rate": 3.6642335766423365e-05,
      "loss": 0.627,
      "step": 251
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8786980276352858,
      "learning_rate": 3.6788321167883217e-05,
      "loss": 0.7724,
      "step": 252
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7809194246896833,
      "learning_rate": 3.693430656934307e-05,
      "loss": 0.6631,
      "step": 253
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8721654825086504,
      "learning_rate": 3.708029197080292e-05,
      "loss": 0.7863,
      "step": 254
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7556796390035293,
      "learning_rate": 3.722627737226278e-05,
      "loss": 0.6549,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.901357995809885,
      "learning_rate": 3.737226277372263e-05,
      "loss": 0.709,
      "step": 256
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.030566364607516,
      "learning_rate": 3.751824817518249e-05,
      "loss": 0.892,
      "step": 257
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8867348824936655,
      "learning_rate": 3.766423357664234e-05,
      "loss": 0.877,
      "step": 258
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.985543783321362,
      "learning_rate": 3.781021897810219e-05,
      "loss": 0.8474,
      "step": 259
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9071408858562116,
      "learning_rate": 3.795620437956204e-05,
      "loss": 0.8303,
      "step": 260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7268676937191998,
      "learning_rate": 3.81021897810219e-05,
      "loss": 0.6717,
      "step": 261
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7889173770814172,
      "learning_rate": 3.824817518248176e-05,
      "loss": 0.7581,
      "step": 262
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7777957847947902,
      "learning_rate": 3.839416058394161e-05,
      "loss": 0.8143,
      "step": 263
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7671321903343677,
      "learning_rate": 3.854014598540147e-05,
      "loss": 0.7566,
      "step": 264
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7969364441763006,
      "learning_rate": 3.868613138686132e-05,
      "loss": 0.7008,
      "step": 265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8587544715131925,
      "learning_rate": 3.883211678832117e-05,
      "loss": 0.7941,
      "step": 266
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8421850880321154,
      "learning_rate": 3.897810218978102e-05,
      "loss": 0.6676,
      "step": 267
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7383172566347763,
      "learning_rate": 3.912408759124088e-05,
      "loss": 0.7128,
      "step": 268
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7890256363217543,
      "learning_rate": 3.927007299270073e-05,
      "loss": 0.7103,
      "step": 269
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8467664487179857,
      "learning_rate": 3.941605839416059e-05,
      "loss": 0.7219,
      "step": 270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8610707977730204,
      "learning_rate": 3.956204379562044e-05,
      "loss": 0.7862,
      "step": 271
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8125713023664098,
      "learning_rate": 3.9708029197080294e-05,
      "loss": 0.7378,
      "step": 272
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8940863689928457,
      "learning_rate": 3.9854014598540145e-05,
      "loss": 0.7886,
      "step": 273
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0050851749746672,
      "learning_rate": 4e-05,
      "loss": 0.7236,
      "step": 274
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8388090635331187,
      "learning_rate": 3.999999873473464e-05,
      "loss": 0.7377,
      "step": 275
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8383014445436562,
      "learning_rate": 3.99999949389387e-05,
      "loss": 0.7259,
      "step": 276
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7510972738292445,
      "learning_rate": 3.999998861261268e-05,
      "loss": 0.7294,
      "step": 277
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.713726901400469,
      "learning_rate": 3.999997975575736e-05,
      "loss": 0.7374,
      "step": 278
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8310339955354938,
      "learning_rate": 3.999996836837388e-05,
      "loss": 0.8416,
      "step": 279
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8503672409522316,
      "learning_rate": 3.9999954450463665e-05,
      "loss": 0.8361,
      "step": 280
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7727501637592222,
      "learning_rate": 3.999993800202848e-05,
      "loss": 0.7638,
      "step": 281
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.752863741310904,
      "learning_rate": 3.9999919023070414e-05,
      "loss": 0.7267,
      "step": 282
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.743777609561108,
      "learning_rate": 3.999989751359186e-05,
      "loss": 0.7195,
      "step": 283
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7491973317001533,
      "learning_rate": 3.999987347359555e-05,
      "loss": 0.6736,
      "step": 284
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8735011410260841,
      "learning_rate": 3.9999846903084514e-05,
      "loss": 0.6922,
      "step": 285
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7681511833998723,
      "learning_rate": 3.999981780206212e-05,
      "loss": 0.6805,
      "step": 286
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7399497982343214,
      "learning_rate": 3.999978617053205e-05,
      "loss": 0.6517,
      "step": 287
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7793165314377594,
      "learning_rate": 3.99997520084983e-05,
      "loss": 0.6802,
      "step": 288
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8442459414980154,
      "learning_rate": 3.9999715315965194e-05,
      "loss": 0.8615,
      "step": 289
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8002566968833671,
      "learning_rate": 3.999967609293739e-05,
      "loss": 0.8276,
      "step": 290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7252502897275802,
      "learning_rate": 3.9999634339419826e-05,
      "loss": 0.6918,
      "step": 291
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7403815533310777,
      "learning_rate": 3.99995900554178e-05,
      "loss": 0.6976,
      "step": 292
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9424472608912726,
      "learning_rate": 3.9999543240936916e-05,
      "loss": 0.77,
      "step": 293
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9165106698438497,
      "learning_rate": 3.999949389598309e-05,
      "loss": 0.8012,
      "step": 294
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8842702662049385,
      "learning_rate": 3.9999442020562575e-05,
      "loss": 0.6576,
      "step": 295
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7214397604572383,
      "learning_rate": 3.999938761468192e-05,
      "loss": 0.6748,
      "step": 296
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8826123913851801,
      "learning_rate": 3.999933067834803e-05,
      "loss": 0.7898,
      "step": 297
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7584184013762938,
      "learning_rate": 3.9999271211568084e-05,
      "loss": 0.6604,
      "step": 298
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7588518232118705,
      "learning_rate": 3.999920921434962e-05,
      "loss": 0.6761,
      "step": 299
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9329383177386635,
      "learning_rate": 3.999914468670048e-05,
      "loss": 0.8267,
      "step": 300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7185478755527461,
      "learning_rate": 3.9999077628628834e-05,
      "loss": 0.6602,
      "step": 301
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.924772525860542,
      "learning_rate": 3.999900804014317e-05,
      "loss": 0.8433,
      "step": 302
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9009317025941854,
      "learning_rate": 3.9998935921252275e-05,
      "loss": 0.8177,
      "step": 303
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8439125681490359,
      "learning_rate": 3.9998861271965285e-05,
      "loss": 0.713,
      "step": 304
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8118401195266246,
      "learning_rate": 3.999878409229164e-05,
      "loss": 0.6455,
      "step": 305
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7715776990627178,
      "learning_rate": 3.999870438224111e-05,
      "loss": 0.7286,
      "step": 306
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9172688514724856,
      "learning_rate": 3.999862214182379e-05,
      "loss": 0.8476,
      "step": 307
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8349527637460848,
      "learning_rate": 3.999853737105007e-05,
      "loss": 0.7318,
      "step": 308
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8912865039256895,
      "learning_rate": 3.999845006993068e-05,
      "loss": 0.7909,
      "step": 309
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8515101425453874,
      "learning_rate": 3.9998360238476655e-05,
      "loss": 0.8353,
      "step": 310
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6814806731373988,
      "learning_rate": 3.9998267876699384e-05,
      "loss": 0.6833,
      "step": 311
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8373271763822236,
      "learning_rate": 3.999817298461054e-05,
      "loss": 0.7619,
      "step": 312
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8559870782638922,
      "learning_rate": 3.9998075562222134e-05,
      "loss": 0.8058,
      "step": 313
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8554773896360247,
      "learning_rate": 3.999797560954649e-05,
      "loss": 0.7257,
      "step": 314
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8004587780228744,
      "learning_rate": 3.9997873126596255e-05,
      "loss": 0.6722,
      "step": 315
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8668054649886653,
      "learning_rate": 3.999776811338439e-05,
      "loss": 0.7144,
      "step": 316
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7637625637987723,
      "learning_rate": 3.999766056992419e-05,
      "loss": 0.6253,
      "step": 317
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8201354426082679,
      "learning_rate": 3.999755049622926e-05,
      "loss": 0.6807,
      "step": 318
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8501310303615375,
      "learning_rate": 3.999743789231353e-05,
      "loss": 0.5714,
      "step": 319
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8012058156011441,
      "learning_rate": 3.9997322758191244e-05,
      "loss": 0.609,
      "step": 320
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.041671632119106,
      "learning_rate": 3.999720509387696e-05,
      "loss": 0.749,
      "step": 321
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7947247234931644,
      "learning_rate": 3.999708489938559e-05,
      "loss": 0.6367,
      "step": 322
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7441789111603445,
      "learning_rate": 3.999696217473231e-05,
      "loss": 0.6123,
      "step": 323
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6904871807413867,
      "learning_rate": 3.999683691993268e-05,
      "loss": 0.541,
      "step": 324
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8181610959733012,
      "learning_rate": 3.9996709135002524e-05,
      "loss": 0.6713,
      "step": 325
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7770591483217604,
      "learning_rate": 3.999657881995802e-05,
      "loss": 0.6928,
      "step": 326
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8852589560278958,
      "learning_rate": 3.999644597481566e-05,
      "loss": 0.8136,
      "step": 327
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8946259627529591,
      "learning_rate": 3.9996310599592244e-05,
      "loss": 0.6859,
      "step": 328
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7997727160648551,
      "learning_rate": 3.999617269430491e-05,
      "loss": 0.6963,
      "step": 329
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7408580719114625,
      "learning_rate": 3.9996032258971097e-05,
      "loss": 0.7109,
      "step": 330
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7875225790510764,
      "learning_rate": 3.999588929360858e-05,
      "loss": 0.8099,
      "step": 331
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.83428135035581,
      "learning_rate": 3.9995743798235445e-05,
      "loss": 0.7305,
      "step": 332
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8211198013194521,
      "learning_rate": 3.9995595772870103e-05,
      "loss": 0.6767,
      "step": 333
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7303975116447102,
      "learning_rate": 3.999544521753128e-05,
      "loss": 0.6688,
      "step": 334
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8747481256090411,
      "learning_rate": 3.999529213223802e-05,
      "loss": 0.7076,
      "step": 335
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7974889297673676,
      "learning_rate": 3.999513651700971e-05,
      "loss": 0.6551,
      "step": 336
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7984550369636046,
      "learning_rate": 3.999497837186603e-05,
      "loss": 0.7061,
      "step": 337
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7393323490355842,
      "learning_rate": 3.999481769682699e-05,
      "loss": 0.6188,
      "step": 338
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7218879533406446,
      "learning_rate": 3.999465449191291e-05,
      "loss": 0.5727,
      "step": 339
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8982077926882093,
      "learning_rate": 3.9994488757144454e-05,
      "loss": 0.7223,
      "step": 340
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.749910826150099,
      "learning_rate": 3.999432049254258e-05,
      "loss": 0.5718,
      "step": 341
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8218926242925816,
      "learning_rate": 3.999414969812859e-05,
      "loss": 0.7052,
      "step": 342
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.702648997779424,
      "learning_rate": 3.999397637392409e-05,
      "loss": 0.5319,
      "step": 343
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.785210792582861,
      "learning_rate": 3.9993800519951e-05,
      "loss": 0.5782,
      "step": 344
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9579061751740375,
      "learning_rate": 3.9993622136231585e-05,
      "loss": 0.8198,
      "step": 345
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7556461950316,
      "learning_rate": 3.99934412227884e-05,
      "loss": 0.5552,
      "step": 346
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7699269473113453,
      "learning_rate": 3.999325777964435e-05,
      "loss": 0.639,
      "step": 347
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7825432944584432,
      "learning_rate": 3.999307180682264e-05,
      "loss": 0.6526,
      "step": 348
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9185256242868887,
      "learning_rate": 3.999288330434681e-05,
      "loss": 0.7195,
      "step": 349
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8126640887726768,
      "learning_rate": 3.9992692272240684e-05,
      "loss": 0.6022,
      "step": 350
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7241460753418988,
      "learning_rate": 3.999249871052845e-05,
      "loss": 0.5674,
      "step": 351
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8330461285310613,
      "learning_rate": 3.99923026192346e-05,
      "loss": 0.6461,
      "step": 352
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9280244072754565,
      "learning_rate": 3.9992103998383946e-05,
      "loss": 0.7077,
      "step": 353
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8357204262364643,
      "learning_rate": 3.999190284800162e-05,
      "loss": 0.6475,
      "step": 354
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7130248061060847,
      "learning_rate": 3.999169916811306e-05,
      "loss": 0.5797,
      "step": 355
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7952051431260259,
      "learning_rate": 3.9991492958744046e-05,
      "loss": 0.6515,
      "step": 356
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7789398369774344,
      "learning_rate": 3.999128421992067e-05,
      "loss": 0.6446,
      "step": 357
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9176244861230839,
      "learning_rate": 3.9991072951669334e-05,
      "loss": 0.6926,
      "step": 358
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9302534416399562,
      "learning_rate": 3.9990859154016785e-05,
      "loss": 0.7405,
      "step": 359
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7284750428542651,
      "learning_rate": 3.999064282699006e-05,
      "loss": 0.5564,
      "step": 360
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9516994885086428,
      "learning_rate": 3.999042397061654e-05,
      "loss": 0.691,
      "step": 361
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9177543876831945,
      "learning_rate": 3.999020258492391e-05,
      "loss": 0.7643,
      "step": 362
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7493901554894491,
      "learning_rate": 3.998997866994017e-05,
      "loss": 0.5473,
      "step": 363
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9026411152779741,
      "learning_rate": 3.998975222569368e-05,
      "loss": 0.7193,
      "step": 364
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8268873214668665,
      "learning_rate": 3.998952325221307e-05,
      "loss": 0.697,
      "step": 365
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8360399201298283,
      "learning_rate": 3.9989291749527314e-05,
      "loss": 0.7089,
      "step": 366
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9191231145135161,
      "learning_rate": 3.998905771766571e-05,
      "loss": 0.7067,
      "step": 367
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8322434848657582,
      "learning_rate": 3.998882115665786e-05,
      "loss": 0.7031,
      "step": 368
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7161513079809682,
      "learning_rate": 3.99885820665337e-05,
      "loss": 0.6038,
      "step": 369
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7122034392638321,
      "learning_rate": 3.998834044732348e-05,
      "loss": 0.617,
      "step": 370
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9640967067789271,
      "learning_rate": 3.998809629905778e-05,
      "loss": 0.6449,
      "step": 371
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8715983413324246,
      "learning_rate": 3.9987849621767473e-05,
      "loss": 0.6974,
      "step": 372
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8641924299950741,
      "learning_rate": 3.998760041548379e-05,
      "loss": 0.5878,
      "step": 373
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0098654957713729,
      "learning_rate": 3.998734868023825e-05,
      "loss": 0.6907,
      "step": 374
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7839792263772378,
      "learning_rate": 3.9987094416062705e-05,
      "loss": 0.5546,
      "step": 375
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7905392566369994,
      "learning_rate": 3.998683762298933e-05,
      "loss": 0.6039,
      "step": 376
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1411723038798014,
      "learning_rate": 3.9986578301050615e-05,
      "loss": 0.7722,
      "step": 377
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7957819004981204,
      "learning_rate": 3.9986316450279365e-05,
      "loss": 0.6741,
      "step": 378
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8070685703921063,
      "learning_rate": 3.998605207070872e-05,
      "loss": 0.6094,
      "step": 379
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7171331131535859,
      "learning_rate": 3.9985785162372135e-05,
      "loss": 0.5898,
      "step": 380
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7022060747325506,
      "learning_rate": 3.998551572530336e-05,
      "loss": 0.542,
      "step": 381
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9226310345388156,
      "learning_rate": 3.998524375953651e-05,
      "loss": 0.7777,
      "step": 382
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8253680708584371,
      "learning_rate": 3.9984969265105984e-05,
      "loss": 0.647,
      "step": 383
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8986164744035455,
      "learning_rate": 3.998469224204652e-05,
      "loss": 0.6549,
      "step": 384
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9477374288157409,
      "learning_rate": 3.998441269039315e-05,
      "loss": 0.7938,
      "step": 385
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7177412168575946,
      "learning_rate": 3.998413061018126e-05,
      "loss": 0.6205,
      "step": 386
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8434483024072181,
      "learning_rate": 3.998384600144655e-05,
      "loss": 0.7054,
      "step": 387
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6939255243213035,
      "learning_rate": 3.9983558864225005e-05,
      "loss": 0.6075,
      "step": 388
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8687155380549173,
      "learning_rate": 3.9983269198552975e-05,
      "loss": 0.6688,
      "step": 389
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8749588547978192,
      "learning_rate": 3.9982977004467106e-05,
      "loss": 0.688,
      "step": 390
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7482574007284549,
      "learning_rate": 3.998268228200437e-05,
      "loss": 0.6648,
      "step": 391
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6874186727798434,
      "learning_rate": 3.998238503120205e-05,
      "loss": 0.527,
      "step": 392
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8229444153046743,
      "learning_rate": 3.9982085252097764e-05,
      "loss": 0.6677,
      "step": 393
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8025994106170787,
      "learning_rate": 3.998178294472944e-05,
      "loss": 0.62,
      "step": 394
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9409990181730299,
      "learning_rate": 3.998147810913532e-05,
      "loss": 0.609,
      "step": 395
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9416416871012572,
      "learning_rate": 3.998117074535398e-05,
      "loss": 0.6682,
      "step": 396
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8087842781314133,
      "learning_rate": 3.998086085342431e-05,
      "loss": 0.6347,
      "step": 397
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8530789443219859,
      "learning_rate": 3.9980548433385525e-05,
      "loss": 0.6421,
      "step": 398
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9054955926953236,
      "learning_rate": 3.9980233485277147e-05,
      "loss": 0.6905,
      "step": 399
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7304986141852277,
      "learning_rate": 3.997991600913903e-05,
      "loss": 0.5146,
      "step": 400
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8920919150008454,
      "learning_rate": 3.997959600501133e-05,
      "loss": 0.6687,
      "step": 401
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9362315816048005,
      "learning_rate": 3.9979273472934556e-05,
      "loss": 0.6466,
      "step": 402
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8421712162517252,
      "learning_rate": 3.99789484129495e-05,
      "loss": 0.6313,
      "step": 403
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7625074949052515,
      "learning_rate": 3.9978620825097306e-05,
      "loss": 0.5929,
      "step": 404
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8694916434234695,
      "learning_rate": 3.997829070941941e-05,
      "loss": 0.6037,
      "step": 405
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7688800360636097,
      "learning_rate": 3.997795806595758e-05,
      "loss": 0.5936,
      "step": 406
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6860970357226771,
      "learning_rate": 3.997762289475392e-05,
      "loss": 0.6259,
      "step": 407
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8702537694869666,
      "learning_rate": 3.9977285195850816e-05,
      "loss": 0.6953,
      "step": 408
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.891297671954456,
      "learning_rate": 3.997694496929102e-05,
      "loss": 0.5534,
      "step": 409
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8962650737266303,
      "learning_rate": 3.9976602215117554e-05,
      "loss": 0.6507,
      "step": 410
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8055955522480743,
      "learning_rate": 3.9976256933373806e-05,
      "loss": 0.5653,
      "step": 411
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7619791221662097,
      "learning_rate": 3.997590912410345e-05,
      "loss": 0.4765,
      "step": 412
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9335185383783076,
      "learning_rate": 3.997555878735051e-05,
      "loss": 0.7551,
      "step": 413
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8748966564730668,
      "learning_rate": 3.997520592315929e-05,
      "loss": 0.5982,
      "step": 414
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8582121177066755,
      "learning_rate": 3.9974850531574444e-05,
      "loss": 0.6746,
      "step": 415
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9448169323544137,
      "learning_rate": 3.997449261264095e-05,
      "loss": 0.6105,
      "step": 416
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8411493488313102,
      "learning_rate": 3.997413216640409e-05,
      "loss": 0.5723,
      "step": 417
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7229822609098383,
      "learning_rate": 3.997376919290946e-05,
      "loss": 0.4986,
      "step": 418
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8178434361578241,
      "learning_rate": 3.9973403692202996e-05,
      "loss": 0.5869,
      "step": 419
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8050939774005232,
      "learning_rate": 3.997303566433094e-05,
      "loss": 0.5368,
      "step": 420
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8669187112031361,
      "learning_rate": 3.997266510933986e-05,
      "loss": 0.5691,
      "step": 421
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7878758033338007,
      "learning_rate": 3.997229202727663e-05,
      "loss": 0.5659,
      "step": 422
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7547933863564875,
      "learning_rate": 3.9971916418188473e-05,
      "loss": 0.5434,
      "step": 423
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9293083251145102,
      "learning_rate": 3.99715382821229e-05,
      "loss": 0.6748,
      "step": 424
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8222121993805745,
      "learning_rate": 3.9971157619127766e-05,
      "loss": 0.5531,
      "step": 425
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9217750689121502,
      "learning_rate": 3.997077442925122e-05,
      "loss": 0.6405,
      "step": 426
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9138232186932482,
      "learning_rate": 3.9970388712541754e-05,
      "loss": 0.7075,
      "step": 427
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7052407116878368,
      "learning_rate": 3.997000046904817e-05,
      "loss": 0.516,
      "step": 428
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7685122254019083,
      "learning_rate": 3.99696096988196e-05,
      "loss": 0.6095,
      "step": 429
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.850810077273434,
      "learning_rate": 3.996921640190547e-05,
      "loss": 0.6907,
      "step": 430
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8651856006110127,
      "learning_rate": 3.996882057835556e-05,
      "loss": 0.5951,
      "step": 431
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.672683761335356,
      "learning_rate": 3.996842222821994e-05,
      "loss": 0.5343,
      "step": 432
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.847871673914895,
      "learning_rate": 3.996802135154901e-05,
      "loss": 0.6006,
      "step": 433
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.986907063906111,
      "learning_rate": 3.9967617948393504e-05,
      "loss": 0.5936,
      "step": 434
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0358227005952716,
      "learning_rate": 3.996721201880445e-05,
      "loss": 0.6529,
      "step": 435
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8039754661664753,
      "learning_rate": 3.996680356283322e-05,
      "loss": 0.5826,
      "step": 436
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8768839309683483,
      "learning_rate": 3.996639258053149e-05,
      "loss": 0.6562,
      "step": 437
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7615953785682957,
      "learning_rate": 3.996597907195126e-05,
      "loss": 0.6211,
      "step": 438
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9001598931377726,
      "learning_rate": 3.996556303714485e-05,
      "loss": 0.7248,
      "step": 439
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7273502386440862,
      "learning_rate": 3.996514447616489e-05,
      "loss": 0.5431,
      "step": 440
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8661286359048096,
      "learning_rate": 3.996472338906436e-05,
      "loss": 0.5844,
      "step": 441
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6760555879151913,
      "learning_rate": 3.996429977589653e-05,
      "loss": 0.6102,
      "step": 442
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8770845990085002,
      "learning_rate": 3.996387363671498e-05,
      "loss": 0.6307,
      "step": 443
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.805713739886529,
      "learning_rate": 3.9963444971573656e-05,
      "loss": 0.5948,
      "step": 444
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.810674156398536,
      "learning_rate": 3.996301378052678e-05,
      "loss": 0.6342,
      "step": 445
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8558961331975985,
      "learning_rate": 3.996258006362891e-05,
      "loss": 0.5744,
      "step": 446
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8517324995387747,
      "learning_rate": 3.9962143820934926e-05,
      "loss": 0.6575,
      "step": 447
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.768192617980181,
      "learning_rate": 3.996170505250002e-05,
      "loss": 0.5407,
      "step": 448
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8158950300205333,
      "learning_rate": 3.996126375837971e-05,
      "loss": 0.5655,
      "step": 449
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9028928332901293,
      "learning_rate": 3.9960819938629834e-05,
      "loss": 0.6456,
      "step": 450
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8508846840103531,
      "learning_rate": 3.996037359330654e-05,
      "loss": 0.5795,
      "step": 451
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8245554231877659,
      "learning_rate": 3.995992472246632e-05,
      "loss": 0.5643,
      "step": 452
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8606073733140465,
      "learning_rate": 3.995947332616594e-05,
      "loss": 0.646,
      "step": 453
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7217634304740731,
      "learning_rate": 3.995901940446254e-05,
      "loss": 0.4658,
      "step": 454
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8505958880642853,
      "learning_rate": 3.995856295741354e-05,
      "loss": 0.6487,
      "step": 455
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8039634558191647,
      "learning_rate": 3.995810398507669e-05,
      "loss": 0.5096,
      "step": 456
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8961179894625264,
      "learning_rate": 3.995764248751007e-05,
      "loss": 0.5975,
      "step": 457
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8722731470790754,
      "learning_rate": 3.995717846477207e-05,
      "loss": 0.5573,
      "step": 458
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9058792408175009,
      "learning_rate": 3.9956711916921405e-05,
      "loss": 0.5735,
      "step": 459
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8868769701988912,
      "learning_rate": 3.9956242844017094e-05,
      "loss": 0.5576,
      "step": 460
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8908005591579613,
      "learning_rate": 3.9955771246118496e-05,
      "loss": 0.5409,
      "step": 461
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7559804499056149,
      "learning_rate": 3.995529712328528e-05,
      "loss": 0.5257,
      "step": 462
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7768366707686001,
      "learning_rate": 3.995482047557743e-05,
      "loss": 0.4991,
      "step": 463
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8049373053784405,
      "learning_rate": 3.995434130305526e-05,
      "loss": 0.5321,
      "step": 464
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0609134721914388,
      "learning_rate": 3.99538596057794e-05,
      "loss": 0.5969,
      "step": 465
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8402193870608702,
      "learning_rate": 3.995337538381079e-05,
      "loss": 0.4918,
      "step": 466
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8196973401437764,
      "learning_rate": 3.99528886372107e-05,
      "loss": 0.5562,
      "step": 467
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.831186214667182,
      "learning_rate": 3.995239936604072e-05,
      "loss": 0.5559,
      "step": 468
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0103769720611149,
      "learning_rate": 3.9951907570362746e-05,
      "loss": 0.6291,
      "step": 469
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8113482454878655,
      "learning_rate": 3.995141325023902e-05,
      "loss": 0.5709,
      "step": 470
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7842536786850128,
      "learning_rate": 3.995091640573207e-05,
      "loss": 0.5358,
      "step": 471
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9247829672368507,
      "learning_rate": 3.995041703690477e-05,
      "loss": 0.5103,
      "step": 472
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8645849266190766,
      "learning_rate": 3.9949915143820295e-05,
      "loss": 0.5109,
      "step": 473
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7540053070296453,
      "learning_rate": 3.994941072654215e-05,
      "loss": 0.5065,
      "step": 474
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.112167229335038,
      "learning_rate": 3.994890378513417e-05,
      "loss": 0.6715,
      "step": 475
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8723396321123827,
      "learning_rate": 3.9948394319660485e-05,
      "loss": 0.5252,
      "step": 476
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0691267253068364,
      "learning_rate": 3.9947882330185554e-05,
      "loss": 0.6188,
      "step": 477
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8231828878637135,
      "learning_rate": 3.994736781677416e-05,
      "loss": 0.5632,
      "step": 478
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9086502123252735,
      "learning_rate": 3.9946850779491406e-05,
      "loss": 0.7469,
      "step": 479
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7043895793715881,
      "learning_rate": 3.994633121840271e-05,
      "loss": 0.4731,
      "step": 480
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7827036113730268,
      "learning_rate": 3.994580913357381e-05,
      "loss": 0.6098,
      "step": 481
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7770199125112138,
      "learning_rate": 3.994528452507076e-05,
      "loss": 0.5309,
      "step": 482
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7678923994822564,
      "learning_rate": 3.994475739295993e-05,
      "loss": 0.4945,
      "step": 483
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6970220402115779,
      "learning_rate": 3.994422773730803e-05,
      "loss": 0.4826,
      "step": 484
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9646826235021102,
      "learning_rate": 3.994369555818208e-05,
      "loss": 0.6833,
      "step": 485
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9048176330897213,
      "learning_rate": 3.99431608556494e-05,
      "loss": 0.5411,
      "step": 486
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.011170758041405,
      "learning_rate": 3.9942623629777645e-05,
      "loss": 0.6149,
      "step": 487
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.730341527947133,
      "learning_rate": 3.99420838806348e-05,
      "loss": 0.5215,
      "step": 488
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7560471885425543,
      "learning_rate": 3.994154160828915e-05,
      "loss": 0.5124,
      "step": 489
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7801216369321703,
      "learning_rate": 3.99409968128093e-05,
      "loss": 0.4579,
      "step": 490
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7303572788964564,
      "learning_rate": 3.994044949426419e-05,
      "loss": 0.4622,
      "step": 491
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9471121437556517,
      "learning_rate": 3.993989965272308e-05,
      "loss": 0.669,
      "step": 492
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8881361467922317,
      "learning_rate": 3.993934728825552e-05,
      "loss": 0.6474,
      "step": 493
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9198785273550014,
      "learning_rate": 3.99387924009314e-05,
      "loss": 0.5895,
      "step": 494
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7581703285069594,
      "learning_rate": 3.993823499082094e-05,
      "loss": 0.5553,
      "step": 495
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.638579872340838,
      "learning_rate": 3.9937675057994666e-05,
      "loss": 0.5458,
      "step": 496
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8006771112973574,
      "learning_rate": 3.993711260252342e-05,
      "loss": 0.5568,
      "step": 497
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7897804317884456,
      "learning_rate": 3.993654762447837e-05,
      "loss": 0.5905,
      "step": 498
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6439291251140656,
      "learning_rate": 3.9935980123930994e-05,
      "loss": 0.3884,
      "step": 499
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6861162999720982,
      "learning_rate": 3.9935410100953105e-05,
      "loss": 0.3851,
      "step": 500
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9067914266218037,
      "learning_rate": 3.9934837555616815e-05,
      "loss": 0.5636,
      "step": 501
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8180201881361133,
      "learning_rate": 3.993426248799458e-05,
      "loss": 0.5483,
      "step": 502
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8958178600187385,
      "learning_rate": 3.993368489815915e-05,
      "loss": 0.598,
      "step": 503
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7910592149568487,
      "learning_rate": 3.993310478618361e-05,
      "loss": 0.474,
      "step": 504
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8537360633393979,
      "learning_rate": 3.993252215214136e-05,
      "loss": 0.5338,
      "step": 505
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.79487838438693,
      "learning_rate": 3.993193699610612e-05,
      "loss": 0.615,
      "step": 506
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9291019679468188,
      "learning_rate": 3.993134931815192e-05,
      "loss": 0.5772,
      "step": 507
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8646719247554688,
      "learning_rate": 3.9930759118353124e-05,
      "loss": 0.5711,
      "step": 508
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9382375359178702,
      "learning_rate": 3.993016639678441e-05,
      "loss": 0.6286,
      "step": 509
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0784286955852909,
      "learning_rate": 3.992957115352077e-05,
      "loss": 0.7173,
      "step": 510
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7641437760325074,
      "learning_rate": 3.9928973388637516e-05,
      "loss": 0.5712,
      "step": 511
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8502802863532842,
      "learning_rate": 3.992837310221028e-05,
      "loss": 0.5857,
      "step": 512
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.774470939327443,
      "learning_rate": 3.992777029431502e-05,
      "loss": 0.501,
      "step": 513
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6755784231993817,
      "learning_rate": 3.9927164965028006e-05,
      "loss": 0.4305,
      "step": 514
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8052972501851107,
      "learning_rate": 3.9926557114425824e-05,
      "loss": 0.4948,
      "step": 515
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.005717028573092,
      "learning_rate": 3.9925946742585385e-05,
      "loss": 0.5787,
      "step": 516
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0117591643638293,
      "learning_rate": 3.992533384958392e-05,
      "loss": 0.4764,
      "step": 517
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1469064484735854,
      "learning_rate": 3.9924718435498964e-05,
      "loss": 0.6591,
      "step": 518
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8531709232476266,
      "learning_rate": 3.9924100500408396e-05,
      "loss": 0.5067,
      "step": 519
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8332974465908206,
      "learning_rate": 3.9923480044390405e-05,
      "loss": 0.5137,
      "step": 520
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8212320532499854,
      "learning_rate": 3.992285706752349e-05,
      "loss": 0.4816,
      "step": 521
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7909313392474584,
      "learning_rate": 3.9922231569886464e-05,
      "loss": 0.46,
      "step": 522
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.926367819341693,
      "learning_rate": 3.992160355155848e-05,
      "loss": 0.566,
      "step": 523
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.66833142702048,
      "learning_rate": 3.9920973012619e-05,
      "loss": 0.3985,
      "step": 524
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8056006944184592,
      "learning_rate": 3.99203399531478e-05,
      "loss": 0.4224,
      "step": 525
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7928465199341881,
      "learning_rate": 3.9919704373224984e-05,
      "loss": 0.5168,
      "step": 526
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7350239422849549,
      "learning_rate": 3.991906627293096e-05,
      "loss": 0.4549,
      "step": 527
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9181886450181526,
      "learning_rate": 3.991842565234647e-05,
      "loss": 0.5915,
      "step": 528
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7769180742569036,
      "learning_rate": 3.991778251155257e-05,
      "loss": 0.4831,
      "step": 529
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9433668937936327,
      "learning_rate": 3.991713685063063e-05,
      "loss": 0.5375,
      "step": 530
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9035488520019831,
      "learning_rate": 3.991648866966235e-05,
      "loss": 0.5769,
      "step": 531
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6780331966999691,
      "learning_rate": 3.991583796872974e-05,
      "loss": 0.422,
      "step": 532
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8719606747115424,
      "learning_rate": 3.991518474791512e-05,
      "loss": 0.4796,
      "step": 533
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9033246014467438,
      "learning_rate": 3.991452900730116e-05,
      "loss": 0.6109,
      "step": 534
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8063878362955893,
      "learning_rate": 3.991387074697081e-05,
      "loss": 0.4728,
      "step": 535
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8953218922280521,
      "learning_rate": 3.991320996700737e-05,
      "loss": 0.4734,
      "step": 536
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9101199687226137,
      "learning_rate": 3.991254666749445e-05,
      "loss": 0.4892,
      "step": 537
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7384766819552022,
      "learning_rate": 3.991188084851596e-05,
      "loss": 0.4767,
      "step": 538
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9491709630420719,
      "learning_rate": 3.991121251015615e-05,
      "loss": 0.5161,
      "step": 539
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.909931728191452,
      "learning_rate": 3.991054165249958e-05,
      "loss": 0.5798,
      "step": 540
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8027158489085024,
      "learning_rate": 3.9909868275631136e-05,
      "loss": 0.4564,
      "step": 541
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9044503726408262,
      "learning_rate": 3.990919237963602e-05,
      "loss": 0.5053,
      "step": 542
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8712957857817895,
      "learning_rate": 3.9908513964599746e-05,
      "loss": 0.4985,
      "step": 543
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9166160844082891,
      "learning_rate": 3.9907833030608153e-05,
      "loss": 0.48,
      "step": 544
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9066680075715005,
      "learning_rate": 3.9907149577747394e-05,
      "loss": 0.6192,
      "step": 545
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.884837233990768,
      "learning_rate": 3.990646360610395e-05,
      "loss": 0.4844,
      "step": 546
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9263782105663679,
      "learning_rate": 3.990577511576461e-05,
      "loss": 0.5146,
      "step": 547
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8135124281066956,
      "learning_rate": 3.9905084106816494e-05,
      "loss": 0.5439,
      "step": 548
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7885608400326016,
      "learning_rate": 3.990439057934702e-05,
      "loss": 0.4937,
      "step": 549
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7520273939834736,
      "learning_rate": 3.990369453344394e-05,
      "loss": 0.4919,
      "step": 550
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.728270911466838,
      "learning_rate": 3.990299596919533e-05,
      "loss": 0.4685,
      "step": 551
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8988349035638267,
      "learning_rate": 3.9902294886689576e-05,
      "loss": 0.56,
      "step": 552
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8178949973186007,
      "learning_rate": 3.990159128601538e-05,
      "loss": 0.4903,
      "step": 553
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1128534398506127,
      "learning_rate": 3.990088516726177e-05,
      "loss": 0.5388,
      "step": 554
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1820542297348275,
      "learning_rate": 3.990017653051808e-05,
      "loss": 0.5334,
      "step": 555
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8657062698939187,
      "learning_rate": 3.9899465375873985e-05,
      "loss": 0.4955,
      "step": 556
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9360997232963838,
      "learning_rate": 3.989875170341945e-05,
      "loss": 0.6707,
      "step": 557
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.99819724187575,
      "learning_rate": 3.989803551324479e-05,
      "loss": 0.5776,
      "step": 558
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8188566822698722,
      "learning_rate": 3.9897316805440606e-05,
      "loss": 0.4849,
      "step": 559
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8046237957376331,
      "learning_rate": 3.989659558009784e-05,
      "loss": 0.5574,
      "step": 560
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7514849902371689,
      "learning_rate": 3.989587183730775e-05,
      "loss": 0.489,
      "step": 561
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8620842968465351,
      "learning_rate": 3.98951455771619e-05,
      "loss": 0.5149,
      "step": 562
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8285908378479403,
      "learning_rate": 3.989441679975219e-05,
      "loss": 0.4171,
      "step": 563
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8647104575828756,
      "learning_rate": 3.989368550517083e-05,
      "loss": 0.5928,
      "step": 564
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7672005615365576,
      "learning_rate": 3.989295169351033e-05,
      "loss": 0.3149,
      "step": 565
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7527293768787586,
      "learning_rate": 3.989221536486357e-05,
      "loss": 0.4782,
      "step": 566
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.749104243051563,
      "learning_rate": 3.989147651932368e-05,
      "loss": 0.3828,
      "step": 567
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0380230056477102,
      "learning_rate": 3.989073515698417e-05,
      "loss": 0.4976,
      "step": 568
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8024857407488464,
      "learning_rate": 3.988999127793883e-05,
      "loss": 0.393,
      "step": 569
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.217959220804909,
      "learning_rate": 3.988924488228178e-05,
      "loss": 0.4469,
      "step": 570
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7581582214800285,
      "learning_rate": 3.988849597010746e-05,
      "loss": 0.3947,
      "step": 571
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9045742919402086,
      "learning_rate": 3.988774454151063e-05,
      "loss": 0.4732,
      "step": 572
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7909181511244426,
      "learning_rate": 3.9886990596586364e-05,
      "loss": 0.4606,
      "step": 573
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7998771960655621,
      "learning_rate": 3.988623413543006e-05,
      "loss": 0.4945,
      "step": 574
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7605340988896442,
      "learning_rate": 3.988547515813742e-05,
      "loss": 0.4552,
      "step": 575
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.79865402898783,
      "learning_rate": 3.9884713664804485e-05,
      "loss": 0.5731,
      "step": 576
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7210486186112235,
      "learning_rate": 3.9883949655527594e-05,
      "loss": 0.5238,
      "step": 577
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7120687685556387,
      "learning_rate": 3.9883183130403424e-05,
      "loss": 0.4322,
      "step": 578
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8625105788093526,
      "learning_rate": 3.9882414089528965e-05,
      "loss": 0.6855,
      "step": 579
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8621647031009402,
      "learning_rate": 3.98816425330015e-05,
      "loss": 0.5546,
      "step": 580
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6415601743125476,
      "learning_rate": 3.988086846091867e-05,
      "loss": 0.3759,
      "step": 581
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8710250485275114,
      "learning_rate": 3.9880091873378416e-05,
      "loss": 0.6254,
      "step": 582
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0054804115834395,
      "learning_rate": 3.9879312770478986e-05,
      "loss": 0.5297,
      "step": 583
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8756127255741422,
      "learning_rate": 3.9878531152318966e-05,
      "loss": 0.4065,
      "step": 584
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.03466047974667,
      "learning_rate": 3.987774701899724e-05,
      "loss": 0.4764,
      "step": 585
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1189225601863473,
      "learning_rate": 3.987696037061304e-05,
      "loss": 0.5426,
      "step": 586
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8213950870039111,
      "learning_rate": 3.987617120726588e-05,
      "loss": 0.4189,
      "step": 587
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7069288991115357,
      "learning_rate": 3.9875379529055624e-05,
      "loss": 0.3873,
      "step": 588
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8552682720062313,
      "learning_rate": 3.987458533608243e-05,
      "loss": 0.4539,
      "step": 589
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8658534605778868,
      "learning_rate": 3.987378862844679e-05,
      "loss": 0.5051,
      "step": 590
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7952194219356961,
      "learning_rate": 3.987298940624951e-05,
      "loss": 0.505,
      "step": 591
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9537159072962105,
      "learning_rate": 3.987218766959171e-05,
      "loss": 0.606,
      "step": 592
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7129695064850823,
      "learning_rate": 3.987138341857483e-05,
      "loss": 0.4469,
      "step": 593
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8865645354351441,
      "learning_rate": 3.987057665330063e-05,
      "loss": 0.5366,
      "step": 594
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7025878549939398,
      "learning_rate": 3.986976737387118e-05,
      "loss": 0.4002,
      "step": 595
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8067180605613338,
      "learning_rate": 3.986895558038889e-05,
      "loss": 0.5418,
      "step": 596
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0673460679216866,
      "learning_rate": 3.9868141272956466e-05,
      "loss": 0.5032,
      "step": 597
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1270499096966131,
      "learning_rate": 3.986732445167694e-05,
      "loss": 0.4783,
      "step": 598
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9116535232080022,
      "learning_rate": 3.986650511665366e-05,
      "loss": 0.4888,
      "step": 599
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8390148952057622,
      "learning_rate": 3.9865683267990295e-05,
      "loss": 0.431,
      "step": 600
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9276603910158733,
      "learning_rate": 3.986485890579083e-05,
      "loss": 0.5032,
      "step": 601
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8495850307440783,
      "learning_rate": 3.986403203015957e-05,
      "loss": 0.4827,
      "step": 602
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.045770888471718,
      "learning_rate": 3.986320264120113e-05,
      "loss": 0.6028,
      "step": 603
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8439683631621782,
      "learning_rate": 3.9862370739020455e-05,
      "loss": 0.4724,
      "step": 604
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.741322505228919,
      "learning_rate": 3.986153632372281e-05,
      "loss": 0.4573,
      "step": 605
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0254748500565236,
      "learning_rate": 3.9860699395413764e-05,
      "loss": 0.5496,
      "step": 606
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7416375783152273,
      "learning_rate": 3.9859859954199206e-05,
      "loss": 0.374,
      "step": 607
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.798779884877756,
      "learning_rate": 3.985901800018535e-05,
      "loss": 0.4794,
      "step": 608
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9340540816437252,
      "learning_rate": 3.985817353347874e-05,
      "loss": 0.5524,
      "step": 609
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8234748918704481,
      "learning_rate": 3.98573265541862e-05,
      "loss": 0.4616,
      "step": 610
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7236925272942816,
      "learning_rate": 3.9856477062414914e-05,
      "loss": 0.361,
      "step": 611
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8200101022563254,
      "learning_rate": 3.985562505827235e-05,
      "loss": 0.4695,
      "step": 612
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7971260572895745,
      "learning_rate": 3.985477054186632e-05,
      "loss": 0.3912,
      "step": 613
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7372485815262885,
      "learning_rate": 3.985391351330494e-05,
      "loss": 0.4484,
      "step": 614
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9282647795051534,
      "learning_rate": 3.985305397269665e-05,
      "loss": 0.5281,
      "step": 615
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6776210459298431,
      "learning_rate": 3.985219192015019e-05,
      "loss": 0.3881,
      "step": 616
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.902390945323282,
      "learning_rate": 3.985132735577466e-05,
      "loss": 0.586,
      "step": 617
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7948999800232522,
      "learning_rate": 3.985046027967943e-05,
      "loss": 0.4314,
      "step": 618
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7553545229375843,
      "learning_rate": 3.9849590691974206e-05,
      "loss": 0.4977,
      "step": 619
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8939691968364601,
      "learning_rate": 3.984871859276902e-05,
      "loss": 0.5062,
      "step": 620
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8110558440011825,
      "learning_rate": 3.984784398217422e-05,
      "loss": 0.5306,
      "step": 621
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8278026313236698,
      "learning_rate": 3.984696686030046e-05,
      "loss": 0.4389,
      "step": 622
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7135830795262094,
      "learning_rate": 3.984608722725873e-05,
      "loss": 0.3995,
      "step": 623
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7439518886925478,
      "learning_rate": 3.9845205083160315e-05,
      "loss": 0.4798,
      "step": 624
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8649653927269196,
      "learning_rate": 3.9844320428116834e-05,
      "loss": 0.5324,
      "step": 625
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8203098842034187,
      "learning_rate": 3.984343326224022e-05,
      "loss": 0.4146,
      "step": 626
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.131329216475481,
      "learning_rate": 3.984254358564272e-05,
      "loss": 0.5946,
      "step": 627
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8553907803054434,
      "learning_rate": 3.9841651398436907e-05,
      "loss": 0.4519,
      "step": 628
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6543785613581563,
      "learning_rate": 3.984075670073566e-05,
      "loss": 0.3514,
      "step": 629
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8992314023769143,
      "learning_rate": 3.983985949265219e-05,
      "loss": 0.55,
      "step": 630
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7436674921768419,
      "learning_rate": 3.9838959774300014e-05,
      "loss": 0.4009,
      "step": 631
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7493322499276991,
      "learning_rate": 3.983805754579297e-05,
      "loss": 0.3852,
      "step": 632
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8291136759849222,
      "learning_rate": 3.983715280724521e-05,
      "loss": 0.4515,
      "step": 633
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7341967427230568,
      "learning_rate": 3.98362455587712e-05,
      "loss": 0.4011,
      "step": 634
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7381493586850031,
      "learning_rate": 3.983533580048576e-05,
      "loss": 0.4676,
      "step": 635
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7715101406768454,
      "learning_rate": 3.9834423532503975e-05,
      "loss": 0.4699,
      "step": 636
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.718094900378569,
      "learning_rate": 3.9833508754941275e-05,
      "loss": 0.4098,
      "step": 637
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8582567048048435,
      "learning_rate": 3.9832591467913405e-05,
      "loss": 0.4472,
      "step": 638
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9982396367024982,
      "learning_rate": 3.983167167153642e-05,
      "loss": 0.4947,
      "step": 639
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8140285126076521,
      "learning_rate": 3.9830749365926716e-05,
      "loss": 0.3669,
      "step": 640
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8844944957434661,
      "learning_rate": 3.982982455120097e-05,
      "loss": 0.4538,
      "step": 641
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8430719123341497,
      "learning_rate": 3.982889722747621e-05,
      "loss": 0.3993,
      "step": 642
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8022406700354929,
      "learning_rate": 3.982796739486976e-05,
      "loss": 0.4895,
      "step": 643
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6767667959698668,
      "learning_rate": 3.9827035053499264e-05,
      "loss": 0.3653,
      "step": 644
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8436771997198442,
      "learning_rate": 3.98261002034827e-05,
      "loss": 0.4349,
      "step": 645
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.764853419553518,
      "learning_rate": 3.982516284493834e-05,
      "loss": 0.3761,
      "step": 646
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6259443301562875,
      "learning_rate": 3.982422297798479e-05,
      "loss": 0.3636,
      "step": 647
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9050084850798686,
      "learning_rate": 3.982328060274097e-05,
      "loss": 0.4513,
      "step": 648
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.727292258368843,
      "learning_rate": 3.982233571932611e-05,
      "loss": 0.3982,
      "step": 649
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.696803989983554,
      "learning_rate": 3.982138832785976e-05,
      "loss": 0.4042,
      "step": 650
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9918835392123475,
      "learning_rate": 3.9820438428461804e-05,
      "loss": 0.4666,
      "step": 651
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8799281981826182,
      "learning_rate": 3.981948602125242e-05,
      "loss": 0.4934,
      "step": 652
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8043596655414714,
      "learning_rate": 3.9818531106352105e-05,
      "loss": 0.4225,
      "step": 653
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8476339592178188,
      "learning_rate": 3.98175736838817e-05,
      "loss": 0.4979,
      "step": 654
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7266337041932338,
      "learning_rate": 3.981661375396233e-05,
      "loss": 0.4047,
      "step": 655
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8436533554676912,
      "learning_rate": 3.981565131671546e-05,
      "loss": 0.5754,
      "step": 656
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7476992284844522,
      "learning_rate": 3.981468637226286e-05,
      "loss": 0.3768,
      "step": 657
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8141812023816225,
      "learning_rate": 3.981371892072661e-05,
      "loss": 0.4427,
      "step": 658
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8764500183721773,
      "learning_rate": 3.981274896222915e-05,
      "loss": 0.4539,
      "step": 659
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9053560649277458,
      "learning_rate": 3.981177649689317e-05,
      "loss": 0.4496,
      "step": 660
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6096858674403417,
      "learning_rate": 3.9810801524841724e-05,
      "loss": 0.3538,
      "step": 661
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7287829933745966,
      "learning_rate": 3.980982404619819e-05,
      "loss": 0.4358,
      "step": 662
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7010359353392834,
      "learning_rate": 3.9808844061086225e-05,
      "loss": 0.385,
      "step": 663
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9392858344403822,
      "learning_rate": 3.9807861569629815e-05,
      "loss": 0.528,
      "step": 664
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.72071478609405,
      "learning_rate": 3.98068765719533e-05,
      "loss": 0.3393,
      "step": 665
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7639010346535613,
      "learning_rate": 3.980588906818129e-05,
      "loss": 0.395,
      "step": 666
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9596415208530081,
      "learning_rate": 3.9804899058438734e-05,
      "loss": 0.4851,
      "step": 667
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8549654799737334,
      "learning_rate": 3.980390654285088e-05,
      "loss": 0.4556,
      "step": 668
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7438391527869523,
      "learning_rate": 3.980291152154334e-05,
      "loss": 0.4144,
      "step": 669
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8095402361134165,
      "learning_rate": 3.980191399464198e-05,
      "loss": 0.3902,
      "step": 670
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7470857905361475,
      "learning_rate": 3.9800913962273036e-05,
      "loss": 0.3759,
      "step": 671
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7369768032579752,
      "learning_rate": 3.979991142456302e-05,
      "loss": 0.3798,
      "step": 672
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.759339396894067,
      "learning_rate": 3.97989063816388e-05,
      "loss": 0.395,
      "step": 673
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8302124632792892,
      "learning_rate": 3.9797898833627514e-05,
      "loss": 0.4224,
      "step": 674
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7278666533211542,
      "learning_rate": 3.979688878065666e-05,
      "loss": 0.3764,
      "step": 675
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7908713880359691,
      "learning_rate": 3.979587622285404e-05,
      "loss": 0.3578,
      "step": 676
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8879192812550103,
      "learning_rate": 3.979486116034776e-05,
      "loss": 0.5345,
      "step": 677
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7621693769852078,
      "learning_rate": 3.979384359326626e-05,
      "loss": 0.4503,
      "step": 678
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7641127306569218,
      "learning_rate": 3.9792823521738285e-05,
      "loss": 0.3742,
      "step": 679
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9125055926948455,
      "learning_rate": 3.97918009458929e-05,
      "loss": 0.4496,
      "step": 680
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7408004649465679,
      "learning_rate": 3.979077586585949e-05,
      "loss": 0.3832,
      "step": 681
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6839487298536606,
      "learning_rate": 3.9789748281767754e-05,
      "loss": 0.3796,
      "step": 682
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.915190823569343,
      "learning_rate": 3.978871819374771e-05,
      "loss": 0.3978,
      "step": 683
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8545008021329147,
      "learning_rate": 3.978768560192969e-05,
      "loss": 0.4422,
      "step": 684
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8117363348663309,
      "learning_rate": 3.978665050644435e-05,
      "loss": 0.3714,
      "step": 685
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9090595432310778,
      "learning_rate": 3.978561290742265e-05,
      "loss": 0.4485,
      "step": 686
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7339571311811625,
      "learning_rate": 3.978457280499587e-05,
      "loss": 0.3612,
      "step": 687
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7724578510131132,
      "learning_rate": 3.978353019929562e-05,
      "loss": 0.4277,
      "step": 688
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8283611896469473,
      "learning_rate": 3.978248509045382e-05,
      "loss": 0.4569,
      "step": 689
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7955528210954658,
      "learning_rate": 3.978143747860269e-05,
      "loss": 0.3966,
      "step": 690
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7690746676254782,
      "learning_rate": 3.978038736387479e-05,
      "loss": 0.4785,
      "step": 691
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7890240877074117,
      "learning_rate": 3.977933474640298e-05,
      "loss": 0.4321,
      "step": 692
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9459794899247427,
      "learning_rate": 3.977827962632046e-05,
      "loss": 0.5031,
      "step": 693
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.793011969530129,
      "learning_rate": 3.9777222003760714e-05,
      "loss": 0.3606,
      "step": 694
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.823145452316263,
      "learning_rate": 3.977616187885757e-05,
      "loss": 0.4653,
      "step": 695
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7724615933768747,
      "learning_rate": 3.977509925174515e-05,
      "loss": 0.396,
      "step": 696
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9428965062846085,
      "learning_rate": 3.9774034122557924e-05,
      "loss": 0.534,
      "step": 697
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6913907324132594,
      "learning_rate": 3.977296649143064e-05,
      "loss": 0.3577,
      "step": 698
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8110297913118387,
      "learning_rate": 3.977189635849839e-05,
      "loss": 0.4093,
      "step": 699
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6609588349057541,
      "learning_rate": 3.9770823723896574e-05,
      "loss": 0.3489,
      "step": 700
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6835736081016998,
      "learning_rate": 3.976974858776091e-05,
      "loss": 0.3421,
      "step": 701
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9106013310798687,
      "learning_rate": 3.976867095022742e-05,
      "loss": 0.5828,
      "step": 702
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9108473482883123,
      "learning_rate": 3.976759081143247e-05,
      "loss": 0.4331,
      "step": 703
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.889465413701839,
      "learning_rate": 3.9766508171512715e-05,
      "loss": 0.4431,
      "step": 704
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7614113211741779,
      "learning_rate": 3.976542303060515e-05,
      "loss": 0.335,
      "step": 705
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7986397369524915,
      "learning_rate": 3.976433538884706e-05,
      "loss": 0.4339,
      "step": 706
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8725885814088392,
      "learning_rate": 3.976324524637606e-05,
      "loss": 0.444,
      "step": 707
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8176339514527683,
      "learning_rate": 3.97621526033301e-05,
      "loss": 0.4381,
      "step": 708
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7352478845871343,
      "learning_rate": 3.976105745984742e-05,
      "loss": 0.4521,
      "step": 709
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7244324172454369,
      "learning_rate": 3.9759959816066575e-05,
      "loss": 0.3966,
      "step": 710
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8091998201591605,
      "learning_rate": 3.9758859672126455e-05,
      "loss": 0.4023,
      "step": 711
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.772083600231714,
      "learning_rate": 3.975775702816625e-05,
      "loss": 0.4155,
      "step": 712
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7369778546612566,
      "learning_rate": 3.975665188432548e-05,
      "loss": 0.3481,
      "step": 713
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9478752769963814,
      "learning_rate": 3.975554424074397e-05,
      "loss": 0.5124,
      "step": 714
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9533006474976025,
      "learning_rate": 3.975443409756188e-05,
      "loss": 0.5248,
      "step": 715
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7704006260724757,
      "learning_rate": 3.975332145491965e-05,
      "loss": 0.3521,
      "step": 716
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8149667853625703,
      "learning_rate": 3.9752206312958086e-05,
      "loss": 0.3601,
      "step": 717
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8175710737030942,
      "learning_rate": 3.975108867181826e-05,
      "loss": 0.4371,
      "step": 718
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7899516544155942,
      "learning_rate": 3.9749968531641594e-05,
      "loss": 0.4082,
      "step": 719
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8570226900130501,
      "learning_rate": 3.974884589256981e-05,
      "loss": 0.3944,
      "step": 720
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9830655365050049,
      "learning_rate": 3.974772075474496e-05,
      "loss": 0.4515,
      "step": 721
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8191672186657225,
      "learning_rate": 3.97465931183094e-05,
      "loss": 0.4095,
      "step": 722
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8742857470165107,
      "learning_rate": 3.974546298340579e-05,
      "loss": 0.3991,
      "step": 723
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7385873714951968,
      "learning_rate": 3.9744330350177156e-05,
      "loss": 0.3548,
      "step": 724
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7437597386339665,
      "learning_rate": 3.974319521876678e-05,
      "loss": 0.3535,
      "step": 725
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9061489048837175,
      "learning_rate": 3.974205758931828e-05,
      "loss": 0.437,
      "step": 726
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6882122858247355,
      "learning_rate": 3.974091746197563e-05,
      "loss": 0.3453,
      "step": 727
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7254883404037363,
      "learning_rate": 3.973977483688305e-05,
      "loss": 0.3604,
      "step": 728
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7676121750284994,
      "learning_rate": 3.973862971418513e-05,
      "loss": 0.3581,
      "step": 729
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.830418128895816,
      "learning_rate": 3.9737482094026764e-05,
      "loss": 0.4385,
      "step": 730
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6877092996532913,
      "learning_rate": 3.9736331976553154e-05,
      "loss": 0.4166,
      "step": 731
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8272879346655926,
      "learning_rate": 3.9735179361909803e-05,
      "loss": 0.4436,
      "step": 732
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0049495396827917,
      "learning_rate": 3.9734024250242564e-05,
      "loss": 0.3879,
      "step": 733
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7703604321529897,
      "learning_rate": 3.9732866641697586e-05,
      "loss": 0.3008,
      "step": 734
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8551568982682785,
      "learning_rate": 3.973170653642133e-05,
      "loss": 0.3793,
      "step": 735
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8550069598979876,
      "learning_rate": 3.9730543934560595e-05,
      "loss": 0.398,
      "step": 736
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8811450503467172,
      "learning_rate": 3.972937883626247e-05,
      "loss": 0.4356,
      "step": 737
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6686100028911549,
      "learning_rate": 3.9728211241674363e-05,
      "loss": 0.2951,
      "step": 738
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7879317084264726,
      "learning_rate": 3.972704115094403e-05,
      "loss": 0.4351,
      "step": 739
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8452377624754119,
      "learning_rate": 3.972586856421949e-05,
      "loss": 0.4758,
      "step": 740
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7554675917828586,
      "learning_rate": 3.9724693481649134e-05,
      "loss": 0.4134,
      "step": 741
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8598670937726027,
      "learning_rate": 3.9723515903381625e-05,
      "loss": 0.4508,
      "step": 742
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8079319332530365,
      "learning_rate": 3.9722335829565955e-05,
      "loss": 0.4001,
      "step": 743
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7432211253732091,
      "learning_rate": 3.9721153260351446e-05,
      "loss": 0.3633,
      "step": 744
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7635748916273694,
      "learning_rate": 3.971996819588771e-05,
      "loss": 0.4162,
      "step": 745
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9134161204270983,
      "learning_rate": 3.971878063632471e-05,
      "loss": 0.3784,
      "step": 746
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7186014187696611,
      "learning_rate": 3.971759058181269e-05,
      "loss": 0.3729,
      "step": 747
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7774807603178935,
      "learning_rate": 3.971639803250221e-05,
      "loss": 0.4337,
      "step": 748
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7224299308432479,
      "learning_rate": 3.971520298854419e-05,
      "loss": 0.3177,
      "step": 749
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7735434565552102,
      "learning_rate": 3.9714005450089815e-05,
      "loss": 0.4021,
      "step": 750
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8583503857248929,
      "learning_rate": 3.971280541729061e-05,
      "loss": 0.3301,
      "step": 751
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6263724992679699,
      "learning_rate": 3.971160289029841e-05,
      "loss": 0.3213,
      "step": 752
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.726769653650323,
      "learning_rate": 3.9710397869265364e-05,
      "loss": 0.3767,
      "step": 753
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8664066476365219,
      "learning_rate": 3.9709190354343936e-05,
      "loss": 0.3703,
      "step": 754
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6975022029500102,
      "learning_rate": 3.9707980345686924e-05,
      "loss": 0.3448,
      "step": 755
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6369118894875103,
      "learning_rate": 3.9706767843447417e-05,
      "loss": 0.3381,
      "step": 756
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5829089926032058,
      "learning_rate": 3.970555284777883e-05,
      "loss": 0.3056,
      "step": 757
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8357335849077621,
      "learning_rate": 3.970433535883489e-05,
      "loss": 0.4038,
      "step": 758
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9163609486185431,
      "learning_rate": 3.970311537676964e-05,
      "loss": 0.4754,
      "step": 759
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8544288829787227,
      "learning_rate": 3.970189290173744e-05,
      "loss": 0.4247,
      "step": 760
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7723366966596901,
      "learning_rate": 3.970066793389297e-05,
      "loss": 0.4092,
      "step": 761
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8365623064293849,
      "learning_rate": 3.969944047339122e-05,
      "loss": 0.3165,
      "step": 762
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8268536219064315,
      "learning_rate": 3.96982105203875e-05,
      "loss": 0.3276,
      "step": 763
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.968372148227198,
      "learning_rate": 3.969697807503742e-05,
      "loss": 0.3886,
      "step": 764
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.895316766059658,
      "learning_rate": 3.969574313749693e-05,
      "loss": 0.3945,
      "step": 765
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7392640208055445,
      "learning_rate": 3.969450570792227e-05,
      "loss": 0.3964,
      "step": 766
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7908308778878117,
      "learning_rate": 3.9693265786470016e-05,
      "loss": 0.3896,
      "step": 767
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8566097506183497,
      "learning_rate": 3.969202337329705e-05,
      "loss": 0.4392,
      "step": 768
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6322339085484171,
      "learning_rate": 3.969077846856057e-05,
      "loss": 0.2962,
      "step": 769
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8539239181826643,
      "learning_rate": 3.968953107241809e-05,
      "loss": 0.4008,
      "step": 770
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.859982085374054,
      "learning_rate": 3.9688281185027434e-05,
      "loss": 0.4318,
      "step": 771
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8618796868366281,
      "learning_rate": 3.9687028806546756e-05,
      "loss": 0.4331,
      "step": 772
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9366000942643986,
      "learning_rate": 3.96857739371345e-05,
      "loss": 0.4672,
      "step": 773
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7585798881042127,
      "learning_rate": 3.968451657694946e-05,
      "loss": 0.3669,
      "step": 774
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.807687223596606,
      "learning_rate": 3.9683256726150707e-05,
      "loss": 0.3918,
      "step": 775
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7919770979002362,
      "learning_rate": 3.9681994384897654e-05,
      "loss": 0.3846,
      "step": 776
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7443223866430635,
      "learning_rate": 3.9680729553350016e-05,
      "loss": 0.3592,
      "step": 777
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.754575758217912,
      "learning_rate": 3.967946223166784e-05,
      "loss": 0.356,
      "step": 778
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7114377619690837,
      "learning_rate": 3.967819242001146e-05,
      "loss": 0.3294,
      "step": 779
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7754772916158532,
      "learning_rate": 3.967692011854155e-05,
      "loss": 0.3101,
      "step": 780
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.034931201583225,
      "learning_rate": 3.967564532741909e-05,
      "loss": 0.5161,
      "step": 781
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7108248527068666,
      "learning_rate": 3.967436804680537e-05,
      "loss": 0.2893,
      "step": 782
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8280630268443923,
      "learning_rate": 3.9673088276862e-05,
      "loss": 0.4229,
      "step": 783
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8426574415031907,
      "learning_rate": 3.9671806017750915e-05,
      "loss": 0.4099,
      "step": 784
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8443906259396801,
      "learning_rate": 3.967052126963435e-05,
      "loss": 0.3673,
      "step": 785
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7433810625305357,
      "learning_rate": 3.966923403267485e-05,
      "loss": 0.3935,
      "step": 786
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.81018830606576,
      "learning_rate": 3.966794430703528e-05,
      "loss": 0.333,
      "step": 787
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7762348182164174,
      "learning_rate": 3.9666652092878856e-05,
      "loss": 0.3622,
      "step": 788
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8275345550767284,
      "learning_rate": 3.966535739036905e-05,
      "loss": 0.387,
      "step": 789
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7738081689270789,
      "learning_rate": 3.966406019966968e-05,
      "loss": 0.3739,
      "step": 790
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8311358366277135,
      "learning_rate": 3.9662760520944886e-05,
      "loss": 0.3782,
      "step": 791
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9774607685148268,
      "learning_rate": 3.9661458354359105e-05,
      "loss": 0.3976,
      "step": 792
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9264658396641045,
      "learning_rate": 3.966015370007709e-05,
      "loss": 0.5481,
      "step": 793
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7644249210603022,
      "learning_rate": 3.9658846558263925e-05,
      "loss": 0.3557,
      "step": 794
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6324907591676214,
      "learning_rate": 3.965753692908499e-05,
      "loss": 0.3275,
      "step": 795
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.765082770650668,
      "learning_rate": 3.965622481270599e-05,
      "loss": 0.3816,
      "step": 796
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6620841682609669,
      "learning_rate": 3.9654910209292943e-05,
      "loss": 0.3236,
      "step": 797
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7204440511864142,
      "learning_rate": 3.9653593119012185e-05,
      "loss": 0.362,
      "step": 798
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7578089212549397,
      "learning_rate": 3.9652273542030355e-05,
      "loss": 0.3905,
      "step": 799
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7888588217555922,
      "learning_rate": 3.965095147851442e-05,
      "loss": 0.4339,
      "step": 800
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8138751716853801,
      "learning_rate": 3.9649626928631655e-05,
      "loss": 0.4155,
      "step": 801
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6165424549384506,
      "learning_rate": 3.9648299892549654e-05,
      "loss": 0.2941,
      "step": 802
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7059527860475053,
      "learning_rate": 3.964697037043632e-05,
      "loss": 0.3268,
      "step": 803
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8911116090523694,
      "learning_rate": 3.964563836245987e-05,
      "loss": 0.4004,
      "step": 804
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8497829917570302,
      "learning_rate": 3.964430386878883e-05,
      "loss": 0.4635,
      "step": 805
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7549620829780918,
      "learning_rate": 3.964296688959208e-05,
      "loss": 0.4025,
      "step": 806
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8601342401971424,
      "learning_rate": 3.9641627425038745e-05,
      "loss": 0.317,
      "step": 807
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9082765282941799,
      "learning_rate": 3.964028547529832e-05,
      "loss": 0.446,
      "step": 808
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6087626290235267,
      "learning_rate": 3.9638941040540606e-05,
      "loss": 0.2865,
      "step": 809
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.900223921364458,
      "learning_rate": 3.9637594120935697e-05,
      "loss": 0.4247,
      "step": 810
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7034242840766348,
      "learning_rate": 3.963624471665402e-05,
      "loss": 0.3271,
      "step": 811
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6772115928261491,
      "learning_rate": 3.9634892827866306e-05,
      "loss": 0.3007,
      "step": 812
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6664765628334043,
      "learning_rate": 3.963353845474361e-05,
      "loss": 0.378,
      "step": 813
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.782829518999284,
      "learning_rate": 3.9632181597457296e-05,
      "loss": 0.3665,
      "step": 814
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8306568596601841,
      "learning_rate": 3.963082225617903e-05,
      "loss": 0.4402,
      "step": 815
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8407522566098939,
      "learning_rate": 3.9629460431080825e-05,
      "loss": 0.3815,
      "step": 816
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7948156187382217,
      "learning_rate": 3.962809612233497e-05,
      "loss": 0.3928,
      "step": 817
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7308868191009538,
      "learning_rate": 3.96267293301141e-05,
      "loss": 0.3259,
      "step": 818
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8296952744746502,
      "learning_rate": 3.9625360054591144e-05,
      "loss": 0.3351,
      "step": 819
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6594169585270185,
      "learning_rate": 3.962398829593935e-05,
      "loss": 0.323,
      "step": 820
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7775576503269125,
      "learning_rate": 3.962261405433229e-05,
      "loss": 0.3453,
      "step": 821
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7960862763028763,
      "learning_rate": 3.962123732994383e-05,
      "loss": 0.4011,
      "step": 822
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9116213225909428,
      "learning_rate": 3.961985812294817e-05,
      "loss": 0.3732,
      "step": 823
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7696309291356568,
      "learning_rate": 3.961847643351981e-05,
      "loss": 0.333,
      "step": 824
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8680071694346081,
      "learning_rate": 3.961709226183359e-05,
      "loss": 0.3697,
      "step": 825
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9742102407712474,
      "learning_rate": 3.961570560806461e-05,
      "loss": 0.4396,
      "step": 826
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9699224944464142,
      "learning_rate": 3.961431647238835e-05,
      "loss": 0.48,
      "step": 827
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.826206035223429,
      "learning_rate": 3.9612924854980556e-05,
      "loss": 0.3786,
      "step": 828
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7994337700661437,
      "learning_rate": 3.9611530756017306e-05,
      "loss": 0.3527,
      "step": 829
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7616547386465274,
      "learning_rate": 3.9610134175675e-05,
      "loss": 0.3833,
      "step": 830
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6806056737826948,
      "learning_rate": 3.960873511413033e-05,
      "loss": 0.3292,
      "step": 831
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6463923547259581,
      "learning_rate": 3.960733357156033e-05,
      "loss": 0.3016,
      "step": 832
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7202256851673425,
      "learning_rate": 3.9605929548142314e-05,
      "loss": 0.3773,
      "step": 833
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8216163808537975,
      "learning_rate": 3.960452304405394e-05,
      "loss": 0.4243,
      "step": 834
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6232743759713391,
      "learning_rate": 3.960311405947317e-05,
      "loss": 0.289,
      "step": 835
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.820963065711702,
      "learning_rate": 3.960170259457826e-05,
      "loss": 0.2993,
      "step": 836
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7917922740138276,
      "learning_rate": 3.960028864954782e-05,
      "loss": 0.3405,
      "step": 837
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6907357882191566,
      "learning_rate": 3.959887222456075e-05,
      "loss": 0.2672,
      "step": 838
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7582841172508993,
      "learning_rate": 3.9597453319796245e-05,
      "loss": 0.2889,
      "step": 839
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6911531836534004,
      "learning_rate": 3.959603193543385e-05,
      "loss": 0.3292,
      "step": 840
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.816896610747596,
      "learning_rate": 3.95946080716534e-05,
      "loss": 0.357,
      "step": 841
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8281384592941847,
      "learning_rate": 3.959318172863506e-05,
      "loss": 0.3597,
      "step": 842
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8258436195690133,
      "learning_rate": 3.95917529065593e-05,
      "loss": 0.4532,
      "step": 843
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7230337445732571,
      "learning_rate": 3.95903216056069e-05,
      "loss": 0.3236,
      "step": 844
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7858822630626577,
      "learning_rate": 3.958888782595895e-05,
      "loss": 0.3141,
      "step": 845
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7437788917635865,
      "learning_rate": 3.958745156779688e-05,
      "loss": 0.3863,
      "step": 846
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6731286883098178,
      "learning_rate": 3.9586012831302396e-05,
      "loss": 0.3209,
      "step": 847
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8627273025091535,
      "learning_rate": 3.9584571616657544e-05,
      "loss": 0.4197,
      "step": 848
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6090186373093468,
      "learning_rate": 3.958312792404468e-05,
      "loss": 0.2961,
      "step": 849
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8078390829113737,
      "learning_rate": 3.958168175364646e-05,
      "loss": 0.3597,
      "step": 850
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6844922509629057,
      "learning_rate": 3.9580233105645874e-05,
      "loss": 0.2671,
      "step": 851
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6578007495401484,
      "learning_rate": 3.957878198022621e-05,
      "loss": 0.2385,
      "step": 852
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7632581365672044,
      "learning_rate": 3.957732837757107e-05,
      "loss": 0.3571,
      "step": 853
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7806615520682209,
      "learning_rate": 3.957587229786437e-05,
      "loss": 0.3126,
      "step": 854
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7183221704702855,
      "learning_rate": 3.957441374129035e-05,
      "loss": 0.3251,
      "step": 855
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6635868837608979,
      "learning_rate": 3.9572952708033564e-05,
      "loss": 0.2768,
      "step": 856
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7650985464811946,
      "learning_rate": 3.9571489198278855e-05,
      "loss": 0.3336,
      "step": 857
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7593707551562814,
      "learning_rate": 3.9570023212211405e-05,
      "loss": 0.3748,
      "step": 858
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8188017603114055,
      "learning_rate": 3.95685547500167e-05,
      "loss": 0.406,
      "step": 859
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7126570227413258,
      "learning_rate": 3.956708381188054e-05,
      "loss": 0.3344,
      "step": 860
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6770653404388245,
      "learning_rate": 3.956561039798903e-05,
      "loss": 0.2968,
      "step": 861
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8083508997578231,
      "learning_rate": 3.95641345085286e-05,
      "loss": 0.366,
      "step": 862
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8739611044026313,
      "learning_rate": 3.956265614368599e-05,
      "loss": 0.3072,
      "step": 863
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8994228101671509,
      "learning_rate": 3.956117530364826e-05,
      "loss": 0.3689,
      "step": 864
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8499506473238314,
      "learning_rate": 3.955969198860276e-05,
      "loss": 0.3983,
      "step": 865
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7764093735467574,
      "learning_rate": 3.955820619873719e-05,
      "loss": 0.3578,
      "step": 866
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7616638946343639,
      "learning_rate": 3.9556717934239516e-05,
      "loss": 0.3146,
      "step": 867
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7857665162251634,
      "learning_rate": 3.955522719529807e-05,
      "loss": 0.396,
      "step": 868
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8501695211447731,
      "learning_rate": 3.955373398210144e-05,
      "loss": 0.3346,
      "step": 869
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7679752943316975,
      "learning_rate": 3.9552238294838584e-05,
      "loss": 0.3314,
      "step": 870
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6558055962421216,
      "learning_rate": 3.9550740133698726e-05,
      "loss": 0.288,
      "step": 871
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7313453595836844,
      "learning_rate": 3.954923949887144e-05,
      "loss": 0.3122,
      "step": 872
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7218615311856481,
      "learning_rate": 3.954773639054659e-05,
      "loss": 0.2875,
      "step": 873
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.739149844147705,
      "learning_rate": 3.954623080891435e-05,
      "loss": 0.259,
      "step": 874
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7460763340060723,
      "learning_rate": 3.9544722754165225e-05,
      "loss": 0.291,
      "step": 875
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8088352008732254,
      "learning_rate": 3.954321222649003e-05,
      "loss": 0.3429,
      "step": 876
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7909876688973838,
      "learning_rate": 3.954169922607987e-05,
      "loss": 0.2892,
      "step": 877
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.877938135268879,
      "learning_rate": 3.95401837531262e-05,
      "loss": 0.2588,
      "step": 878
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7311048950598087,
      "learning_rate": 3.953866580782075e-05,
      "loss": 0.2772,
      "step": 879
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8246951421374188,
      "learning_rate": 3.953714539035558e-05,
      "loss": 0.3546,
      "step": 880
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6775861462584851,
      "learning_rate": 3.953562250092308e-05,
      "loss": 0.2771,
      "step": 881
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6792306131748974,
      "learning_rate": 3.9534097139715926e-05,
      "loss": 0.3082,
      "step": 882
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8128290977088691,
      "learning_rate": 3.9532569306927115e-05,
      "loss": 0.3386,
      "step": 883
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.709903510078536,
      "learning_rate": 3.9531039002749955e-05,
      "loss": 0.3249,
      "step": 884
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7870809015088726,
      "learning_rate": 3.9529506227378085e-05,
      "loss": 0.4089,
      "step": 885
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7636376102497198,
      "learning_rate": 3.952797098100543e-05,
      "loss": 0.3527,
      "step": 886
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8253743175045637,
      "learning_rate": 3.952643326382624e-05,
      "loss": 0.3307,
      "step": 887
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6402923371550692,
      "learning_rate": 3.952489307603507e-05,
      "loss": 0.2392,
      "step": 888
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.764620647680578,
      "learning_rate": 3.9523350417826816e-05,
      "loss": 0.3716,
      "step": 889
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.064539858582937,
      "learning_rate": 3.9521805289396645e-05,
      "loss": 0.2258,
      "step": 890
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7584962807767207,
      "learning_rate": 3.952025769094006e-05,
      "loss": 0.3336,
      "step": 891
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9684532695519109,
      "learning_rate": 3.951870762265288e-05,
      "loss": 0.3657,
      "step": 892
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7111334059203872,
      "learning_rate": 3.951715508473124e-05,
      "loss": 0.2955,
      "step": 893
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7075510379997415,
      "learning_rate": 3.9515600077371545e-05,
      "loss": 0.2746,
      "step": 894
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8876017068960823,
      "learning_rate": 3.9514042600770576e-05,
      "loss": 0.4418,
      "step": 895
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6422516245511461,
      "learning_rate": 3.951248265512538e-05,
      "loss": 0.2797,
      "step": 896
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7476030033806428,
      "learning_rate": 3.951092024063333e-05,
      "loss": 0.3639,
      "step": 897
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8450392304291887,
      "learning_rate": 3.950935535749213e-05,
      "loss": 0.347,
      "step": 898
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7068815547055709,
      "learning_rate": 3.9507788005899756e-05,
      "loss": 0.326,
      "step": 899
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8299403536288089,
      "learning_rate": 3.950621818605453e-05,
      "loss": 0.2721,
      "step": 900
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6629675870512527,
      "learning_rate": 3.950464589815508e-05,
      "loss": 0.2502,
      "step": 901
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7484625714588448,
      "learning_rate": 3.950307114240034e-05,
      "loss": 0.314,
      "step": 902
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8145970645017154,
      "learning_rate": 3.950149391898955e-05,
      "loss": 0.3533,
      "step": 903
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8147193235818403,
      "learning_rate": 3.9499914228122286e-05,
      "loss": 0.3479,
      "step": 904
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8505544733300423,
      "learning_rate": 3.9498332069998405e-05,
      "loss": 0.3788,
      "step": 905
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7673949588504672,
      "learning_rate": 3.9496747444818105e-05,
      "loss": 0.2771,
      "step": 906
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7690665686483074,
      "learning_rate": 3.9495160352781875e-05,
      "loss": 0.2982,
      "step": 907
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7372941376392371,
      "learning_rate": 3.9493570794090524e-05,
      "loss": 0.2931,
      "step": 908
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8846684080359255,
      "learning_rate": 3.9491978768945184e-05,
      "loss": 0.3135,
      "step": 909
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7435395624272256,
      "learning_rate": 3.9490384277547266e-05,
      "loss": 0.3314,
      "step": 910
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6994417357116672,
      "learning_rate": 3.948878732009854e-05,
      "loss": 0.3881,
      "step": 911
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6923609462454279,
      "learning_rate": 3.9487187896801054e-05,
      "loss": 0.288,
      "step": 912
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7272337407261249,
      "learning_rate": 3.9485586007857174e-05,
      "loss": 0.3146,
      "step": 913
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8734351197348492,
      "learning_rate": 3.9483981653469586e-05,
      "loss": 0.4029,
      "step": 914
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8283377320181136,
      "learning_rate": 3.948237483384128e-05,
      "loss": 0.3308,
      "step": 915
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5984747274985789,
      "learning_rate": 3.948076554917556e-05,
      "loss": 0.249,
      "step": 916
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8565072849518371,
      "learning_rate": 3.947915379967605e-05,
      "loss": 0.321,
      "step": 917
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.814407530315674,
      "learning_rate": 3.9477539585546676e-05,
      "loss": 0.327,
      "step": 918
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9619628893857703,
      "learning_rate": 3.947592290699168e-05,
      "loss": 0.4363,
      "step": 919
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6454757705561542,
      "learning_rate": 3.9474303764215606e-05,
      "loss": 0.2605,
      "step": 920
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9926341095455815,
      "learning_rate": 3.947268215742333e-05,
      "loss": 0.3469,
      "step": 921
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7241534009171295,
      "learning_rate": 3.9471058086820024e-05,
      "loss": 0.3321,
      "step": 922
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7240218996824527,
      "learning_rate": 3.946943155261117e-05,
      "loss": 0.2985,
      "step": 923
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7046570934155437,
      "learning_rate": 3.9467802555002584e-05,
      "loss": 0.2776,
      "step": 924
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7198740007168696,
      "learning_rate": 3.9466171094200356e-05,
      "loss": 0.2917,
      "step": 925
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7669508605967125,
      "learning_rate": 3.946453717041093e-05,
      "loss": 0.3446,
      "step": 926
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.907258327819075,
      "learning_rate": 3.946290078384103e-05,
      "loss": 0.3825,
      "step": 927
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7321825373925491,
      "learning_rate": 3.94612619346977e-05,
      "loss": 0.3469,
      "step": 928
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7435893822990791,
      "learning_rate": 3.9459620623188294e-05,
      "loss": 0.3749,
      "step": 929
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6541312392374807,
      "learning_rate": 3.94579768495205e-05,
      "loss": 0.2811,
      "step": 930
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7122762462564227,
      "learning_rate": 3.9456330613902286e-05,
      "loss": 0.2973,
      "step": 931
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6039775821065473,
      "learning_rate": 3.9454681916541936e-05,
      "loss": 0.223,
      "step": 932
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7453116266977244,
      "learning_rate": 3.945303075764807e-05,
      "loss": 0.2829,
      "step": 933
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.812096275381738,
      "learning_rate": 3.94513771374296e-05,
      "loss": 0.3244,
      "step": 934
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7949422697782987,
      "learning_rate": 3.9449721056095746e-05,
      "loss": 0.3314,
      "step": 935
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6782805910351128,
      "learning_rate": 3.9448062513856056e-05,
      "loss": 0.2472,
      "step": 936
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7304765017501041,
      "learning_rate": 3.9446401510920365e-05,
      "loss": 0.277,
      "step": 937
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8840028620129856,
      "learning_rate": 3.944473804749885e-05,
      "loss": 0.3494,
      "step": 938
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7815671658443387,
      "learning_rate": 3.9443072123801975e-05,
      "loss": 0.2972,
      "step": 939
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.687822266387582,
      "learning_rate": 3.944140374004052e-05,
      "loss": 0.2074,
      "step": 940
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7464770386084335,
      "learning_rate": 3.943973289642559e-05,
      "loss": 0.3356,
      "step": 941
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.763183867099979,
      "learning_rate": 3.9438059593168586e-05,
      "loss": 0.3204,
      "step": 942
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6546027080826027,
      "learning_rate": 3.9436383830481226e-05,
      "loss": 0.2803,
      "step": 943
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7755412764936565,
      "learning_rate": 3.943470560857553e-05,
      "loss": 0.3101,
      "step": 944
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7884411395424002,
      "learning_rate": 3.9433024927663856e-05,
      "loss": 0.2722,
      "step": 945
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6560500158136222,
      "learning_rate": 3.943134178795883e-05,
      "loss": 0.2805,
      "step": 946
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8125042548435016,
      "learning_rate": 3.942965618967344e-05,
      "loss": 0.3715,
      "step": 947
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.849678475211711,
      "learning_rate": 3.942796813302094e-05,
      "loss": 0.4147,
      "step": 948
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7345623020992047,
      "learning_rate": 3.942627761821492e-05,
      "loss": 0.3156,
      "step": 949
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6994124842085044,
      "learning_rate": 3.942458464546928e-05,
      "loss": 0.3267,
      "step": 950
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7415999646856553,
      "learning_rate": 3.9422889214998214e-05,
      "loss": 0.3421,
      "step": 951
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.674673418231147,
      "learning_rate": 3.942119132701625e-05,
      "loss": 0.325,
      "step": 952
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6002050913527098,
      "learning_rate": 3.941949098173821e-05,
      "loss": 0.2828,
      "step": 953
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8729606090826311,
      "learning_rate": 3.9417788179379245e-05,
      "loss": 0.3874,
      "step": 954
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6830203457981324,
      "learning_rate": 3.941608292015478e-05,
      "loss": 0.2398,
      "step": 955
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6871211134974333,
      "learning_rate": 3.941437520428061e-05,
      "loss": 0.2924,
      "step": 956
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.661933200319762,
      "learning_rate": 3.941266503197277e-05,
      "loss": 0.2484,
      "step": 957
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5907171994860796,
      "learning_rate": 3.941095240344766e-05,
      "loss": 0.2307,
      "step": 958
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7480759976049202,
      "learning_rate": 3.9409237318921975e-05,
      "loss": 0.3177,
      "step": 959
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7428933250738426,
      "learning_rate": 3.940751977861272e-05,
      "loss": 0.3098,
      "step": 960
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8061759663383855,
      "learning_rate": 3.9405799782737196e-05,
      "loss": 0.344,
      "step": 961
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6340489023855357,
      "learning_rate": 3.9404077331513044e-05,
      "loss": 0.238,
      "step": 962
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6411403351048823,
      "learning_rate": 3.940235242515819e-05,
      "loss": 0.2586,
      "step": 963
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6791197718674111,
      "learning_rate": 3.940062506389089e-05,
      "loss": 0.2914,
      "step": 964
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7169401353081453,
      "learning_rate": 3.939889524792968e-05,
      "loss": 0.2625,
      "step": 965
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.713993170980712,
      "learning_rate": 3.9397162977493455e-05,
      "loss": 0.2781,
      "step": 966
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8212851434675372,
      "learning_rate": 3.939542825280137e-05,
      "loss": 0.3637,
      "step": 967
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7417147686893449,
      "learning_rate": 3.939369107407293e-05,
      "loss": 0.2795,
      "step": 968
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6280562538749723,
      "learning_rate": 3.939195144152792e-05,
      "loss": 0.2363,
      "step": 969
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7004002865637144,
      "learning_rate": 3.939020935538647e-05,
      "loss": 0.2463,
      "step": 970
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6972561576478352,
      "learning_rate": 3.938846481586898e-05,
      "loss": 0.2859,
      "step": 971
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.572818996311442,
      "learning_rate": 3.938671782319619e-05,
      "loss": 0.2364,
      "step": 972
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7532835449234433,
      "learning_rate": 3.938496837758914e-05,
      "loss": 0.3212,
      "step": 973
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6375030648400928,
      "learning_rate": 3.938321647926918e-05,
      "loss": 0.3274,
      "step": 974
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6013369447017474,
      "learning_rate": 3.938146212845797e-05,
      "loss": 0.2656,
      "step": 975
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7198242369669776,
      "learning_rate": 3.937970532537749e-05,
      "loss": 0.3267,
      "step": 976
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6414430094325005,
      "learning_rate": 3.9377946070250006e-05,
      "loss": 0.2564,
      "step": 977
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8464511796175019,
      "learning_rate": 3.937618436329813e-05,
      "loss": 0.3334,
      "step": 978
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7920613099965732,
      "learning_rate": 3.937442020474475e-05,
      "loss": 0.2926,
      "step": 979
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.80756319873438,
      "learning_rate": 3.937265359481309e-05,
      "loss": 0.303,
      "step": 980
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6979425387545215,
      "learning_rate": 3.9370884533726664e-05,
      "loss": 0.2471,
      "step": 981
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7290781830279354,
      "learning_rate": 3.936911302170931e-05,
      "loss": 0.2995,
      "step": 982
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8006790468077764,
      "learning_rate": 3.936733905898517e-05,
      "loss": 0.3339,
      "step": 983
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7596170628296051,
      "learning_rate": 3.936556264577869e-05,
      "loss": 0.3366,
      "step": 984
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6322558745340952,
      "learning_rate": 3.936378378231465e-05,
      "loss": 0.2878,
      "step": 985
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6589092688958328,
      "learning_rate": 3.9362002468818105e-05,
      "loss": 0.2734,
      "step": 986
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5840876051077623,
      "learning_rate": 3.9360218705514456e-05,
      "loss": 0.2268,
      "step": 987
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8223636595939524,
      "learning_rate": 3.935843249262939e-05,
      "loss": 0.358,
      "step": 988
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8171113542196695,
      "learning_rate": 3.93566438303889e-05,
      "loss": 0.3505,
      "step": 989
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6316890162495712,
      "learning_rate": 3.9354852719019306e-05,
      "loss": 0.221,
      "step": 990
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7426172317409356,
      "learning_rate": 3.935305915874724e-05,
      "loss": 0.3201,
      "step": 991
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8140781920928644,
      "learning_rate": 3.935126314979962e-05,
      "loss": 0.2949,
      "step": 992
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6333975736345722,
      "learning_rate": 3.93494646924037e-05,
      "loss": 0.2761,
      "step": 993
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6260929088810602,
      "learning_rate": 3.934766378678704e-05,
      "loss": 0.2609,
      "step": 994
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7287030835368725,
      "learning_rate": 3.934586043317748e-05,
      "loss": 0.2768,
      "step": 995
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6525172385500014,
      "learning_rate": 3.93440546318032e-05,
      "loss": 0.2232,
      "step": 996
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7513320141168345,
      "learning_rate": 3.934224638289268e-05,
      "loss": 0.2951,
      "step": 997
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5344745721683354,
      "learning_rate": 3.934043568667473e-05,
      "loss": 0.2579,
      "step": 998
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7747366042306933,
      "learning_rate": 3.9338622543378436e-05,
      "loss": 0.3363,
      "step": 999
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7486739356581735,
      "learning_rate": 3.933680695323321e-05,
      "loss": 0.284,
      "step": 1000
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7217638846747562,
      "learning_rate": 3.933498891646877e-05,
      "loss": 0.3034,
      "step": 1001
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5502334467760756,
      "learning_rate": 3.9333168433315144e-05,
      "loss": 0.2358,
      "step": 1002
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6679734793155404,
      "learning_rate": 3.933134550400268e-05,
      "loss": 0.2488,
      "step": 1003
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.744859003255334,
      "learning_rate": 3.932952012876203e-05,
      "loss": 0.3261,
      "step": 1004
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4895547662341082,
      "learning_rate": 3.932769230782414e-05,
      "loss": 0.2033,
      "step": 1005
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7013920515343732,
      "learning_rate": 3.9325862041420275e-05,
      "loss": 0.3107,
      "step": 1006
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6422877249739573,
      "learning_rate": 3.932402932978203e-05,
      "loss": 0.2577,
      "step": 1007
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6516262837391291,
      "learning_rate": 3.9322194173141284e-05,
      "loss": 0.2396,
      "step": 1008
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8268493327729397,
      "learning_rate": 3.932035657173023e-05,
      "loss": 0.2831,
      "step": 1009
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.709380236366876,
      "learning_rate": 3.931851652578137e-05,
      "loss": 0.2742,
      "step": 1010
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6892083573752351,
      "learning_rate": 3.931667403552753e-05,
      "loss": 0.2584,
      "step": 1011
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7666191730784749,
      "learning_rate": 3.9314829101201814e-05,
      "loss": 0.2843,
      "step": 1012
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6516448748726327,
      "learning_rate": 3.931298172303768e-05,
      "loss": 0.2307,
      "step": 1013
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.591878098557286,
      "learning_rate": 3.9311131901268855e-05,
      "loss": 0.204,
      "step": 1014
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8590507242305281,
      "learning_rate": 3.9309279636129396e-05,
      "loss": 0.3197,
      "step": 1015
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7266935004812294,
      "learning_rate": 3.930742492785366e-05,
      "loss": 0.2782,
      "step": 1016
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6321222756077973,
      "learning_rate": 3.930556777667632e-05,
      "loss": 0.2293,
      "step": 1017
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7886974132195063,
      "learning_rate": 3.930370818283235e-05,
      "loss": 0.2876,
      "step": 1018
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5970742377444473,
      "learning_rate": 3.9301846146557045e-05,
      "loss": 0.2525,
      "step": 1019
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6257482817653303,
      "learning_rate": 3.9299981668085997e-05,
      "loss": 0.287,
      "step": 1020
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7252019633782599,
      "learning_rate": 3.929811474765512e-05,
      "loss": 0.2785,
      "step": 1021
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6737315675411617,
      "learning_rate": 3.929624538550061e-05,
      "loss": 0.3161,
      "step": 1022
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.733700625238442,
      "learning_rate": 3.929437358185901e-05,
      "loss": 0.3127,
      "step": 1023
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7339977756513314,
      "learning_rate": 3.929249933696715e-05,
      "loss": 0.3156,
      "step": 1024
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7095123930985195,
      "learning_rate": 3.9290622651062164e-05,
      "loss": 0.2758,
      "step": 1025
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6859423157125364,
      "learning_rate": 3.92887435243815e-05,
      "loss": 0.2841,
      "step": 1026
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7578136404756439,
      "learning_rate": 3.928686195716294e-05,
      "loss": 0.3359,
      "step": 1027
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.630251163151722,
      "learning_rate": 3.928497794964452e-05,
      "loss": 0.2465,
      "step": 1028
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6332205645861801,
      "learning_rate": 3.928309150206464e-05,
      "loss": 0.3327,
      "step": 1029
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6105684429125092,
      "learning_rate": 3.928120261466198e-05,
      "loss": 0.2322,
      "step": 1030
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5737584403503849,
      "learning_rate": 3.9279311287675535e-05,
      "loss": 0.2143,
      "step": 1031
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.588643450642452,
      "learning_rate": 3.92774175213446e-05,
      "loss": 0.209,
      "step": 1032
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8643382234397027,
      "learning_rate": 3.92755213159088e-05,
      "loss": 0.2968,
      "step": 1033
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6109670724565266,
      "learning_rate": 3.927362267160804e-05,
      "loss": 0.2428,
      "step": 1034
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.784660372551906,
      "learning_rate": 3.927172158868257e-05,
      "loss": 0.1984,
      "step": 1035
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8481179663100128,
      "learning_rate": 3.92698180673729e-05,
      "loss": 0.322,
      "step": 1036
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6891565738740009,
      "learning_rate": 3.92679121079199e-05,
      "loss": 0.2695,
      "step": 1037
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8321120889999258,
      "learning_rate": 3.9266003710564706e-05,
      "loss": 0.4166,
      "step": 1038
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6893674456821418,
      "learning_rate": 3.92640928755488e-05,
      "loss": 0.3054,
      "step": 1039
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6900694904372378,
      "learning_rate": 3.9262179603113934e-05,
      "loss": 0.1942,
      "step": 1040
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6848669783360392,
      "learning_rate": 3.9260263893502204e-05,
      "loss": 0.2819,
      "step": 1041
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5351295742988001,
      "learning_rate": 3.925834574695599e-05,
      "loss": 0.2224,
      "step": 1042
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8952740246559334,
      "learning_rate": 3.925642516371799e-05,
      "loss": 0.3992,
      "step": 1043
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6412168420617952,
      "learning_rate": 3.9254502144031204e-05,
      "loss": 0.2896,
      "step": 1044
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6373657730386285,
      "learning_rate": 3.925257668813895e-05,
      "loss": 0.2555,
      "step": 1045
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6261593556238331,
      "learning_rate": 3.925064879628485e-05,
      "loss": 0.2483,
      "step": 1046
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8885453018764059,
      "learning_rate": 3.924871846871283e-05,
      "loss": 0.3659,
      "step": 1047
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6511353361774384,
      "learning_rate": 3.924678570566714e-05,
      "loss": 0.2772,
      "step": 1048
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7091998989090091,
      "learning_rate": 3.9244850507392296e-05,
      "loss": 0.2509,
      "step": 1049
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7040334554845372,
      "learning_rate": 3.9242912874133186e-05,
      "loss": 0.2606,
      "step": 1050
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7719942363736338,
      "learning_rate": 3.924097280613495e-05,
      "loss": 0.3191,
      "step": 1051
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6697367865671184,
      "learning_rate": 3.9239030303643074e-05,
      "loss": 0.2613,
      "step": 1052
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8264958172673278,
      "learning_rate": 3.923708536690332e-05,
      "loss": 0.3564,
      "step": 1053
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6571493344328319,
      "learning_rate": 3.9235137996161786e-05,
      "loss": 0.2708,
      "step": 1054
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.710064068509379,
      "learning_rate": 3.9233188191664856e-05,
      "loss": 0.2523,
      "step": 1055
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.608062798513221,
      "learning_rate": 3.9231235953659244e-05,
      "loss": 0.2325,
      "step": 1056
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6773015819830156,
      "learning_rate": 3.922928128239195e-05,
      "loss": 0.2123,
      "step": 1057
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6601524804363184,
      "learning_rate": 3.9227324178110295e-05,
      "loss": 0.2513,
      "step": 1058
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5563992428679837,
      "learning_rate": 3.9225364641061904e-05,
      "loss": 0.2206,
      "step": 1059
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6595045405413055,
      "learning_rate": 3.922340267149472e-05,
      "loss": 0.1919,
      "step": 1060
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7530465239480391,
      "learning_rate": 3.922143826965697e-05,
      "loss": 0.2667,
      "step": 1061
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9347659616752685,
      "learning_rate": 3.9219471435797205e-05,
      "loss": 0.3505,
      "step": 1062
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.622338229803054,
      "learning_rate": 3.921750217016429e-05,
      "loss": 0.2342,
      "step": 1063
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7586102563415626,
      "learning_rate": 3.921553047300739e-05,
      "loss": 0.3498,
      "step": 1064
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6776757998259934,
      "learning_rate": 3.9213556344575964e-05,
      "loss": 0.3132,
      "step": 1065
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.809506918860515,
      "learning_rate": 3.9211579785119804e-05,
      "loss": 0.3646,
      "step": 1066
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7208531334717293,
      "learning_rate": 3.920960079488899e-05,
      "loss": 0.2787,
      "step": 1067
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6928812662013678,
      "learning_rate": 3.9207619374133917e-05,
      "loss": 0.3179,
      "step": 1068
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5637700524749805,
      "learning_rate": 3.920563552310529e-05,
      "loss": 0.2183,
      "step": 1069
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5474515328227216,
      "learning_rate": 3.920364924205412e-05,
      "loss": 0.2715,
      "step": 1070
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6813817885429322,
      "learning_rate": 3.920166053123172e-05,
      "loss": 0.2607,
      "step": 1071
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6183924441432396,
      "learning_rate": 3.9199669390889725e-05,
      "loss": 0.2305,
      "step": 1072
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6323170311714446,
      "learning_rate": 3.919767582128005e-05,
      "loss": 0.1967,
      "step": 1073
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6699121991673812,
      "learning_rate": 3.919567982265495e-05,
      "loss": 0.2598,
      "step": 1074
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8344633348545905,
      "learning_rate": 3.919368139526697e-05,
      "loss": 0.3114,
      "step": 1075
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8664136991189226,
      "learning_rate": 3.9191680539368956e-05,
      "loss": 0.2834,
      "step": 1076
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8080766286630513,
      "learning_rate": 3.918967725521407e-05,
      "loss": 0.2887,
      "step": 1077
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9076625076317271,
      "learning_rate": 3.9187671543055785e-05,
      "loss": 0.286,
      "step": 1078
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8725319865246761,
      "learning_rate": 3.918566340314788e-05,
      "loss": 0.3357,
      "step": 1079
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6330280172436185,
      "learning_rate": 3.918365283574443e-05,
      "loss": 0.2758,
      "step": 1080
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6839189028598982,
      "learning_rate": 3.9181639841099825e-05,
      "loss": 0.2807,
      "step": 1081
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6554614279227786,
      "learning_rate": 3.9179624419468766e-05,
      "loss": 0.3104,
      "step": 1082
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6251213432774099,
      "learning_rate": 3.9177606571106265e-05,
      "loss": 0.2422,
      "step": 1083
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7579948756414638,
      "learning_rate": 3.917558629626762e-05,
      "loss": 0.3519,
      "step": 1084
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5614815074358583,
      "learning_rate": 3.917356359520846e-05,
      "loss": 0.2381,
      "step": 1085
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5754216369020166,
      "learning_rate": 3.917153846818471e-05,
      "loss": 0.251,
      "step": 1086
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7220614072886674,
      "learning_rate": 3.916951091545258e-05,
      "loss": 0.2917,
      "step": 1087
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6868942149410308,
      "learning_rate": 3.916748093726864e-05,
      "loss": 0.2941,
      "step": 1088
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6636968166790539,
      "learning_rate": 3.9165448533889726e-05,
      "loss": 0.2215,
      "step": 1089
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7906859664630741,
      "learning_rate": 3.9163413705572984e-05,
      "loss": 0.3339,
      "step": 1090
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6765223061689623,
      "learning_rate": 3.916137645257587e-05,
      "loss": 0.3126,
      "step": 1091
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5569111752132542,
      "learning_rate": 3.9159336775156165e-05,
      "loss": 0.2029,
      "step": 1092
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6471343446616511,
      "learning_rate": 3.915729467357194e-05,
      "loss": 0.23,
      "step": 1093
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5604740055583318,
      "learning_rate": 3.9155250148081564e-05,
      "loss": 0.2222,
      "step": 1094
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6115950224215414,
      "learning_rate": 3.915320319894373e-05,
      "loss": 0.2503,
      "step": 1095
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6607794895180239,
      "learning_rate": 3.9151153826417436e-05,
      "loss": 0.246,
      "step": 1096
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.814486788778958,
      "learning_rate": 3.914910203076198e-05,
      "loss": 0.282,
      "step": 1097
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0233962654950661,
      "learning_rate": 3.914704781223696e-05,
      "loss": 0.405,
      "step": 1098
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7525338048724047,
      "learning_rate": 3.91449911711023e-05,
      "loss": 0.3131,
      "step": 1099
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6456261322956921,
      "learning_rate": 3.9142932107618214e-05,
      "loss": 0.2325,
      "step": 1100
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5747696446201337,
      "learning_rate": 3.914087062204523e-05,
      "loss": 0.1888,
      "step": 1101
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6842055280740315,
      "learning_rate": 3.913880671464418e-05,
      "loss": 0.2599,
      "step": 1102
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6431547320048512,
      "learning_rate": 3.91367403856762e-05,
      "loss": 0.2516,
      "step": 1103
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7544973161420873,
      "learning_rate": 3.9134671635402745e-05,
      "loss": 0.2657,
      "step": 1104
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7207657288030248,
      "learning_rate": 3.9132600464085556e-05,
      "loss": 0.3575,
      "step": 1105
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7417425729362398,
      "learning_rate": 3.91305268719867e-05,
      "loss": 0.3183,
      "step": 1106
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8137674349720427,
      "learning_rate": 3.912845085936853e-05,
      "loss": 0.3183,
      "step": 1107
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.749358021636869,
      "learning_rate": 3.912637242649373e-05,
      "loss": 0.2676,
      "step": 1108
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.967900857707362,
      "learning_rate": 3.9124291573625263e-05,
      "loss": 0.3343,
      "step": 1109
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7044682916703117,
      "learning_rate": 3.912220830102643e-05,
      "loss": 0.292,
      "step": 1110
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7130506780010349,
      "learning_rate": 3.91201226089608e-05,
      "loss": 0.292,
      "step": 1111
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.620463309224544,
      "learning_rate": 3.911803449769228e-05,
      "loss": 0.2499,
      "step": 1112
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6598372281091182,
      "learning_rate": 3.9115943967485085e-05,
      "loss": 0.2533,
      "step": 1113
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5889333052074396,
      "learning_rate": 3.911385101860369e-05,
      "loss": 0.1957,
      "step": 1114
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6469707215478148,
      "learning_rate": 3.911175565131293e-05,
      "loss": 0.2776,
      "step": 1115
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6861570854030776,
      "learning_rate": 3.9109657865877924e-05,
      "loss": 0.2886,
      "step": 1116
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6399747292923705,
      "learning_rate": 3.9107557662564086e-05,
      "loss": 0.2394,
      "step": 1117
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7348381367687417,
      "learning_rate": 3.910545504163716e-05,
      "loss": 0.2749,
      "step": 1118
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.76546893183466,
      "learning_rate": 3.910335000336319e-05,
      "loss": 0.2976,
      "step": 1119
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5289544268124914,
      "learning_rate": 3.9101242548008496e-05,
      "loss": 0.2018,
      "step": 1120
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7797200672021879,
      "learning_rate": 3.9099132675839736e-05,
      "loss": 0.286,
      "step": 1121
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7877839938752322,
      "learning_rate": 3.9097020387123876e-05,
      "loss": 0.3017,
      "step": 1122
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6544094159713822,
      "learning_rate": 3.9094905682128166e-05,
      "loss": 0.2666,
      "step": 1123
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5002055937556827,
      "learning_rate": 3.9092788561120174e-05,
      "loss": 0.1852,
      "step": 1124
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5951012494388362,
      "learning_rate": 3.909066902436777e-05,
      "loss": 0.2243,
      "step": 1125
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6183167520606488,
      "learning_rate": 3.9088547072139145e-05,
      "loss": 0.2042,
      "step": 1126
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8181522808378632,
      "learning_rate": 3.9086422704702754e-05,
      "loss": 0.2826,
      "step": 1127
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9041795098290607,
      "learning_rate": 3.9084295922327414e-05,
      "loss": 0.3177,
      "step": 1128
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6192185987597003,
      "learning_rate": 3.908216672528221e-05,
      "loss": 0.1909,
      "step": 1129
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7426067172209254,
      "learning_rate": 3.908003511383654e-05,
      "loss": 0.2949,
      "step": 1130
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5395648661022168,
      "learning_rate": 3.907790108826011e-05,
      "loss": 0.2393,
      "step": 1131
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6930545190418199,
      "learning_rate": 3.907576464882294e-05,
      "loss": 0.2748,
      "step": 1132
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7622005406163825,
      "learning_rate": 3.907362579579532e-05,
      "loss": 0.3442,
      "step": 1133
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6376383640629019,
      "learning_rate": 3.90714845294479e-05,
      "loss": 0.2322,
      "step": 1134
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7228179750676318,
      "learning_rate": 3.9069340850051595e-05,
      "loss": 0.2668,
      "step": 1135
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5861811576310614,
      "learning_rate": 3.9067194757877635e-05,
      "loss": 0.2324,
      "step": 1136
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5741847540724454,
      "learning_rate": 3.906504625319756e-05,
      "loss": 0.1767,
      "step": 1137
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6147131755762856,
      "learning_rate": 3.906289533628322e-05,
      "loss": 0.2624,
      "step": 1138
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5390624170717921,
      "learning_rate": 3.906074200740675e-05,
      "loss": 0.2086,
      "step": 1139
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7785584816474902,
      "learning_rate": 3.9058586266840614e-05,
      "loss": 0.2773,
      "step": 1140
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5692274929299811,
      "learning_rate": 3.905642811485757e-05,
      "loss": 0.2317,
      "step": 1141
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6684624258622031,
      "learning_rate": 3.905426755173068e-05,
      "loss": 0.2385,
      "step": 1142
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5692816785341807,
      "learning_rate": 3.90521045777333e-05,
      "loss": 0.2181,
      "step": 1143
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7373620421478237,
      "learning_rate": 3.904993919313912e-05,
      "loss": 0.2796,
      "step": 1144
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6851066197305056,
      "learning_rate": 3.9047771398222115e-05,
      "loss": 0.2429,
      "step": 1145
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46115732608511156,
      "learning_rate": 3.9045601193256564e-05,
      "loss": 0.1981,
      "step": 1146
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.572500187319929,
      "learning_rate": 3.9043428578517055e-05,
      "loss": 0.2363,
      "step": 1147
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6352037062517615,
      "learning_rate": 3.9041253554278486e-05,
      "loss": 0.2702,
      "step": 1148
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7629694915961622,
      "learning_rate": 3.9039076120816055e-05,
      "loss": 0.3771,
      "step": 1149
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6471239136275394,
      "learning_rate": 3.9036896278405264e-05,
      "loss": 0.3015,
      "step": 1150
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8248461464270617,
      "learning_rate": 3.903471402732193e-05,
      "loss": 0.3045,
      "step": 1151
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.763132163161792,
      "learning_rate": 3.9032529367842145e-05,
      "loss": 0.2647,
      "step": 1152
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.663105128688938,
      "learning_rate": 3.9030342300242334e-05,
      "loss": 0.2612,
      "step": 1153
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7474917351383821,
      "learning_rate": 3.902815282479923e-05,
      "loss": 0.2497,
      "step": 1154
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.549185504151279,
      "learning_rate": 3.902596094178985e-05,
      "loss": 0.2355,
      "step": 1155
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5437948679773392,
      "learning_rate": 3.902376665149153e-05,
      "loss": 0.2059,
      "step": 1156
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7314118198116689,
      "learning_rate": 3.9021569954181905e-05,
      "loss": 0.3453,
      "step": 1157
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6924006751679204,
      "learning_rate": 3.9019370850138915e-05,
      "loss": 0.2964,
      "step": 1158
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8438484699575118,
      "learning_rate": 3.90171693396408e-05,
      "loss": 0.3169,
      "step": 1159
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6313163826285553,
      "learning_rate": 3.9014965422966115e-05,
      "loss": 0.253,
      "step": 1160
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5590317616080528,
      "learning_rate": 3.9012759100393714e-05,
      "loss": 0.1806,
      "step": 1161
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6003285670674611,
      "learning_rate": 3.9010550372202756e-05,
      "loss": 0.2258,
      "step": 1162
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.68521455714584,
      "learning_rate": 3.90083392386727e-05,
      "loss": 0.2908,
      "step": 1163
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6515214957455261,
      "learning_rate": 3.900612570008331e-05,
      "loss": 0.204,
      "step": 1164
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7735844867408266,
      "learning_rate": 3.900390975671467e-05,
      "loss": 0.2509,
      "step": 1165
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7204671333806125,
      "learning_rate": 3.900169140884715e-05,
      "loss": 0.2897,
      "step": 1166
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7108858529918054,
      "learning_rate": 3.8999470656761425e-05,
      "loss": 0.2336,
      "step": 1167
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6739129939052515,
      "learning_rate": 3.899724750073848e-05,
      "loss": 0.2398,
      "step": 1168
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5950638639812865,
      "learning_rate": 3.89950219410596e-05,
      "loss": 0.2394,
      "step": 1169
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6235646216257108,
      "learning_rate": 3.899279397800639e-05,
      "loss": 0.2312,
      "step": 1170
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7044447487092723,
      "learning_rate": 3.899056361186074e-05,
      "loss": 0.2502,
      "step": 1171
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6785046160940794,
      "learning_rate": 3.8988330842904844e-05,
      "loss": 0.2393,
      "step": 1172
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6668498835530737,
      "learning_rate": 3.8986095671421214e-05,
      "loss": 0.2567,
      "step": 1173
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7345434360917417,
      "learning_rate": 3.8983858097692656e-05,
      "loss": 0.277,
      "step": 1174
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6933838745946668,
      "learning_rate": 3.8981618122002284e-05,
      "loss": 0.3151,
      "step": 1175
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6841713780958396,
      "learning_rate": 3.8979375744633515e-05,
      "loss": 0.3232,
      "step": 1176
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5284133800955243,
      "learning_rate": 3.897713096587007e-05,
      "loss": 0.2942,
      "step": 1177
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7390750069396702,
      "learning_rate": 3.897488378599596e-05,
      "loss": 0.2582,
      "step": 1178
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6034182374271129,
      "learning_rate": 3.897263420529553e-05,
      "loss": 0.2431,
      "step": 1179
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5839404341583044,
      "learning_rate": 3.8970382224053414e-05,
      "loss": 0.2155,
      "step": 1180
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6694830299559975,
      "learning_rate": 3.8968127842554526e-05,
      "loss": 0.2396,
      "step": 1181
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6697326481882796,
      "learning_rate": 3.8965871061084126e-05,
      "loss": 0.2595,
      "step": 1182
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8205411683121762,
      "learning_rate": 3.896361187992774e-05,
      "loss": 0.3036,
      "step": 1183
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6065592144381388,
      "learning_rate": 3.896135029937123e-05,
      "loss": 0.2003,
      "step": 1184
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0616656000280686,
      "learning_rate": 3.895908631970074e-05,
      "loss": 0.3972,
      "step": 1185
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.543240979786405,
      "learning_rate": 3.895681994120272e-05,
      "loss": 0.2014,
      "step": 1186
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5940428312879357,
      "learning_rate": 3.8954551164163924e-05,
      "loss": 0.2251,
      "step": 1187
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7983022138493923,
      "learning_rate": 3.8952279988871425e-05,
      "loss": 0.3055,
      "step": 1188
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5650046271160005,
      "learning_rate": 3.895000641561258e-05,
      "loss": 0.2271,
      "step": 1189
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6216246297402951,
      "learning_rate": 3.894773044467505e-05,
      "loss": 0.2454,
      "step": 1190
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7433994637212084,
      "learning_rate": 3.894545207634681e-05,
      "loss": 0.3221,
      "step": 1191
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6981473418155327,
      "learning_rate": 3.8943171310916146e-05,
      "loss": 0.2652,
      "step": 1192
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6493088256925852,
      "learning_rate": 3.8940888148671615e-05,
      "loss": 0.2261,
      "step": 1193
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6113396299211942,
      "learning_rate": 3.893860258990212e-05,
      "loss": 0.1989,
      "step": 1194
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7437777698365386,
      "learning_rate": 3.893631463489682e-05,
      "loss": 0.2734,
      "step": 1195
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6566140663635328,
      "learning_rate": 3.893402428394522e-05,
      "loss": 0.2117,
      "step": 1196
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7000718267238578,
      "learning_rate": 3.8931731537337105e-05,
      "loss": 0.2229,
      "step": 1197
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.658687877150711,
      "learning_rate": 3.892943639536257e-05,
      "loss": 0.2393,
      "step": 1198
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7123417661873115,
      "learning_rate": 3.892713885831201e-05,
      "loss": 0.283,
      "step": 1199
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7320268303117896,
      "learning_rate": 3.8924838926476114e-05,
      "loss": 0.2446,
      "step": 1200
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6106401417396183,
      "learning_rate": 3.8922536600145914e-05,
      "loss": 0.2123,
      "step": 1201
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6672925817209624,
      "learning_rate": 3.892023187961268e-05,
      "loss": 0.2638,
      "step": 1202
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6572337043906973,
      "learning_rate": 3.891792476516804e-05,
      "loss": 0.272,
      "step": 1203
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6910810406193486,
      "learning_rate": 3.891561525710389e-05,
      "loss": 0.2475,
      "step": 1204
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6645591673647715,
      "learning_rate": 3.8913303355712476e-05,
      "loss": 0.2441,
      "step": 1205
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4280210257361357,
      "learning_rate": 3.891098906128628e-05,
      "loss": 0.1758,
      "step": 1206
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6574830777561546,
      "learning_rate": 3.890867237411814e-05,
      "loss": 0.2962,
      "step": 1207
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6820573161629137,
      "learning_rate": 3.890635329450118e-05,
      "loss": 0.239,
      "step": 1208
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6226431994823505,
      "learning_rate": 3.8904031822728804e-05,
      "loss": 0.2933,
      "step": 1209
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6802266930930283,
      "learning_rate": 3.890170795909477e-05,
      "loss": 0.2117,
      "step": 1210
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5477393812354234,
      "learning_rate": 3.889938170389309e-05,
      "loss": 0.2366,
      "step": 1211
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5213901277449603,
      "learning_rate": 3.88970530574181e-05,
      "loss": 0.1795,
      "step": 1212
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8229024056917156,
      "learning_rate": 3.889472201996444e-05,
      "loss": 0.2598,
      "step": 1213
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8605302848030582,
      "learning_rate": 3.889238859182703e-05,
      "loss": 0.28,
      "step": 1214
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7069294050008,
      "learning_rate": 3.889005277330114e-05,
      "loss": 0.2642,
      "step": 1215
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8320750909994648,
      "learning_rate": 3.888771456468229e-05,
      "loss": 0.2667,
      "step": 1216
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7472166229788278,
      "learning_rate": 3.888537396626634e-05,
      "loss": 0.2413,
      "step": 1217
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5918464513981199,
      "learning_rate": 3.8883030978349416e-05,
      "loss": 0.1717,
      "step": 1218
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6290008047353451,
      "learning_rate": 3.8880685601228e-05,
      "loss": 0.2442,
      "step": 1219
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6693512735872327,
      "learning_rate": 3.887833783519882e-05,
      "loss": 0.2768,
      "step": 1220
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6337689365665258,
      "learning_rate": 3.887598768055894e-05,
      "loss": 0.2543,
      "step": 1221
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5820862916868048,
      "learning_rate": 3.887363513760571e-05,
      "loss": 0.2686,
      "step": 1222
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.654137435389576,
      "learning_rate": 3.88712802066368e-05,
      "loss": 0.2651,
      "step": 1223
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6068287495708279,
      "learning_rate": 3.8868922887950165e-05,
      "loss": 0.2478,
      "step": 1224
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6847562224574402,
      "learning_rate": 3.886656318184407e-05,
      "loss": 0.2938,
      "step": 1225
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6221632476325742,
      "learning_rate": 3.886420108861708e-05,
      "loss": 0.2376,
      "step": 1226
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5977840411625438,
      "learning_rate": 3.886183660856806e-05,
      "loss": 0.2174,
      "step": 1227
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5322449847191587,
      "learning_rate": 3.885946974199618e-05,
      "loss": 0.1947,
      "step": 1228
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6127247563426347,
      "learning_rate": 3.8857100489200915e-05,
      "loss": 0.213,
      "step": 1229
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5792430170994114,
      "learning_rate": 3.8854728850482034e-05,
      "loss": 0.1908,
      "step": 1230
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6969818905676577,
      "learning_rate": 3.885235482613962e-05,
      "loss": 0.2522,
      "step": 1231
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7777228652968459,
      "learning_rate": 3.884997841647404e-05,
      "loss": 0.2884,
      "step": 1232
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8085412639592999,
      "learning_rate": 3.8847599621785976e-05,
      "loss": 0.2158,
      "step": 1233
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8555021932675256,
      "learning_rate": 3.8845218442376416e-05,
      "loss": 0.3035,
      "step": 1234
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.60544748576571,
      "learning_rate": 3.8842834878546625e-05,
      "loss": 0.2205,
      "step": 1235
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6872595019608175,
      "learning_rate": 3.8840448930598216e-05,
      "loss": 0.2765,
      "step": 1236
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6201364108733016,
      "learning_rate": 3.883806059883304e-05,
      "loss": 0.2321,
      "step": 1237
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5343775408946525,
      "learning_rate": 3.8835669883553315e-05,
      "loss": 0.1676,
      "step": 1238
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.661309268004355,
      "learning_rate": 3.8833276785061504e-05,
      "loss": 0.2325,
      "step": 1239
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6090929405626248,
      "learning_rate": 3.883088130366042e-05,
      "loss": 0.233,
      "step": 1240
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6542144267957732,
      "learning_rate": 3.882848343965314e-05,
      "loss": 0.2596,
      "step": 1241
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.554565819302052,
      "learning_rate": 3.882608319334306e-05,
      "loss": 0.2281,
      "step": 1242
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6197919338022719,
      "learning_rate": 3.8823680565033884e-05,
      "loss": 0.2126,
      "step": 1243
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7289902927893132,
      "learning_rate": 3.88212755550296e-05,
      "loss": 0.3425,
      "step": 1244
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6156219472664214,
      "learning_rate": 3.881886816363451e-05,
      "loss": 0.2354,
      "step": 1245
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6649824614112222,
      "learning_rate": 3.88164583911532e-05,
      "loss": 0.2607,
      "step": 1246
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6089744718772897,
      "learning_rate": 3.881404623789059e-05,
      "loss": 0.2025,
      "step": 1247
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5898090377597667,
      "learning_rate": 3.881163170415186e-05,
      "loss": 0.2179,
      "step": 1248
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.622741911582533,
      "learning_rate": 3.880921479024253e-05,
      "loss": 0.249,
      "step": 1249
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6799072261241708,
      "learning_rate": 3.88067954964684e-05,
      "loss": 0.2457,
      "step": 1250
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.964697946504875,
      "learning_rate": 3.8804373823135576e-05,
      "loss": 0.3046,
      "step": 1251
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.746965069307749,
      "learning_rate": 3.880194977055045e-05,
      "loss": 0.2465,
      "step": 1252
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.668794979828834,
      "learning_rate": 3.879952333901975e-05,
      "loss": 0.3104,
      "step": 1253
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44904798496177784,
      "learning_rate": 3.8797094528850474e-05,
      "loss": 0.2025,
      "step": 1254
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5123551175554063,
      "learning_rate": 3.8794663340349915e-05,
      "loss": 0.2133,
      "step": 1255
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6448259315774166,
      "learning_rate": 3.8792229773825716e-05,
      "loss": 0.2481,
      "step": 1256
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4628615229347241,
      "learning_rate": 3.878979382958576e-05,
      "loss": 0.155,
      "step": 1257
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5933499494335127,
      "learning_rate": 3.878735550793827e-05,
      "loss": 0.2285,
      "step": 1258
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5237221584340037,
      "learning_rate": 3.8784914809191764e-05,
      "loss": 0.1666,
      "step": 1259
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6483043396047382,
      "learning_rate": 3.8782471733655044e-05,
      "loss": 0.249,
      "step": 1260
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.606080736540311,
      "learning_rate": 3.8780026281637224e-05,
      "loss": 0.2158,
      "step": 1261
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6714757797818425,
      "learning_rate": 3.877757845344773e-05,
      "loss": 0.2476,
      "step": 1262
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7585959170278128,
      "learning_rate": 3.8775128249396265e-05,
      "loss": 0.2354,
      "step": 1263
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6891021699796219,
      "learning_rate": 3.8772675669792855e-05,
      "loss": 0.2292,
      "step": 1264
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7685840024289295,
      "learning_rate": 3.877022071494781e-05,
      "loss": 0.2447,
      "step": 1265
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5877327437697002,
      "learning_rate": 3.876776338517174e-05,
      "loss": 0.1985,
      "step": 1266
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.847615025902723,
      "learning_rate": 3.876530368077558e-05,
      "loss": 0.2218,
      "step": 1267
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8589034433997359,
      "learning_rate": 3.876284160207053e-05,
      "loss": 0.3489,
      "step": 1268
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7512883802692587,
      "learning_rate": 3.876037714936813e-05,
      "loss": 0.2595,
      "step": 1269
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7304878436362356,
      "learning_rate": 3.875791032298017e-05,
      "loss": 0.3265,
      "step": 1270
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6648290361329455,
      "learning_rate": 3.875544112321879e-05,
      "loss": 0.281,
      "step": 1271
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7660220537397362,
      "learning_rate": 3.875296955039641e-05,
      "loss": 0.331,
      "step": 1272
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5581666073556578,
      "learning_rate": 3.8750495604825733e-05,
      "loss": 0.1946,
      "step": 1273
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5847558631668265,
      "learning_rate": 3.874801928681979e-05,
      "loss": 0.1857,
      "step": 1274
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5725991640872297,
      "learning_rate": 3.8745540596691905e-05,
      "loss": 0.2284,
      "step": 1275
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6228902015890331,
      "learning_rate": 3.87430595347557e-05,
      "loss": 0.2098,
      "step": 1276
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6430056694620289,
      "learning_rate": 3.874057610132508e-05,
      "loss": 0.2459,
      "step": 1277
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6038958391918268,
      "learning_rate": 3.873809029671427e-05,
      "loss": 0.2406,
      "step": 1278
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6131309519487615,
      "learning_rate": 3.87356021212378e-05,
      "loss": 0.2314,
      "step": 1279
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6606580418504284,
      "learning_rate": 3.873311157521048e-05,
      "loss": 0.2363,
      "step": 1280
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8186205171889596,
      "learning_rate": 3.873061865894744e-05,
      "loss": 0.3057,
      "step": 1281
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7306612138608514,
      "learning_rate": 3.8728123372764085e-05,
      "loss": 0.2202,
      "step": 1282
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5884113476938773,
      "learning_rate": 3.872562571697615e-05,
      "loss": 0.2006,
      "step": 1283
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.49841716689009385,
      "learning_rate": 3.8723125691899646e-05,
      "loss": 0.1889,
      "step": 1284
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5397632853581106,
      "learning_rate": 3.8720623297850895e-05,
      "loss": 0.1834,
      "step": 1285
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7106599370710558,
      "learning_rate": 3.871811853514652e-05,
      "loss": 0.2299,
      "step": 1286
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7194856941969401,
      "learning_rate": 3.871561140410343e-05,
      "loss": 0.2389,
      "step": 1287
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6744908708614878,
      "learning_rate": 3.871310190503886e-05,
      "loss": 0.2422,
      "step": 1288
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6636175123271209,
      "learning_rate": 3.8710590038270305e-05,
      "loss": 0.2576,
      "step": 1289
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7140115782362624,
      "learning_rate": 3.870807580411561e-05,
      "loss": 0.2728,
      "step": 1290
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7325567095940773,
      "learning_rate": 3.870555920289287e-05,
      "loss": 0.2456,
      "step": 1291
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7164992876282243,
      "learning_rate": 3.870304023492051e-05,
      "loss": 0.3254,
      "step": 1292
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4950636133105119,
      "learning_rate": 3.870051890051725e-05,
      "loss": 0.2218,
      "step": 1293
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5366939533396958,
      "learning_rate": 3.8697995200002105e-05,
      "loss": 0.2285,
      "step": 1294
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6601583040645628,
      "learning_rate": 3.8695469133694384e-05,
      "loss": 0.2085,
      "step": 1295
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6514617759736432,
      "learning_rate": 3.8692940701913706e-05,
      "loss": 0.271,
      "step": 1296
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6292324286528318,
      "learning_rate": 3.869040990497998e-05,
      "loss": 0.19,
      "step": 1297
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.645190370100445,
      "learning_rate": 3.868787674321343e-05,
      "loss": 0.2397,
      "step": 1298
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6137624662807509,
      "learning_rate": 3.868534121693455e-05,
      "loss": 0.2095,
      "step": 1299
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.603864895883944,
      "learning_rate": 3.868280332646417e-05,
      "loss": 0.2651,
      "step": 1300
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5939647135753702,
      "learning_rate": 3.868026307212339e-05,
      "loss": 0.2023,
      "step": 1301
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.587893514531027,
      "learning_rate": 3.867772045423362e-05,
      "loss": 0.2385,
      "step": 1302
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.628995782390761,
      "learning_rate": 3.867517547311657e-05,
      "loss": 0.2385,
      "step": 1303
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.670993803706319,
      "learning_rate": 3.8672628129094255e-05,
      "loss": 0.2629,
      "step": 1304
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5956606990527377,
      "learning_rate": 3.867007842248897e-05,
      "loss": 0.2059,
      "step": 1305
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6449072463596339,
      "learning_rate": 3.8667526353623326e-05,
      "loss": 0.2044,
      "step": 1306
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7756324602332358,
      "learning_rate": 3.866497192282023e-05,
      "loss": 0.2972,
      "step": 1307
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5729374766182814,
      "learning_rate": 3.866241513040288e-05,
      "loss": 0.1879,
      "step": 1308
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6493318204503121,
      "learning_rate": 3.865985597669478e-05,
      "loss": 0.1981,
      "step": 1309
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9166138084661662,
      "learning_rate": 3.8657294462019735e-05,
      "loss": 0.3181,
      "step": 1310
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6929765315946769,
      "learning_rate": 3.865473058670183e-05,
      "loss": 0.2,
      "step": 1311
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.80787036622088,
      "learning_rate": 3.865216435106549e-05,
      "loss": 0.2661,
      "step": 1312
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5621348891521877,
      "learning_rate": 3.8649595755435383e-05,
      "loss": 0.1811,
      "step": 1313
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5045417799667435,
      "learning_rate": 3.8647024800136524e-05,
      "loss": 0.1838,
      "step": 1314
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.683540603751585,
      "learning_rate": 3.86444514854942e-05,
      "loss": 0.1961,
      "step": 1315
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6355911755259537,
      "learning_rate": 3.8641875811834004e-05,
      "loss": 0.2462,
      "step": 1316
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5304297678673695,
      "learning_rate": 3.8639297779481825e-05,
      "loss": 0.194,
      "step": 1317
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5986492658784093,
      "learning_rate": 3.863671738876385e-05,
      "loss": 0.2387,
      "step": 1318
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.528911238946868,
      "learning_rate": 3.8634134640006585e-05,
      "loss": 0.1738,
      "step": 1319
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7341278553106145,
      "learning_rate": 3.863154953353679e-05,
      "loss": 0.2954,
      "step": 1320
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6440245052348925,
      "learning_rate": 3.862896206968157e-05,
      "loss": 0.2334,
      "step": 1321
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6774699989939744,
      "learning_rate": 3.8626372248768295e-05,
      "loss": 0.2287,
      "step": 1322
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9057785649545252,
      "learning_rate": 3.862378007112465e-05,
      "loss": 0.3634,
      "step": 1323
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7203096385576256,
      "learning_rate": 3.862118553707863e-05,
      "loss": 0.2639,
      "step": 1324
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7933059877269436,
      "learning_rate": 3.8618588646958483e-05,
      "loss": 0.3027,
      "step": 1325
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6650534696880952,
      "learning_rate": 3.86159894010928e-05,
      "loss": 0.2746,
      "step": 1326
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43670280236348946,
      "learning_rate": 3.8613387799810464e-05,
      "loss": 0.1619,
      "step": 1327
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5615618617967515,
      "learning_rate": 3.8610783843440626e-05,
      "loss": 0.163,
      "step": 1328
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.629366025434526,
      "learning_rate": 3.860817753231278e-05,
      "loss": 0.214,
      "step": 1329
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44901373787616894,
      "learning_rate": 3.8605568866756666e-05,
      "loss": 0.1559,
      "step": 1330
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5633414121289285,
      "learning_rate": 3.860295784710237e-05,
      "loss": 0.2219,
      "step": 1331
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5336170755540967,
      "learning_rate": 3.860034447368024e-05,
      "loss": 0.2078,
      "step": 1332
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7093887226945248,
      "learning_rate": 3.8597728746820955e-05,
      "loss": 0.2702,
      "step": 1333
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8586704834239652,
      "learning_rate": 3.8595110666855466e-05,
      "loss": 0.3336,
      "step": 1334
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7121096579111179,
      "learning_rate": 3.859249023411503e-05,
      "loss": 0.2298,
      "step": 1335
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6523963313126454,
      "learning_rate": 3.858986744893119e-05,
      "loss": 0.2655,
      "step": 1336
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7057387408818833,
      "learning_rate": 3.8587242311635805e-05,
      "loss": 0.2454,
      "step": 1337
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5975010739001595,
      "learning_rate": 3.858461482256103e-05,
      "loss": 0.2242,
      "step": 1338
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7903046603585147,
      "learning_rate": 3.858198498203931e-05,
      "loss": 0.2764,
      "step": 1339
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6611386725702717,
      "learning_rate": 3.8579352790403395e-05,
      "loss": 0.1945,
      "step": 1340
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.644720015193557,
      "learning_rate": 3.857671824798631e-05,
      "loss": 0.2159,
      "step": 1341
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6836654189551152,
      "learning_rate": 3.857408135512142e-05,
      "loss": 0.2078,
      "step": 1342
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5925070149774991,
      "learning_rate": 3.857144211214233e-05,
      "loss": 0.2131,
      "step": 1343
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6736078759298164,
      "learning_rate": 3.8568800519383e-05,
      "loss": 0.2433,
      "step": 1344
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6608968113488545,
      "learning_rate": 3.856615657717765e-05,
      "loss": 0.205,
      "step": 1345
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5574863367159772,
      "learning_rate": 3.856351028586082e-05,
      "loss": 0.162,
      "step": 1346
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5952551986871059,
      "learning_rate": 3.856086164576731e-05,
      "loss": 0.2107,
      "step": 1347
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6002483132246118,
      "learning_rate": 3.855821065723228e-05,
      "loss": 0.2072,
      "step": 1348
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7489851602375327,
      "learning_rate": 3.8555557320591125e-05,
      "loss": 0.3085,
      "step": 1349
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5825581821559521,
      "learning_rate": 3.855290163617956e-05,
      "loss": 0.1934,
      "step": 1350
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5336214318083551,
      "learning_rate": 3.855024360433362e-05,
      "loss": 0.1938,
      "step": 1351
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5675469323636004,
      "learning_rate": 3.8547583225389596e-05,
      "loss": 0.2049,
      "step": 1352
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8589140956521026,
      "learning_rate": 3.8544920499684115e-05,
      "loss": 0.3192,
      "step": 1353
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5594412769713898,
      "learning_rate": 3.8542255427554065e-05,
      "loss": 0.2138,
      "step": 1354
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6392457116482576,
      "learning_rate": 3.853958800933666e-05,
      "loss": 0.3116,
      "step": 1355
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5834558449115287,
      "learning_rate": 3.85369182453694e-05,
      "loss": 0.23,
      "step": 1356
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.564808136594211,
      "learning_rate": 3.853424613599007e-05,
      "loss": 0.1764,
      "step": 1357
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.47273115084798006,
      "learning_rate": 3.853157168153677e-05,
      "loss": 0.1842,
      "step": 1358
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.679784000605634,
      "learning_rate": 3.852889488234789e-05,
      "loss": 0.2744,
      "step": 1359
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45843104021842784,
      "learning_rate": 3.852621573876212e-05,
      "loss": 0.163,
      "step": 1360
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5392846465212989,
      "learning_rate": 3.8523534251118426e-05,
      "loss": 0.2324,
      "step": 1361
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.645274617928227,
      "learning_rate": 3.8520850419756104e-05,
      "loss": 0.2286,
      "step": 1362
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8955659837039383,
      "learning_rate": 3.851816424501473e-05,
      "loss": 0.3426,
      "step": 1363
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5718774097162559,
      "learning_rate": 3.851547572723416e-05,
      "loss": 0.1985,
      "step": 1364
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6350843363047664,
      "learning_rate": 3.851278486675459e-05,
      "loss": 0.1971,
      "step": 1365
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6870865662450322,
      "learning_rate": 3.851009166391646e-05,
      "loss": 0.2976,
      "step": 1366
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6039437315685586,
      "learning_rate": 3.850739611906054e-05,
      "loss": 0.2193,
      "step": 1367
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6965816359289543,
      "learning_rate": 3.850469823252789e-05,
      "loss": 0.2746,
      "step": 1368
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5924091005010735,
      "learning_rate": 3.850199800465987e-05,
      "loss": 0.1989,
      "step": 1369
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5739712603498505,
      "learning_rate": 3.849929543579812e-05,
      "loss": 0.1996,
      "step": 1370
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6416951289820902,
      "learning_rate": 3.849659052628459e-05,
      "loss": 0.1992,
      "step": 1371
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6425116600284796,
      "learning_rate": 3.849388327646152e-05,
      "loss": 0.2582,
      "step": 1372
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5995535560496725,
      "learning_rate": 3.849117368667145e-05,
      "loss": 0.1936,
      "step": 1373
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7262496274595898,
      "learning_rate": 3.848846175725722e-05,
      "loss": 0.2981,
      "step": 1374
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.520994635082087,
      "learning_rate": 3.848574748856196e-05,
      "loss": 0.2027,
      "step": 1375
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6204295895855063,
      "learning_rate": 3.84830308809291e-05,
      "loss": 0.2178,
      "step": 1376
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5023853207676777,
      "learning_rate": 3.848031193470234e-05,
      "loss": 0.1797,
      "step": 1377
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5656745657094889,
      "learning_rate": 3.8477590650225735e-05,
      "loss": 0.1695,
      "step": 1378
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5448724047447405,
      "learning_rate": 3.847486702784358e-05,
      "loss": 0.1843,
      "step": 1379
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6005323432642046,
      "learning_rate": 3.8472141067900485e-05,
      "loss": 0.2546,
      "step": 1380
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5501483229564694,
      "learning_rate": 3.8469412770741355e-05,
      "loss": 0.2079,
      "step": 1381
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5533484293226613,
      "learning_rate": 3.84666821367114e-05,
      "loss": 0.2008,
      "step": 1382
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6177398286566418,
      "learning_rate": 3.846394916615611e-05,
      "loss": 0.1941,
      "step": 1383
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6260502573990595,
      "learning_rate": 3.846121385942128e-05,
      "loss": 0.243,
      "step": 1384
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7802480757963685,
      "learning_rate": 3.8458476216853e-05,
      "loss": 0.2483,
      "step": 1385
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.497857360686356,
      "learning_rate": 3.845573623879766e-05,
      "loss": 0.15,
      "step": 1386
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.647615969409576,
      "learning_rate": 3.8452993925601935e-05,
      "loss": 0.2322,
      "step": 1387
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7513537904255497,
      "learning_rate": 3.845024927761279e-05,
      "loss": 0.2723,
      "step": 1388
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5922225579425241,
      "learning_rate": 3.844750229517751e-05,
      "loss": 0.1969,
      "step": 1389
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6564419329487305,
      "learning_rate": 3.844475297864366e-05,
      "loss": 0.1896,
      "step": 1390
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5620477765909019,
      "learning_rate": 3.84420013283591e-05,
      "loss": 0.1986,
      "step": 1391
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7264961704617914,
      "learning_rate": 3.843924734467199e-05,
      "loss": 0.2349,
      "step": 1392
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6200766482581344,
      "learning_rate": 3.843649102793077e-05,
      "loss": 0.2405,
      "step": 1393
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5409938931834414,
      "learning_rate": 3.843373237848419e-05,
      "loss": 0.1651,
      "step": 1394
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7545101928932502,
      "learning_rate": 3.843097139668131e-05,
      "loss": 0.2639,
      "step": 1395
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5616662946675293,
      "learning_rate": 3.842820808287144e-05,
      "loss": 0.1942,
      "step": 1396
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6924023968465035,
      "learning_rate": 3.842544243740424e-05,
      "loss": 0.2813,
      "step": 1397
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5803282227489259,
      "learning_rate": 3.842267446062962e-05,
      "loss": 0.1823,
      "step": 1398
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5079350837057682,
      "learning_rate": 3.84199041528978e-05,
      "loss": 0.1484,
      "step": 1399
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6176581941897777,
      "learning_rate": 3.841713151455931e-05,
      "loss": 0.2529,
      "step": 1400
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5198121422943797,
      "learning_rate": 3.8414356545964955e-05,
      "loss": 0.2114,
      "step": 1401
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5765381763982369,
      "learning_rate": 3.8411579247465845e-05,
      "loss": 0.2078,
      "step": 1402
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5100605429448324,
      "learning_rate": 3.840879961941338e-05,
      "loss": 0.195,
      "step": 1403
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5536489569742455,
      "learning_rate": 3.840601766215926e-05,
      "loss": 0.1969,
      "step": 1404
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5270076952504538,
      "learning_rate": 3.840323337605548e-05,
      "loss": 0.1979,
      "step": 1405
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6129683407002526,
      "learning_rate": 3.840044676145431e-05,
      "loss": 0.2249,
      "step": 1406
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5335944001178495,
      "learning_rate": 3.839765781870834e-05,
      "loss": 0.1946,
      "step": 1407
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5991831703017667,
      "learning_rate": 3.839486654817045e-05,
      "loss": 0.2184,
      "step": 1408
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6384652179633203,
      "learning_rate": 3.8392072950193806e-05,
      "loss": 0.2342,
      "step": 1409
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6515416679436807,
      "learning_rate": 3.838927702513187e-05,
      "loss": 0.1825,
      "step": 1410
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6086950176247553,
      "learning_rate": 3.8386478773338407e-05,
      "loss": 0.1963,
      "step": 1411
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5317600830764958,
      "learning_rate": 3.838367819516746e-05,
      "loss": 0.1645,
      "step": 1412
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5675356424372892,
      "learning_rate": 3.8380875290973384e-05,
      "loss": 0.1917,
      "step": 1413
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.755198976305292,
      "learning_rate": 3.837807006111082e-05,
      "loss": 0.269,
      "step": 1414
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.597506834813055,
      "learning_rate": 3.837526250593471e-05,
      "loss": 0.1558,
      "step": 1415
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6141618856902178,
      "learning_rate": 3.837245262580027e-05,
      "loss": 0.2322,
      "step": 1416
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7328624570040942,
      "learning_rate": 3.8369640421063035e-05,
      "loss": 0.1851,
      "step": 1417
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5636229961137138,
      "learning_rate": 3.836682589207882e-05,
      "loss": 0.1771,
      "step": 1418
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5690682306896664,
      "learning_rate": 3.836400903920373e-05,
      "loss": 0.2086,
      "step": 1419
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7196154566892891,
      "learning_rate": 3.836118986279419e-05,
      "loss": 0.2744,
      "step": 1420
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.651059674331848,
      "learning_rate": 3.835836836320689e-05,
      "loss": 0.2084,
      "step": 1421
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6362643346314623,
      "learning_rate": 3.835554454079882e-05,
      "loss": 0.2203,
      "step": 1422
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5161123284770505,
      "learning_rate": 3.835271839592727e-05,
      "loss": 0.2167,
      "step": 1423
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5383637640484493,
      "learning_rate": 3.834988992894983e-05,
      "loss": 0.1836,
      "step": 1424
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6469490709144389,
      "learning_rate": 3.834705914022437e-05,
      "loss": 0.2632,
      "step": 1425
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5233903835214193,
      "learning_rate": 3.834422603010906e-05,
      "loss": 0.1933,
      "step": 1426
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5650177082308085,
      "learning_rate": 3.834139059896237e-05,
      "loss": 0.2063,
      "step": 1427
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5327804622918416,
      "learning_rate": 3.833855284714305e-05,
      "loss": 0.2033,
      "step": 1428
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6364780275218678,
      "learning_rate": 3.8335712775010155e-05,
      "loss": 0.2276,
      "step": 1429
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7360250319728143,
      "learning_rate": 3.833287038292303e-05,
      "loss": 0.2505,
      "step": 1430
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5623256890103914,
      "learning_rate": 3.8330025671241304e-05,
      "loss": 0.1774,
      "step": 1431
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.625355929115744,
      "learning_rate": 3.832717864032492e-05,
      "loss": 0.2196,
      "step": 1432
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5848711610404335,
      "learning_rate": 3.83243292905341e-05,
      "loss": 0.201,
      "step": 1433
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6313675289463488,
      "learning_rate": 3.832147762222936e-05,
      "loss": 0.2112,
      "step": 1434
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5584681976710097,
      "learning_rate": 3.831862363577151e-05,
      "loss": 0.1603,
      "step": 1435
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.619959869854945,
      "learning_rate": 3.8315767331521655e-05,
      "loss": 0.2016,
      "step": 1436
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7049163040910001,
      "learning_rate": 3.83129087098412e-05,
      "loss": 0.2849,
      "step": 1437
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6509255623604149,
      "learning_rate": 3.831004777109183e-05,
      "loss": 0.2381,
      "step": 1438
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6440643931679848,
      "learning_rate": 3.830718451563553e-05,
      "loss": 0.2237,
      "step": 1439
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.535347368404155,
      "learning_rate": 3.8304318943834584e-05,
      "loss": 0.2123,
      "step": 1440
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5214642010047621,
      "learning_rate": 3.830145105605156e-05,
      "loss": 0.2057,
      "step": 1441
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5755475816221367,
      "learning_rate": 3.8298580852649316e-05,
      "loss": 0.2069,
      "step": 1442
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5464044453845005,
      "learning_rate": 3.8295708333991015e-05,
      "loss": 0.171,
      "step": 1443
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4922906979349393,
      "learning_rate": 3.82928335004401e-05,
      "loss": 0.2265,
      "step": 1444
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5361928437190463,
      "learning_rate": 3.8289956352360326e-05,
      "loss": 0.2056,
      "step": 1445
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6365842852752848,
      "learning_rate": 3.828707689011572e-05,
      "loss": 0.2376,
      "step": 1446
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5581368131337875,
      "learning_rate": 3.828419511407062e-05,
      "loss": 0.1818,
      "step": 1447
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6162437273058337,
      "learning_rate": 3.828131102458962e-05,
      "loss": 0.2223,
      "step": 1448
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7446113759463118,
      "learning_rate": 3.8278424622037674e-05,
      "loss": 0.1992,
      "step": 1449
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6518321515317013,
      "learning_rate": 3.827553590677996e-05,
      "loss": 0.1876,
      "step": 1450
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4870119304820982,
      "learning_rate": 3.827264487918198e-05,
      "loss": 0.2002,
      "step": 1451
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5429456417737526,
      "learning_rate": 3.8269751539609525e-05,
      "loss": 0.1738,
      "step": 1452
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7087697106098125,
      "learning_rate": 3.8266855888428696e-05,
      "loss": 0.2639,
      "step": 1453
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5556042040611595,
      "learning_rate": 3.8263957926005855e-05,
      "loss": 0.2325,
      "step": 1454
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6686974727643898,
      "learning_rate": 3.826105765270767e-05,
      "loss": 0.2586,
      "step": 1455
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5034636098483164,
      "learning_rate": 3.825815506890111e-05,
      "loss": 0.1693,
      "step": 1456
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5640029167766091,
      "learning_rate": 3.825525017495343e-05,
      "loss": 0.2198,
      "step": 1457
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4213213289697031,
      "learning_rate": 3.825234297123216e-05,
      "loss": 0.1295,
      "step": 1458
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5488004575658648,
      "learning_rate": 3.824943345810516e-05,
      "loss": 0.2173,
      "step": 1459
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6090241914056581,
      "learning_rate": 3.824652163594056e-05,
      "loss": 0.2189,
      "step": 1460
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6371886268968706,
      "learning_rate": 3.824360750510676e-05,
      "loss": 0.2531,
      "step": 1461
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5803139460909753,
      "learning_rate": 3.8240691065972486e-05,
      "loss": 0.1876,
      "step": 1462
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6438605315490373,
      "learning_rate": 3.823777231890676e-05,
      "loss": 0.2456,
      "step": 1463
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6280611651043215,
      "learning_rate": 3.823485126427886e-05,
      "loss": 0.2457,
      "step": 1464
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4604206743828294,
      "learning_rate": 3.823192790245839e-05,
      "loss": 0.1692,
      "step": 1465
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5265680895771866,
      "learning_rate": 3.822900223381522e-05,
      "loss": 0.1847,
      "step": 1466
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6115156136045458,
      "learning_rate": 3.822607425871954e-05,
      "loss": 0.2119,
      "step": 1467
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7013657920693921,
      "learning_rate": 3.8223143977541806e-05,
      "loss": 0.2722,
      "step": 1468
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8230019215391029,
      "learning_rate": 3.822021139065279e-05,
      "loss": 0.2843,
      "step": 1469
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6136283863958748,
      "learning_rate": 3.821727649842352e-05,
      "loss": 0.2236,
      "step": 1470
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.695559147153752,
      "learning_rate": 3.821433930122536e-05,
      "loss": 0.2726,
      "step": 1471
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5096054117696908,
      "learning_rate": 3.821139979942992e-05,
      "loss": 0.1532,
      "step": 1472
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5443504920111407,
      "learning_rate": 3.820845799340915e-05,
      "loss": 0.2026,
      "step": 1473
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6468763286927072,
      "learning_rate": 3.820551388353525e-05,
      "loss": 0.2244,
      "step": 1474
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6628971420983781,
      "learning_rate": 3.8202567470180734e-05,
      "loss": 0.236,
      "step": 1475
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9479873581960843,
      "learning_rate": 3.819961875371839e-05,
      "loss": 0.2186,
      "step": 1476
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5568327187478903,
      "learning_rate": 3.819666773452133e-05,
      "loss": 0.1855,
      "step": 1477
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6491061979524193,
      "learning_rate": 3.819371441296292e-05,
      "loss": 0.2201,
      "step": 1478
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6765599257351926,
      "learning_rate": 3.819075878941685e-05,
      "loss": 0.2224,
      "step": 1479
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5868176271550007,
      "learning_rate": 3.8187800864257065e-05,
      "loss": 0.2025,
      "step": 1480
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6483809664797203,
      "learning_rate": 3.818484063785783e-05,
      "loss": 0.1787,
      "step": 1481
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.607151965064967,
      "learning_rate": 3.818187811059369e-05,
      "loss": 0.2227,
      "step": 1482
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6682745919331851,
      "learning_rate": 3.817891328283949e-05,
      "loss": 0.2553,
      "step": 1483
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6850585284950187,
      "learning_rate": 3.817594615497035e-05,
      "loss": 0.2702,
      "step": 1484
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6719758379472411,
      "learning_rate": 3.817297672736169e-05,
      "loss": 0.2198,
      "step": 1485
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6036882616546373,
      "learning_rate": 3.817000500038924e-05,
      "loss": 0.2197,
      "step": 1486
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7633477032723397,
      "learning_rate": 3.816703097442898e-05,
      "loss": 0.2201,
      "step": 1487
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.48237705408604487,
      "learning_rate": 3.8164054649857206e-05,
      "loss": 0.1883,
      "step": 1488
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6028399095093094,
      "learning_rate": 3.816107602705052e-05,
      "loss": 0.2148,
      "step": 1489
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5465009500036333,
      "learning_rate": 3.815809510638578e-05,
      "loss": 0.1569,
      "step": 1490
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9283049104015987,
      "learning_rate": 3.815511188824016e-05,
      "loss": 0.3973,
      "step": 1491
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.544770195746384,
      "learning_rate": 3.81521263729911e-05,
      "loss": 0.1509,
      "step": 1492
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4282058138346209,
      "learning_rate": 3.8149138561016377e-05,
      "loss": 0.1703,
      "step": 1493
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5242653463606947,
      "learning_rate": 3.8146148452694e-05,
      "loss": 0.1906,
      "step": 1494
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5849806538667813,
      "learning_rate": 3.814315604840231e-05,
      "loss": 0.2413,
      "step": 1495
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5687919622179289,
      "learning_rate": 3.8140161348519924e-05,
      "loss": 0.225,
      "step": 1496
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5605160641592873,
      "learning_rate": 3.813716435342575e-05,
      "loss": 0.2032,
      "step": 1497
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5818536730749441,
      "learning_rate": 3.813416506349899e-05,
      "loss": 0.1825,
      "step": 1498
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5871388136741383,
      "learning_rate": 3.813116347911913e-05,
      "loss": 0.212,
      "step": 1499
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5426386097095636,
      "learning_rate": 3.8128159600665954e-05,
      "loss": 0.1559,
      "step": 1500
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5433230444844735,
      "learning_rate": 3.812515342851953e-05,
      "loss": 0.1368,
      "step": 1501
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6322249522408151,
      "learning_rate": 3.812214496306022e-05,
      "loss": 0.2408,
      "step": 1502
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6238523436773569,
      "learning_rate": 3.8119134204668665e-05,
      "loss": 0.1961,
      "step": 1503
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5798299239167037,
      "learning_rate": 3.8116121153725824e-05,
      "loss": 0.211,
      "step": 1504
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6820964655333405,
      "learning_rate": 3.811310581061291e-05,
      "loss": 0.2239,
      "step": 1505
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6070979195882269,
      "learning_rate": 3.8110088175711456e-05,
      "loss": 0.2033,
      "step": 1506
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5318706477308779,
      "learning_rate": 3.8107068249403267e-05,
      "loss": 0.2144,
      "step": 1507
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6492636370146587,
      "learning_rate": 3.810404603207045e-05,
      "loss": 0.2298,
      "step": 1508
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7436351911904817,
      "learning_rate": 3.810102152409539e-05,
      "loss": 0.2862,
      "step": 1509
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6365678760599308,
      "learning_rate": 3.809799472586077e-05,
      "loss": 0.227,
      "step": 1510
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7379404320776938,
      "learning_rate": 3.8094965637749556e-05,
      "loss": 0.3129,
      "step": 1511
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.550846258344613,
      "learning_rate": 3.809193426014501e-05,
      "loss": 0.1946,
      "step": 1512
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5805506724157818,
      "learning_rate": 3.808890059343069e-05,
      "loss": 0.19,
      "step": 1513
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5085162055322425,
      "learning_rate": 3.808586463799042e-05,
      "loss": 0.1545,
      "step": 1514
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5153072996785115,
      "learning_rate": 3.808282639420834e-05,
      "loss": 0.2043,
      "step": 1515
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5088192863053964,
      "learning_rate": 3.807978586246887e-05,
      "loss": 0.1531,
      "step": 1516
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5022200828194179,
      "learning_rate": 3.807674304315671e-05,
      "loss": 0.1609,
      "step": 1517
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.49465972412747927,
      "learning_rate": 3.8073697936656866e-05,
      "loss": 0.1758,
      "step": 1518
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5887060498478996,
      "learning_rate": 3.807065054335461e-05,
      "loss": 0.2244,
      "step": 1519
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6424893952767317,
      "learning_rate": 3.806760086363554e-05,
      "loss": 0.2094,
      "step": 1520
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6669745156668481,
      "learning_rate": 3.8064548897885505e-05,
      "loss": 0.2297,
      "step": 1521
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5918228596940724,
      "learning_rate": 3.806149464649066e-05,
      "loss": 0.2126,
      "step": 1522
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5942133551128759,
      "learning_rate": 3.805843810983745e-05,
      "loss": 0.178,
      "step": 1523
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5799344588112726,
      "learning_rate": 3.8055379288312625e-05,
      "loss": 0.1831,
      "step": 1524
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6139264934073163,
      "learning_rate": 3.805231818230318e-05,
      "loss": 0.2034,
      "step": 1525
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8576155250514091,
      "learning_rate": 3.8049254792196443e-05,
      "loss": 0.2191,
      "step": 1526
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5857760397658246,
      "learning_rate": 3.804618911838001e-05,
      "loss": 0.2073,
      "step": 1527
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6263960029675492,
      "learning_rate": 3.804312116124177e-05,
      "loss": 0.211,
      "step": 1528
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6773208763897873,
      "learning_rate": 3.80400509211699e-05,
      "loss": 0.241,
      "step": 1529
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6706675836103072,
      "learning_rate": 3.8036978398552876e-05,
      "loss": 0.2997,
      "step": 1530
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5079727506876048,
      "learning_rate": 3.803390359377944e-05,
      "loss": 0.1758,
      "step": 1531
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5475943058261229,
      "learning_rate": 3.803082650723864e-05,
      "loss": 0.1848,
      "step": 1532
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.48245659212915665,
      "learning_rate": 3.802774713931982e-05,
      "loss": 0.1871,
      "step": 1533
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5889031951169206,
      "learning_rate": 3.802466549041258e-05,
      "loss": 0.261,
      "step": 1534
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6305125792063027,
      "learning_rate": 3.802158156090685e-05,
      "loss": 0.2286,
      "step": 1535
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.587873312863154,
      "learning_rate": 3.8018495351192825e-05,
      "loss": 0.2003,
      "step": 1536
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7375998203160486,
      "learning_rate": 3.801540686166099e-05,
      "loss": 0.2746,
      "step": 1537
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5119382664075481,
      "learning_rate": 3.801231609270212e-05,
      "loss": 0.1534,
      "step": 1538
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5164101720248367,
      "learning_rate": 3.800922304470728e-05,
      "loss": 0.1495,
      "step": 1539
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44401728344153946,
      "learning_rate": 3.800612771806781e-05,
      "loss": 0.1461,
      "step": 1540
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5497660377699297,
      "learning_rate": 3.8003030113175376e-05,
      "loss": 0.1828,
      "step": 1541
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5944255699843255,
      "learning_rate": 3.79999302304219e-05,
      "loss": 0.1602,
      "step": 1542
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6211709988136527,
      "learning_rate": 3.799682807019958e-05,
      "loss": 0.1605,
      "step": 1543
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6444652639192185,
      "learning_rate": 3.799372363290095e-05,
      "loss": 0.1967,
      "step": 1544
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5991779726320862,
      "learning_rate": 3.799061691891878e-05,
      "loss": 0.2011,
      "step": 1545
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.50835618478452,
      "learning_rate": 3.798750792864617e-05,
      "loss": 0.1349,
      "step": 1546
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.588203777668044,
      "learning_rate": 3.798439666247648e-05,
      "loss": 0.2122,
      "step": 1547
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6501399980680594,
      "learning_rate": 3.798128312080336e-05,
      "loss": 0.3295,
      "step": 1548
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5657072292010945,
      "learning_rate": 3.797816730402077e-05,
      "loss": 0.1815,
      "step": 1549
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5442527025722538,
      "learning_rate": 3.7975049212522934e-05,
      "loss": 0.2255,
      "step": 1550
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5618929766536301,
      "learning_rate": 3.7971928846704385e-05,
      "loss": 0.1894,
      "step": 1551
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6628438426174117,
      "learning_rate": 3.7968806206959915e-05,
      "loss": 0.2312,
      "step": 1552
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.582174910666719,
      "learning_rate": 3.7965681293684635e-05,
      "loss": 0.1811,
      "step": 1553
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7423726704142294,
      "learning_rate": 3.7962554107273926e-05,
      "loss": 0.2482,
      "step": 1554
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6158164767391765,
      "learning_rate": 3.7959424648123455e-05,
      "loss": 0.2327,
      "step": 1555
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6030979130646366,
      "learning_rate": 3.795629291662919e-05,
      "loss": 0.2259,
      "step": 1556
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6052458075796123,
      "learning_rate": 3.795315891318737e-05,
      "loss": 0.2314,
      "step": 1557
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5678576060273508,
      "learning_rate": 3.795002263819453e-05,
      "loss": 0.1581,
      "step": 1558
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5745456310108428,
      "learning_rate": 3.79468840920475e-05,
      "loss": 0.2081,
      "step": 1559
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.589193807077525,
      "learning_rate": 3.7943743275143384e-05,
      "loss": 0.2069,
      "step": 1560
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6307064851378155,
      "learning_rate": 3.794060018787958e-05,
      "loss": 0.2159,
      "step": 1561
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6045940356508995,
      "learning_rate": 3.793745483065377e-05,
      "loss": 0.1928,
      "step": 1562
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6465696526376529,
      "learning_rate": 3.793430720386392e-05,
      "loss": 0.2001,
      "step": 1563
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5094112172737284,
      "learning_rate": 3.7931157307908304e-05,
      "loss": 0.1706,
      "step": 1564
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.739874196055498,
      "learning_rate": 3.792800514318546e-05,
      "loss": 0.2881,
      "step": 1565
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5597013150399909,
      "learning_rate": 3.792485071009421e-05,
      "loss": 0.1647,
      "step": 1566
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5684475125798846,
      "learning_rate": 3.792169400903368e-05,
      "loss": 0.1863,
      "step": 1567
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.611883234391413,
      "learning_rate": 3.7918535040403284e-05,
      "loss": 0.2414,
      "step": 1568
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6093756846888682,
      "learning_rate": 3.791537380460271e-05,
      "loss": 0.2236,
      "step": 1569
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5269051599977148,
      "learning_rate": 3.791221030203193e-05,
      "loss": 0.1379,
      "step": 1570
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5963777817849354,
      "learning_rate": 3.790904453309123e-05,
      "loss": 0.2163,
      "step": 1571
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6536606613442913,
      "learning_rate": 3.790587649818115e-05,
      "loss": 0.2195,
      "step": 1572
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.601181033178103,
      "learning_rate": 3.790270619770253e-05,
      "loss": 0.184,
      "step": 1573
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6645972398919309,
      "learning_rate": 3.78995336320565e-05,
      "loss": 0.253,
      "step": 1574
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7148911507321741,
      "learning_rate": 3.789635880164448e-05,
      "loss": 0.224,
      "step": 1575
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5395672687755909,
      "learning_rate": 3.789318170686816e-05,
      "loss": 0.1898,
      "step": 1576
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6881055765762845,
      "learning_rate": 3.789000234812953e-05,
      "loss": 0.1896,
      "step": 1577
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7216790267672889,
      "learning_rate": 3.788682072583087e-05,
      "loss": 0.2412,
      "step": 1578
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6937682974180902,
      "learning_rate": 3.788363684037473e-05,
      "loss": 0.2679,
      "step": 1579
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6132995942889292,
      "learning_rate": 3.788045069216396e-05,
      "loss": 0.2104,
      "step": 1580
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5429407290905746,
      "learning_rate": 3.787726228160169e-05,
      "loss": 0.1936,
      "step": 1581
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5499775751917583,
      "learning_rate": 3.787407160909134e-05,
      "loss": 0.2203,
      "step": 1582
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7308398441414706,
      "learning_rate": 3.787087867503663e-05,
      "loss": 0.2363,
      "step": 1583
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5083751788921056,
      "learning_rate": 3.786768347984152e-05,
      "loss": 0.1754,
      "step": 1584
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5710690327316926,
      "learning_rate": 3.786448602391031e-05,
      "loss": 0.1857,
      "step": 1585
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5625356027144318,
      "learning_rate": 3.7861286307647555e-05,
      "loss": 0.2163,
      "step": 1586
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5384405667156221,
      "learning_rate": 3.7858084331458106e-05,
      "loss": 0.1521,
      "step": 1587
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5099848425707089,
      "learning_rate": 3.78548800957471e-05,
      "loss": 0.1541,
      "step": 1588
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.48413366027421945,
      "learning_rate": 3.785167360091995e-05,
      "loss": 0.1805,
      "step": 1589
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6015975248370843,
      "learning_rate": 3.7848464847382376e-05,
      "loss": 0.1688,
      "step": 1590
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6369656511205274,
      "learning_rate": 3.7845253835540357e-05,
      "loss": 0.1963,
      "step": 1591
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6443084794201295,
      "learning_rate": 3.7842040565800184e-05,
      "loss": 0.1942,
      "step": 1592
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5279845970916258,
      "learning_rate": 3.78388250385684e-05,
      "loss": 0.2134,
      "step": 1593
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6921526441559828,
      "learning_rate": 3.783560725425188e-05,
      "loss": 0.2127,
      "step": 1594
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.510994930714361,
      "learning_rate": 3.783238721325774e-05,
      "loss": 0.1868,
      "step": 1595
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7549485823194225,
      "learning_rate": 3.782916491599341e-05,
      "loss": 0.277,
      "step": 1596
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45856505374059575,
      "learning_rate": 3.78259403628666e-05,
      "loss": 0.1668,
      "step": 1597
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4931505711286947,
      "learning_rate": 3.782271355428529e-05,
      "loss": 0.1889,
      "step": 1598
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6379140621815674,
      "learning_rate": 3.781948449065777e-05,
      "loss": 0.2453,
      "step": 1599
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6456815684851703,
      "learning_rate": 3.781625317239258e-05,
      "loss": 0.2175,
      "step": 1600
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5364593734237316,
      "learning_rate": 3.7813019599898593e-05,
      "loss": 0.1845,
      "step": 1601
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5400679237420941,
      "learning_rate": 3.780978377358493e-05,
      "loss": 0.1469,
      "step": 1602
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5315345955532083,
      "learning_rate": 3.7806545693861006e-05,
      "loss": 0.1706,
      "step": 1603
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5308179781662627,
      "learning_rate": 3.7803305361136534e-05,
      "loss": 0.197,
      "step": 1604
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5067890123463344,
      "learning_rate": 3.780006277582149e-05,
      "loss": 0.1841,
      "step": 1605
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5677038915851197,
      "learning_rate": 3.7796817938326155e-05,
      "loss": 0.1967,
      "step": 1606
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6025447999752158,
      "learning_rate": 3.7793570849061084e-05,
      "loss": 0.1941,
      "step": 1607
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6173391276625572,
      "learning_rate": 3.7790321508437124e-05,
      "loss": 0.2143,
      "step": 1608
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41026736979468215,
      "learning_rate": 3.77870699168654e-05,
      "loss": 0.148,
      "step": 1609
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5183573489232199,
      "learning_rate": 3.778381607475732e-05,
      "loss": 0.1707,
      "step": 1610
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7336652044529826,
      "learning_rate": 3.778055998252459e-05,
      "loss": 0.2424,
      "step": 1611
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7065554540621118,
      "learning_rate": 3.777730164057919e-05,
      "loss": 0.2664,
      "step": 1612
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6193998979246913,
      "learning_rate": 3.7774041049333374e-05,
      "loss": 0.169,
      "step": 1613
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5427465738990496,
      "learning_rate": 3.777077820919972e-05,
      "loss": 0.1421,
      "step": 1614
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5094403511672391,
      "learning_rate": 3.776751312059103e-05,
      "loss": 0.1523,
      "step": 1615
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7173040191476101,
      "learning_rate": 3.776424578392045e-05,
      "loss": 0.2806,
      "step": 1616
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5899660350026443,
      "learning_rate": 3.7760976199601384e-05,
      "loss": 0.202,
      "step": 1617
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5727841108314073,
      "learning_rate": 3.775770436804751e-05,
      "loss": 0.1982,
      "step": 1618
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5425038714644294,
      "learning_rate": 3.77544302896728e-05,
      "loss": 0.1573,
      "step": 1619
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.441808432943059,
      "learning_rate": 3.775115396489153e-05,
      "loss": 0.1605,
      "step": 1620
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5942919416186491,
      "learning_rate": 3.774787539411821e-05,
      "loss": 0.1775,
      "step": 1621
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7083838996415665,
      "learning_rate": 3.77445945777677e-05,
      "loss": 0.2439,
      "step": 1622
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.503312167307887,
      "learning_rate": 3.7741311516255096e-05,
      "loss": 0.1497,
      "step": 1623
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6952062589938142,
      "learning_rate": 3.773802620999579e-05,
      "loss": 0.2034,
      "step": 1624
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45336381604052195,
      "learning_rate": 3.7734738659405467e-05,
      "loss": 0.1233,
      "step": 1625
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7498056239335655,
      "learning_rate": 3.773144886490007e-05,
      "loss": 0.2313,
      "step": 1626
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6833520080371666,
      "learning_rate": 3.772815682689587e-05,
      "loss": 0.2595,
      "step": 1627
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5355445399883849,
      "learning_rate": 3.7724862545809394e-05,
      "loss": 0.1624,
      "step": 1628
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6489583876249027,
      "learning_rate": 3.772156602205744e-05,
      "loss": 0.2153,
      "step": 1629
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5397262357009679,
      "learning_rate": 3.771826725605713e-05,
      "loss": 0.1741,
      "step": 1630
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6080042509617394,
      "learning_rate": 3.771496624822582e-05,
      "loss": 0.2596,
      "step": 1631
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40624558006229716,
      "learning_rate": 3.771166299898118e-05,
      "loss": 0.1246,
      "step": 1632
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.49826294942834465,
      "learning_rate": 3.770835750874117e-05,
      "loss": 0.1721,
      "step": 1633
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5338173147988655,
      "learning_rate": 3.770504977792402e-05,
      "loss": 0.1818,
      "step": 1634
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6435841392990507,
      "learning_rate": 3.770173980694824e-05,
      "loss": 0.2426,
      "step": 1635
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4587496542864336,
      "learning_rate": 3.7698427596232636e-05,
      "loss": 0.1546,
      "step": 1636
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.536350671526269,
      "learning_rate": 3.769511314619628e-05,
      "loss": 0.1534,
      "step": 1637
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5897124693839668,
      "learning_rate": 3.7691796457258546e-05,
      "loss": 0.2191,
      "step": 1638
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.547712502195856,
      "learning_rate": 3.7688477529839085e-05,
      "loss": 0.1918,
      "step": 1639
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5148067339497486,
      "learning_rate": 3.7685156364357825e-05,
      "loss": 0.1872,
      "step": 1640
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8071489187391565,
      "learning_rate": 3.7681832961234974e-05,
      "loss": 0.2559,
      "step": 1641
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.515770689293681,
      "learning_rate": 3.767850732089105e-05,
      "loss": 0.2022,
      "step": 1642
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.646041626309522,
      "learning_rate": 3.7675179443746816e-05,
      "loss": 0.1904,
      "step": 1643
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8126071345781976,
      "learning_rate": 3.7671849330223345e-05,
      "loss": 0.2483,
      "step": 1644
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6605583185414945,
      "learning_rate": 3.7668516980741995e-05,
      "loss": 0.2469,
      "step": 1645
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5948719667476394,
      "learning_rate": 3.766518239572437e-05,
      "loss": 0.1931,
      "step": 1646
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5043237124983753,
      "learning_rate": 3.7661845575592405e-05,
      "loss": 0.1832,
      "step": 1647
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6381973930565653,
      "learning_rate": 3.76585065207683e-05,
      "loss": 0.2653,
      "step": 1648
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5121307170735959,
      "learning_rate": 3.765516523167452e-05,
      "loss": 0.1394,
      "step": 1649
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4798069093921884,
      "learning_rate": 3.765182170873383e-05,
      "loss": 0.1774,
      "step": 1650
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.46785134958780095,
      "learning_rate": 3.7648475952369275e-05,
      "loss": 0.1238,
      "step": 1651
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5255992149170073,
      "learning_rate": 3.7645127963004176e-05,
      "loss": 0.1604,
      "step": 1652
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.570791513646763,
      "learning_rate": 3.764177774106216e-05,
      "loss": 0.1968,
      "step": 1653
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.566266930644921,
      "learning_rate": 3.76384252869671e-05,
      "loss": 0.2006,
      "step": 1654
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.47890697746781674,
      "learning_rate": 3.763507060114319e-05,
      "loss": 0.1639,
      "step": 1655
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6090882188936215,
      "learning_rate": 3.7631713684014866e-05,
      "loss": 0.2415,
      "step": 1656
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5919254780143545,
      "learning_rate": 3.762835453600688e-05,
      "loss": 0.2112,
      "step": 1657
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5127920876856659,
      "learning_rate": 3.7624993157544246e-05,
      "loss": 0.1378,
      "step": 1658
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7165278207844192,
      "learning_rate": 3.762162954905228e-05,
      "loss": 0.2499,
      "step": 1659
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5861346104005456,
      "learning_rate": 3.761826371095655e-05,
      "loss": 0.2304,
      "step": 1660
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.36341851215175336,
      "learning_rate": 3.761489564368294e-05,
      "loss": 0.1097,
      "step": 1661
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4778832129676479,
      "learning_rate": 3.7611525347657584e-05,
      "loss": 0.1464,
      "step": 1662
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4530170246255666,
      "learning_rate": 3.760815282330694e-05,
      "loss": 0.1391,
      "step": 1663
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5287639567379668,
      "learning_rate": 3.7604778071057685e-05,
      "loss": 0.1809,
      "step": 1664
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5484551625246934,
      "learning_rate": 3.760140109133684e-05,
      "loss": 0.1664,
      "step": 1665
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5592885797756405,
      "learning_rate": 3.759802188457168e-05,
      "loss": 0.198,
      "step": 1666
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4924674373357225,
      "learning_rate": 3.759464045118976e-05,
      "loss": 0.1862,
      "step": 1667
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8881317844977353,
      "learning_rate": 3.759125679161893e-05,
      "loss": 0.3135,
      "step": 1668
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5645354055868844,
      "learning_rate": 3.75878709062873e-05,
      "loss": 0.1924,
      "step": 1669
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41503335387847956,
      "learning_rate": 3.758448279562327e-05,
      "loss": 0.1237,
      "step": 1670
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5282994405462973,
      "learning_rate": 3.758109246005554e-05,
      "loss": 0.1439,
      "step": 1671
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4725423313030016,
      "learning_rate": 3.757769990001308e-05,
      "loss": 0.1197,
      "step": 1672
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6196819069321887,
      "learning_rate": 3.757430511592513e-05,
      "loss": 0.2403,
      "step": 1673
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.486460533654453,
      "learning_rate": 3.757090810822122e-05,
      "loss": 0.1762,
      "step": 1674
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6898722900846119,
      "learning_rate": 3.756750887733116e-05,
      "loss": 0.28,
      "step": 1675
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4578229942030372,
      "learning_rate": 3.756410742368505e-05,
      "loss": 0.1451,
      "step": 1676
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5826117162035149,
      "learning_rate": 3.7560703747713255e-05,
      "loss": 0.1839,
      "step": 1677
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6236636896371596,
      "learning_rate": 3.7557297849846444e-05,
      "loss": 0.1963,
      "step": 1678
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.49030850522496977,
      "learning_rate": 3.7553889730515545e-05,
      "loss": 0.1598,
      "step": 1679
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6157030046489759,
      "learning_rate": 3.7550479390151766e-05,
      "loss": 0.2048,
      "step": 1680
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6096990164055758,
      "learning_rate": 3.7547066829186626e-05,
      "loss": 0.14,
      "step": 1681
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.565142438327975,
      "learning_rate": 3.754365204805189e-05,
      "loss": 0.1619,
      "step": 1682
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7016368163918355,
      "learning_rate": 3.754023504717962e-05,
      "loss": 0.2143,
      "step": 1683
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.625176309513068,
      "learning_rate": 3.753681582700216e-05,
      "loss": 0.1814,
      "step": 1684
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.590324307147436,
      "learning_rate": 3.7533394387952134e-05,
      "loss": 0.1537,
      "step": 1685
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6442467263804328,
      "learning_rate": 3.752997073046244e-05,
      "loss": 0.1872,
      "step": 1686
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.59178876713905,
      "learning_rate": 3.752654485496627e-05,
      "loss": 0.2091,
      "step": 1687
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5640764926854976,
      "learning_rate": 3.752311676189708e-05,
      "loss": 0.157,
      "step": 1688
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6923073968316125,
      "learning_rate": 3.7519686451688614e-05,
      "loss": 0.2069,
      "step": 1689
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7447985846636116,
      "learning_rate": 3.75162539247749e-05,
      "loss": 0.207,
      "step": 1690
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5239470549545716,
      "learning_rate": 3.751281918159025e-05,
      "loss": 0.1244,
      "step": 1691
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5805904040177313,
      "learning_rate": 3.750938222256924e-05,
      "loss": 0.2058,
      "step": 1692
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6153837476779249,
      "learning_rate": 3.7505943048146736e-05,
      "loss": 0.2012,
      "step": 1693
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5391845288577851,
      "learning_rate": 3.75025016587579e-05,
      "loss": 0.2016,
      "step": 1694
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40337075940915457,
      "learning_rate": 3.749905805483814e-05,
      "loss": 0.1364,
      "step": 1695
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5676900061585956,
      "learning_rate": 3.7495612236823175e-05,
      "loss": 0.1815,
      "step": 1696
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6269227968290026,
      "learning_rate": 3.749216420514898e-05,
      "loss": 0.2448,
      "step": 1697
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39955308355498004,
      "learning_rate": 3.7488713960251845e-05,
      "loss": 0.1234,
      "step": 1698
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5478316794746595,
      "learning_rate": 3.74852615025683e-05,
      "loss": 0.1723,
      "step": 1699
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6297119855261379,
      "learning_rate": 3.748180683253518e-05,
      "loss": 0.1661,
      "step": 1700
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6537571061296992,
      "learning_rate": 3.747834995058959e-05,
      "loss": 0.1773,
      "step": 1701
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6101084232512916,
      "learning_rate": 3.747489085716891e-05,
      "loss": 0.179,
      "step": 1702
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5618122187537455,
      "learning_rate": 3.747142955271081e-05,
      "loss": 0.159,
      "step": 1703
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6883530742740928,
      "learning_rate": 3.746796603765325e-05,
      "loss": 0.2091,
      "step": 1704
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6575283134369131,
      "learning_rate": 3.7464500312434445e-05,
      "loss": 0.22,
      "step": 1705
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4634539560494647,
      "learning_rate": 3.7461032377492905e-05,
      "loss": 0.1117,
      "step": 1706
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5572836391072831,
      "learning_rate": 3.745756223326741e-05,
      "loss": 0.2117,
      "step": 1707
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5723314995981625,
      "learning_rate": 3.745408988019703e-05,
      "loss": 0.1732,
      "step": 1708
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.599343679005534,
      "learning_rate": 3.7450615318721115e-05,
      "loss": 0.1726,
      "step": 1709
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6137821557348229,
      "learning_rate": 3.744713854927928e-05,
      "loss": 0.2148,
      "step": 1710
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6366886061571443,
      "learning_rate": 3.744365957231143e-05,
      "loss": 0.2418,
      "step": 1711
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6688885964066988,
      "learning_rate": 3.7440178388257746e-05,
      "loss": 0.1802,
      "step": 1712
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6196204651602569,
      "learning_rate": 3.74366949975587e-05,
      "loss": 0.2114,
      "step": 1713
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5522774199244669,
      "learning_rate": 3.743320940065503e-05,
      "loss": 0.1842,
      "step": 1714
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4462534683691138,
      "learning_rate": 3.742972159798775e-05,
      "loss": 0.1449,
      "step": 1715
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6269447825224383,
      "learning_rate": 3.7426231589998166e-05,
      "loss": 0.2665,
      "step": 1716
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6496119835778836,
      "learning_rate": 3.742273937712785e-05,
      "loss": 0.2205,
      "step": 1717
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7138850970441734,
      "learning_rate": 3.741924495981867e-05,
      "loss": 0.2088,
      "step": 1718
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44922088954250167,
      "learning_rate": 3.7415748338512746e-05,
      "loss": 0.1352,
      "step": 1719
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6247904664711037,
      "learning_rate": 3.741224951365251e-05,
      "loss": 0.2075,
      "step": 1720
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5113550014441116,
      "learning_rate": 3.740874848568065e-05,
      "loss": 0.1684,
      "step": 1721
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6298551804743383,
      "learning_rate": 3.740524525504014e-05,
      "loss": 0.18,
      "step": 1722
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.48636890531479665,
      "learning_rate": 3.740173982217423e-05,
      "loss": 0.1576,
      "step": 1723
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6872908534191354,
      "learning_rate": 3.739823218752645e-05,
      "loss": 0.2019,
      "step": 1724
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5770330168435931,
      "learning_rate": 3.739472235154061e-05,
      "loss": 0.1737,
      "step": 1725
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5272731592474038,
      "learning_rate": 3.7391210314660796e-05,
      "loss": 0.1494,
      "step": 1726
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5441748033950204,
      "learning_rate": 3.738769607733138e-05,
      "loss": 0.155,
      "step": 1727
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.565452403928258,
      "learning_rate": 3.7384179639996997e-05,
      "loss": 0.1688,
      "step": 1728
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5876894128109672,
      "learning_rate": 3.738066100310257e-05,
      "loss": 0.1604,
      "step": 1729
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5313991169056789,
      "learning_rate": 3.7377140167093316e-05,
      "loss": 0.1362,
      "step": 1730
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5479566503907015,
      "learning_rate": 3.7373617132414686e-05,
      "loss": 0.1948,
      "step": 1731
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6559345531715473,
      "learning_rate": 3.7370091899512464e-05,
      "loss": 0.2034,
      "step": 1732
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4808759746893655,
      "learning_rate": 3.736656446883267e-05,
      "loss": 0.1685,
      "step": 1733
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.55910713772435,
      "learning_rate": 3.736303484082163e-05,
      "loss": 0.1424,
      "step": 1734
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4510887531813745,
      "learning_rate": 3.735950301592592e-05,
      "loss": 0.1314,
      "step": 1735
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6666557837631132,
      "learning_rate": 3.7355968994592414e-05,
      "loss": 0.228,
      "step": 1736
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7322084650024974,
      "learning_rate": 3.735243277726828e-05,
      "loss": 0.2417,
      "step": 1737
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5170557228768408,
      "learning_rate": 3.7348894364400914e-05,
      "loss": 0.1377,
      "step": 1738
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5642680356376106,
      "learning_rate": 3.7345353756438025e-05,
      "loss": 0.1888,
      "step": 1739
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.621761179336891,
      "learning_rate": 3.734181095382761e-05,
      "loss": 0.2171,
      "step": 1740
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5121415699165283,
      "learning_rate": 3.733826595701792e-05,
      "loss": 0.1633,
      "step": 1741
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6046477134685961,
      "learning_rate": 3.733471876645749e-05,
      "loss": 0.1302,
      "step": 1742
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5619993631047071,
      "learning_rate": 3.733116938259513e-05,
      "loss": 0.1937,
      "step": 1743
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5190376288640828,
      "learning_rate": 3.732761780587993e-05,
      "loss": 0.1653,
      "step": 1744
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6022405394954778,
      "learning_rate": 3.732406403676126e-05,
      "loss": 0.1967,
      "step": 1745
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5974996274284574,
      "learning_rate": 3.732050807568878e-05,
      "loss": 0.1833,
      "step": 1746
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7429531765658418,
      "learning_rate": 3.731694992311239e-05,
      "loss": 0.2606,
      "step": 1747
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43786074887801946,
      "learning_rate": 3.7313389579482315e-05,
      "loss": 0.1441,
      "step": 1748
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6534081643742562,
      "learning_rate": 3.730982704524901e-05,
      "loss": 0.1926,
      "step": 1749
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4990355309142661,
      "learning_rate": 3.7306262320863245e-05,
      "loss": 0.1571,
      "step": 1750
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4616424536237072,
      "learning_rate": 3.730269540677606e-05,
      "loss": 0.16,
      "step": 1751
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45670773394020703,
      "learning_rate": 3.729912630343874e-05,
      "loss": 0.121,
      "step": 1752
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4940831750335586,
      "learning_rate": 3.729555501130288e-05,
      "loss": 0.1705,
      "step": 1753
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5267602871676524,
      "learning_rate": 3.729198153082036e-05,
      "loss": 0.1549,
      "step": 1754
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4261168111303581,
      "learning_rate": 3.7288405862443296e-05,
      "loss": 0.1308,
      "step": 1755
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4561122229018286,
      "learning_rate": 3.7284828006624125e-05,
      "loss": 0.1284,
      "step": 1756
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.840420050613432,
      "learning_rate": 3.728124796381553e-05,
      "loss": 0.2685,
      "step": 1757
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4494623767491376,
      "learning_rate": 3.7277665734470476e-05,
      "loss": 0.1298,
      "step": 1758
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5002308848883505,
      "learning_rate": 3.727408131904223e-05,
      "loss": 0.1509,
      "step": 1759
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5140394628864617,
      "learning_rate": 3.727049471798429e-05,
      "loss": 0.1425,
      "step": 1760
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4895479480319752,
      "learning_rate": 3.726690593175048e-05,
      "loss": 0.1351,
      "step": 1761
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5520572896128711,
      "learning_rate": 3.726331496079486e-05,
      "loss": 0.1739,
      "step": 1762
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7585251200595686,
      "learning_rate": 3.725972180557179e-05,
      "loss": 0.2004,
      "step": 1763
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4590632918671075,
      "learning_rate": 3.7256126466535896e-05,
      "loss": 0.1442,
      "step": 1764
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6287297066206385,
      "learning_rate": 3.725252894414209e-05,
      "loss": 0.2479,
      "step": 1765
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5816323366208689,
      "learning_rate": 3.724892923884555e-05,
      "loss": 0.2193,
      "step": 1766
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5379167072563048,
      "learning_rate": 3.724532735110174e-05,
      "loss": 0.1883,
      "step": 1767
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6668542856218728,
      "learning_rate": 3.724172328136638e-05,
      "loss": 0.2434,
      "step": 1768
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45049135482867525,
      "learning_rate": 3.723811703009549e-05,
      "loss": 0.1624,
      "step": 1769
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45224688387187884,
      "learning_rate": 3.723450859774536e-05,
      "loss": 0.1191,
      "step": 1770
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5132096970849164,
      "learning_rate": 3.723089798477254e-05,
      "loss": 0.1608,
      "step": 1771
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5183383756292147,
      "learning_rate": 3.7227285191633894e-05,
      "loss": 0.1645,
      "step": 1772
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5725595806385939,
      "learning_rate": 3.722367021878651e-05,
      "loss": 0.1745,
      "step": 1773
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6510874363387364,
      "learning_rate": 3.722005306668778e-05,
      "loss": 0.1843,
      "step": 1774
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.548747996932699,
      "learning_rate": 3.721643373579538e-05,
      "loss": 0.158,
      "step": 1775
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5105141592014346,
      "learning_rate": 3.721281222656725e-05,
      "loss": 0.1735,
      "step": 1776
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6106253048243001,
      "learning_rate": 3.7209188539461606e-05,
      "loss": 0.2014,
      "step": 1777
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.49390342356794187,
      "learning_rate": 3.7205562674936945e-05,
      "loss": 0.1507,
      "step": 1778
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7171510242502983,
      "learning_rate": 3.720193463345202e-05,
      "loss": 0.2032,
      "step": 1779
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5795244599430791,
      "learning_rate": 3.719830441546589e-05,
      "loss": 0.1691,
      "step": 1780
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4928726090448924,
      "learning_rate": 3.719467202143786e-05,
      "loss": 0.1702,
      "step": 1781
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5353202011792603,
      "learning_rate": 3.7191037451827545e-05,
      "loss": 0.1768,
      "step": 1782
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5293257589306753,
      "learning_rate": 3.71874007070948e-05,
      "loss": 0.2061,
      "step": 1783
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6033963576307103,
      "learning_rate": 3.718376178769976e-05,
      "loss": 0.2347,
      "step": 1784
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.563174823119198,
      "learning_rate": 3.7180120694102864e-05,
      "loss": 0.1657,
      "step": 1785
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6567917585757848,
      "learning_rate": 3.71764774267648e-05,
      "loss": 0.254,
      "step": 1786
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5453203173407706,
      "learning_rate": 3.717283198614654e-05,
      "loss": 0.1697,
      "step": 1787
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5568532169675561,
      "learning_rate": 3.716918437270932e-05,
      "loss": 0.1702,
      "step": 1788
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4480953433120678,
      "learning_rate": 3.7165534586914674e-05,
      "loss": 0.1469,
      "step": 1789
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5939011632334655,
      "learning_rate": 3.7161882629224386e-05,
      "loss": 0.1693,
      "step": 1790
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5252234040362398,
      "learning_rate": 3.715822850010053e-05,
      "loss": 0.1802,
      "step": 1791
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6148327682270539,
      "learning_rate": 3.7154572200005446e-05,
      "loss": 0.1868,
      "step": 1792
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5877340368036394,
      "learning_rate": 3.7150913729401754e-05,
      "loss": 0.2021,
      "step": 1793
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6031817404586958,
      "learning_rate": 3.714725308875236e-05,
      "loss": 0.2198,
      "step": 1794
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5666927533830229,
      "learning_rate": 3.714359027852041e-05,
      "loss": 0.1631,
      "step": 1795
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5845618821239438,
      "learning_rate": 3.713992529916936e-05,
      "loss": 0.1885,
      "step": 1796
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4468047920508807,
      "learning_rate": 3.713625815116293e-05,
      "loss": 0.1596,
      "step": 1797
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6655990326880111,
      "learning_rate": 3.7132588834965104e-05,
      "loss": 0.2263,
      "step": 1798
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42030102668049585,
      "learning_rate": 3.712891735104015e-05,
      "loss": 0.1408,
      "step": 1799
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.48522803348345994,
      "learning_rate": 3.712524369985262e-05,
      "loss": 0.1396,
      "step": 1800
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5247253687394966,
      "learning_rate": 3.712156788186731e-05,
      "loss": 0.1711,
      "step": 1801
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46044394367711705,
      "learning_rate": 3.711788989754931e-05,
      "loss": 0.1771,
      "step": 1802
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6248716222523348,
      "learning_rate": 3.711420974736399e-05,
      "loss": 0.1769,
      "step": 1803
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5399353314018481,
      "learning_rate": 3.711052743177699e-05,
      "loss": 0.1872,
      "step": 1804
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5948829883633042,
      "learning_rate": 3.7106842951254216e-05,
      "loss": 0.2361,
      "step": 1805
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5504957500747597,
      "learning_rate": 3.710315630626185e-05,
      "loss": 0.1893,
      "step": 1806
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5724746270366653,
      "learning_rate": 3.7099467497266354e-05,
      "loss": 0.1768,
      "step": 1807
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4662508144831766,
      "learning_rate": 3.7095776524734464e-05,
      "loss": 0.1377,
      "step": 1808
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5384959962123979,
      "learning_rate": 3.7092083389133174e-05,
      "loss": 0.1632,
      "step": 1809
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46400492982156966,
      "learning_rate": 3.7088388090929776e-05,
      "loss": 0.1385,
      "step": 1810
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6797547142732374,
      "learning_rate": 3.708469063059181e-05,
      "loss": 0.1942,
      "step": 1811
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6242689625736262,
      "learning_rate": 3.708099100858712e-05,
      "loss": 0.1859,
      "step": 1812
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5202288425768358,
      "learning_rate": 3.70772892253838e-05,
      "loss": 0.1377,
      "step": 1813
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5689157862651102,
      "learning_rate": 3.7073585281450206e-05,
      "loss": 0.1725,
      "step": 1814
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4489943233972253,
      "learning_rate": 3.706987917725501e-05,
      "loss": 0.1281,
      "step": 1815
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4661901513313358,
      "learning_rate": 3.706617091326712e-05,
      "loss": 0.1585,
      "step": 1816
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5624394914083349,
      "learning_rate": 3.7062460489955736e-05,
      "loss": 0.16,
      "step": 1817
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5653668040883995,
      "learning_rate": 3.705874790779032e-05,
      "loss": 0.1983,
      "step": 1818
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7204777641783155,
      "learning_rate": 3.705503316724062e-05,
      "loss": 0.2443,
      "step": 1819
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.512092570269932,
      "learning_rate": 3.705131626877664e-05,
      "loss": 0.1823,
      "step": 1820
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4458199504228133,
      "learning_rate": 3.704759721286866e-05,
      "loss": 0.155,
      "step": 1821
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5639042280905548,
      "learning_rate": 3.7043875999987254e-05,
      "loss": 0.1822,
      "step": 1822
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4959610588857517,
      "learning_rate": 3.704015263060325e-05,
      "loss": 0.1723,
      "step": 1823
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5557884251450709,
      "learning_rate": 3.7036427105187754e-05,
      "loss": 0.1898,
      "step": 1824
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5829874670929166,
      "learning_rate": 3.703269942421214e-05,
      "loss": 0.2062,
      "step": 1825
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5341867728870532,
      "learning_rate": 3.7028969588148056e-05,
      "loss": 0.1755,
      "step": 1826
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5358242960781303,
      "learning_rate": 3.7025237597467436e-05,
      "loss": 0.1705,
      "step": 1827
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6544398365126588,
      "learning_rate": 3.702150345264247e-05,
      "loss": 0.2301,
      "step": 1828
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.631420653475568,
      "learning_rate": 3.701776715414562e-05,
      "loss": 0.2188,
      "step": 1829
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.49352888455739213,
      "learning_rate": 3.701402870244963e-05,
      "loss": 0.1504,
      "step": 1830
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6030738472864781,
      "learning_rate": 3.701028809802752e-05,
      "loss": 0.1979,
      "step": 1831
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5364972597338402,
      "learning_rate": 3.700654534135257e-05,
      "loss": 0.1594,
      "step": 1832
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4877486414047239,
      "learning_rate": 3.7002800432898344e-05,
      "loss": 0.1747,
      "step": 1833
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5407351651707459,
      "learning_rate": 3.699905337313866e-05,
      "loss": 0.1424,
      "step": 1834
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.558463875137577,
      "learning_rate": 3.6995304162547634e-05,
      "loss": 0.163,
      "step": 1835
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5606333702423673,
      "learning_rate": 3.699155280159964e-05,
      "loss": 0.16,
      "step": 1836
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6374742671036454,
      "learning_rate": 3.698779929076931e-05,
      "loss": 0.2115,
      "step": 1837
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5123410990403009,
      "learning_rate": 3.698404363053158e-05,
      "loss": 0.1207,
      "step": 1838
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6002579452452738,
      "learning_rate": 3.6980285821361636e-05,
      "loss": 0.2014,
      "step": 1839
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5124499796671309,
      "learning_rate": 3.697652586373493e-05,
      "loss": 0.1419,
      "step": 1840
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4431069281353897,
      "learning_rate": 3.69727637581272e-05,
      "loss": 0.1213,
      "step": 1841
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5455442586456671,
      "learning_rate": 3.696899950501447e-05,
      "loss": 0.1432,
      "step": 1842
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5126649102779077,
      "learning_rate": 3.696523310487299e-05,
      "loss": 0.1456,
      "step": 1843
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6071043503194643,
      "learning_rate": 3.6961464558179333e-05,
      "loss": 0.2241,
      "step": 1844
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5503258953381277,
      "learning_rate": 3.695769386541031e-05,
      "loss": 0.1799,
      "step": 1845
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5988472683528538,
      "learning_rate": 3.695392102704302e-05,
      "loss": 0.1783,
      "step": 1846
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6539689338200037,
      "learning_rate": 3.695014604355482e-05,
      "loss": 0.1655,
      "step": 1847
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5720937946119732,
      "learning_rate": 3.694636891542334e-05,
      "loss": 0.1653,
      "step": 1848
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4841327061235579,
      "learning_rate": 3.694258964312649e-05,
      "loss": 0.1502,
      "step": 1849
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6239236145889462,
      "learning_rate": 3.693880822714247e-05,
      "loss": 0.2152,
      "step": 1850
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5698756805168184,
      "learning_rate": 3.6935024667949694e-05,
      "loss": 0.226,
      "step": 1851
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46871614333907585,
      "learning_rate": 3.693123896602692e-05,
      "loss": 0.1581,
      "step": 1852
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4838594338493423,
      "learning_rate": 3.6927451121853104e-05,
      "loss": 0.1349,
      "step": 1853
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5893581993968948,
      "learning_rate": 3.692366113590754e-05,
      "loss": 0.2224,
      "step": 1854
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6799030400669556,
      "learning_rate": 3.6919869008669735e-05,
      "loss": 0.2278,
      "step": 1855
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4517636072403157,
      "learning_rate": 3.691607474061951e-05,
      "loss": 0.1279,
      "step": 1856
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7196763744479866,
      "learning_rate": 3.691227833223693e-05,
      "loss": 0.1242,
      "step": 1857
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5461823027363832,
      "learning_rate": 3.690847978400236e-05,
      "loss": 0.1869,
      "step": 1858
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.49258810724464014,
      "learning_rate": 3.69046790963964e-05,
      "loss": 0.1564,
      "step": 1859
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6603150938446183,
      "learning_rate": 3.690087626989994e-05,
      "loss": 0.2154,
      "step": 1860
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5477166919305028,
      "learning_rate": 3.6897071304994145e-05,
      "loss": 0.1551,
      "step": 1861
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4150405704246504,
      "learning_rate": 3.689326420216044e-05,
      "loss": 0.1295,
      "step": 1862
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4571953503344737,
      "learning_rate": 3.688945496188052e-05,
      "loss": 0.1285,
      "step": 1863
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4778201596211847,
      "learning_rate": 3.6885643584636366e-05,
      "loss": 0.1578,
      "step": 1864
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5003209276222216,
      "learning_rate": 3.688183007091021e-05,
      "loss": 0.1817,
      "step": 1865
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4884668226464562,
      "learning_rate": 3.6878014421184565e-05,
      "loss": 0.1615,
      "step": 1866
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5215038624526275,
      "learning_rate": 3.687419663594221e-05,
      "loss": 0.1533,
      "step": 1867
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5356753817691832,
      "learning_rate": 3.68703767156662e-05,
      "loss": 0.1841,
      "step": 1868
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4813866569892626,
      "learning_rate": 3.686655466083986e-05,
      "loss": 0.1092,
      "step": 1869
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.515183722353877,
      "learning_rate": 3.6862730471946766e-05,
      "loss": 0.1662,
      "step": 1870
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.527457726488649,
      "learning_rate": 3.685890414947079e-05,
      "loss": 0.1326,
      "step": 1871
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.48394821191988074,
      "learning_rate": 3.685507569389606e-05,
      "loss": 0.1471,
      "step": 1872
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5079871830900327,
      "learning_rate": 3.685124510570699e-05,
      "loss": 0.1696,
      "step": 1873
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43550581802703897,
      "learning_rate": 3.6847412385388236e-05,
      "loss": 0.1207,
      "step": 1874
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5934227995535719,
      "learning_rate": 3.684357753342474e-05,
      "loss": 0.1493,
      "step": 1875
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5339223182410467,
      "learning_rate": 3.683974055030172e-05,
      "loss": 0.2084,
      "step": 1876
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5214670014076911,
      "learning_rate": 3.683590143650465e-05,
      "loss": 0.1462,
      "step": 1877
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6247407852506875,
      "learning_rate": 3.6832060192519286e-05,
      "loss": 0.1922,
      "step": 1878
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5663147424902499,
      "learning_rate": 3.682821681883164e-05,
      "loss": 0.1726,
      "step": 1879
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5244185804460151,
      "learning_rate": 3.6824371315928e-05,
      "loss": 0.1455,
      "step": 1880
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5249089866040416,
      "learning_rate": 3.6820523684294934e-05,
      "loss": 0.1265,
      "step": 1881
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5999200370274578,
      "learning_rate": 3.681667392441926e-05,
      "loss": 0.1754,
      "step": 1882
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6153279551225889,
      "learning_rate": 3.6812822036788085e-05,
      "loss": 0.1729,
      "step": 1883
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5995136147242454,
      "learning_rate": 3.680896802188876e-05,
      "loss": 0.1877,
      "step": 1884
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4937274547451559,
      "learning_rate": 3.680511188020893e-05,
      "loss": 0.1444,
      "step": 1885
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6131234178512956,
      "learning_rate": 3.6801253612236506e-05,
      "loss": 0.2362,
      "step": 1886
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5421036339354265,
      "learning_rate": 3.679739321845965e-05,
      "loss": 0.1697,
      "step": 1887
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.45804696637575143,
      "learning_rate": 3.679353069936681e-05,
      "loss": 0.1897,
      "step": 1888
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5354349844454791,
      "learning_rate": 3.678966605544669e-05,
      "loss": 0.1602,
      "step": 1889
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5799963793970424,
      "learning_rate": 3.678579928718827e-05,
      "loss": 0.1542,
      "step": 1890
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4359307244160548,
      "learning_rate": 3.678193039508081e-05,
      "loss": 0.1438,
      "step": 1891
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.47411293784923936,
      "learning_rate": 3.6778059379613815e-05,
      "loss": 0.131,
      "step": 1892
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5953754211574861,
      "learning_rate": 3.6774186241277084e-05,
      "loss": 0.1793,
      "step": 1893
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.499161135792009,
      "learning_rate": 3.6770310980560654e-05,
      "loss": 0.1456,
      "step": 1894
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6174845403210941,
      "learning_rate": 3.676643359795487e-05,
      "loss": 0.2249,
      "step": 1895
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.49905216261274066,
      "learning_rate": 3.676255409395031e-05,
      "loss": 0.1721,
      "step": 1896
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7396950883560727,
      "learning_rate": 3.6758672469037834e-05,
      "loss": 0.2002,
      "step": 1897
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5055737075440553,
      "learning_rate": 3.675478872370858e-05,
      "loss": 0.1593,
      "step": 1898
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4943929638594947,
      "learning_rate": 3.675090285845393e-05,
      "loss": 0.1387,
      "step": 1899
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5991777488079773,
      "learning_rate": 3.674701487376557e-05,
      "loss": 0.1728,
      "step": 1900
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6470686010078436,
      "learning_rate": 3.6743124770135416e-05,
      "loss": 0.1846,
      "step": 1901
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5272540828624529,
      "learning_rate": 3.673923254805566e-05,
      "loss": 0.1325,
      "step": 1902
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5076941132096551,
      "learning_rate": 3.673533820801881e-05,
      "loss": 0.1435,
      "step": 1903
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5036932620807516,
      "learning_rate": 3.6731441750517566e-05,
      "loss": 0.1402,
      "step": 1904
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.571876263226224,
      "learning_rate": 3.6727543176044944e-05,
      "loss": 0.1761,
      "step": 1905
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5024522904383996,
      "learning_rate": 3.672364248509422e-05,
      "loss": 0.1502,
      "step": 1906
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5882948439447192,
      "learning_rate": 3.671973967815894e-05,
      "loss": 0.1908,
      "step": 1907
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4973732675223558,
      "learning_rate": 3.67158347557329e-05,
      "loss": 0.2043,
      "step": 1908
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5413165581021107,
      "learning_rate": 3.671192771831019e-05,
      "loss": 0.1678,
      "step": 1909
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4680968820864525,
      "learning_rate": 3.670801856638514e-05,
      "loss": 0.1414,
      "step": 1910
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4835948198396006,
      "learning_rate": 3.670410730045238e-05,
      "loss": 0.1576,
      "step": 1911
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5014166432721385,
      "learning_rate": 3.6700193921006766e-05,
      "loss": 0.1481,
      "step": 1912
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5560035138019156,
      "learning_rate": 3.669627842854346e-05,
      "loss": 0.175,
      "step": 1913
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6116576357892887,
      "learning_rate": 3.669236082355787e-05,
      "loss": 0.2143,
      "step": 1914
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4306543050685672,
      "learning_rate": 3.668844110654568e-05,
      "loss": 0.1309,
      "step": 1915
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5837273686042967,
      "learning_rate": 3.668451927800283e-05,
      "loss": 0.1632,
      "step": 1916
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41713692433543653,
      "learning_rate": 3.668059533842556e-05,
      "loss": 0.1335,
      "step": 1917
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3997542930489403,
      "learning_rate": 3.667666928831032e-05,
      "loss": 0.1159,
      "step": 1918
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6479163316028705,
      "learning_rate": 3.667274112815387e-05,
      "loss": 0.248,
      "step": 1919
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6217802560221917,
      "learning_rate": 3.666881085845324e-05,
      "loss": 0.1791,
      "step": 1920
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43443352764664045,
      "learning_rate": 3.666487847970571e-05,
      "loss": 0.1211,
      "step": 1921
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.47964265408707923,
      "learning_rate": 3.6660943992408817e-05,
      "loss": 0.1367,
      "step": 1922
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5425305691030274,
      "learning_rate": 3.665700739706038e-05,
      "loss": 0.1529,
      "step": 1923
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5446673383280621,
      "learning_rate": 3.66530686941585e-05,
      "loss": 0.1331,
      "step": 1924
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5984780638968301,
      "learning_rate": 3.664912788420151e-05,
      "loss": 0.1877,
      "step": 1925
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5801971005277564,
      "learning_rate": 3.664518496768802e-05,
      "loss": 0.1695,
      "step": 1926
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5705829658360405,
      "learning_rate": 3.664123994511695e-05,
      "loss": 0.175,
      "step": 1927
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5007147746368455,
      "learning_rate": 3.663729281698741e-05,
      "loss": 0.1513,
      "step": 1928
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6023429168022518,
      "learning_rate": 3.6633343583798836e-05,
      "loss": 0.1747,
      "step": 1929
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5204841651340057,
      "learning_rate": 3.662939224605091e-05,
      "loss": 0.1446,
      "step": 1930
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5331089090242225,
      "learning_rate": 3.6625438804243574e-05,
      "loss": 0.1453,
      "step": 1931
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4469172857858662,
      "learning_rate": 3.6621483258877055e-05,
      "loss": 0.1362,
      "step": 1932
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5490063686696296,
      "learning_rate": 3.661752561045182e-05,
      "loss": 0.1558,
      "step": 1933
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5997796826381633,
      "learning_rate": 3.6613565859468626e-05,
      "loss": 0.2151,
      "step": 1934
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5764123447144598,
      "learning_rate": 3.6609604006428486e-05,
      "loss": 0.2078,
      "step": 1935
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5391405021513669,
      "learning_rate": 3.660564005183268e-05,
      "loss": 0.1846,
      "step": 1936
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4838369827038705,
      "learning_rate": 3.660167399618275e-05,
      "loss": 0.1449,
      "step": 1937
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5380862144289499,
      "learning_rate": 3.659770583998051e-05,
      "loss": 0.189,
      "step": 1938
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6434882076119558,
      "learning_rate": 3.659373558372803e-05,
      "loss": 0.2672,
      "step": 1939
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5245802631949871,
      "learning_rate": 3.658976322792766e-05,
      "loss": 0.22,
      "step": 1940
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.460259796046616,
      "learning_rate": 3.658578877308201e-05,
      "loss": 0.1327,
      "step": 1941
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5019063371159829,
      "learning_rate": 3.658181221969395e-05,
      "loss": 0.167,
      "step": 1942
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4179564322679788,
      "learning_rate": 3.657783356826662e-05,
      "loss": 0.1643,
      "step": 1943
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5073032466603314,
      "learning_rate": 3.657385281930343e-05,
      "loss": 0.1379,
      "step": 1944
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4739893249887038,
      "learning_rate": 3.656986997330804e-05,
      "loss": 0.1877,
      "step": 1945
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.45973387862208753,
      "learning_rate": 3.65658850307844e-05,
      "loss": 0.1388,
      "step": 1946
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.570273254298344,
      "learning_rate": 3.656189799223669e-05,
      "loss": 0.1568,
      "step": 1947
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6250386702972167,
      "learning_rate": 3.65579088581694e-05,
      "loss": 0.1904,
      "step": 1948
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.47787019447407003,
      "learning_rate": 3.6553917629087246e-05,
      "loss": 0.1496,
      "step": 1949
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.47813375377744693,
      "learning_rate": 3.6549924305495225e-05,
      "loss": 0.1475,
      "step": 1950
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6565636838805152,
      "learning_rate": 3.654592888789861e-05,
      "loss": 0.1703,
      "step": 1951
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6033249340825082,
      "learning_rate": 3.6541931376802906e-05,
      "loss": 0.1664,
      "step": 1952
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4774719066244637,
      "learning_rate": 3.653793177271393e-05,
      "loss": 0.1291,
      "step": 1953
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4556537965483882,
      "learning_rate": 3.653393007613771e-05,
      "loss": 0.1302,
      "step": 1954
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6308377619486693,
      "learning_rate": 3.65299262875806e-05,
      "loss": 0.214,
      "step": 1955
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5604642481499463,
      "learning_rate": 3.652592040754917e-05,
      "loss": 0.1741,
      "step": 1956
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46423169721782687,
      "learning_rate": 3.652191243655025e-05,
      "loss": 0.1395,
      "step": 1957
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5776067421262083,
      "learning_rate": 3.651790237509098e-05,
      "loss": 0.2144,
      "step": 1958
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6092808234972482,
      "learning_rate": 3.651389022367874e-05,
      "loss": 0.1702,
      "step": 1959
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5484238897303066,
      "learning_rate": 3.650987598282116e-05,
      "loss": 0.1499,
      "step": 1960
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5255099590022039,
      "learning_rate": 3.650585965302614e-05,
      "loss": 0.1514,
      "step": 1961
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5224774794767366,
      "learning_rate": 3.6501841234801886e-05,
      "loss": 0.1648,
      "step": 1962
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.558116308659357,
      "learning_rate": 3.64978207286568e-05,
      "loss": 0.168,
      "step": 1963
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5608751884002743,
      "learning_rate": 3.649379813509961e-05,
      "loss": 0.1614,
      "step": 1964
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5931777455118002,
      "learning_rate": 3.648977345463926e-05,
      "loss": 0.1398,
      "step": 1965
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5711772320182702,
      "learning_rate": 3.648574668778499e-05,
      "loss": 0.1986,
      "step": 1966
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5072716987747973,
      "learning_rate": 3.6481717835046286e-05,
      "loss": 0.1617,
      "step": 1967
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5965541576443644,
      "learning_rate": 3.647768689693291e-05,
      "loss": 0.1623,
      "step": 1968
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6849475207726617,
      "learning_rate": 3.647365387395488e-05,
      "loss": 0.2328,
      "step": 1969
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5811336431633901,
      "learning_rate": 3.646961876662248e-05,
      "loss": 0.1932,
      "step": 1970
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4679220994677543,
      "learning_rate": 3.646558157544626e-05,
      "loss": 0.1411,
      "step": 1971
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4810113711569421,
      "learning_rate": 3.6461542300937035e-05,
      "loss": 0.1718,
      "step": 1972
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5762250972671092,
      "learning_rate": 3.645750094360588e-05,
      "loss": 0.221,
      "step": 1973
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5818175877356747,
      "learning_rate": 3.645345750396412e-05,
      "loss": 0.1829,
      "step": 1974
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5473105739565279,
      "learning_rate": 3.6449411982523376e-05,
      "loss": 0.1891,
      "step": 1975
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6094371323821574,
      "learning_rate": 3.64453643797955e-05,
      "loss": 0.1784,
      "step": 1976
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4716501957300015,
      "learning_rate": 3.644131469629264e-05,
      "loss": 0.1588,
      "step": 1977
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4981673898932009,
      "learning_rate": 3.643726293252717e-05,
      "loss": 0.1199,
      "step": 1978
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4690944359852931,
      "learning_rate": 3.6433209089011745e-05,
      "loss": 0.1416,
      "step": 1979
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5811417201921631,
      "learning_rate": 3.642915316625929e-05,
      "loss": 0.1804,
      "step": 1980
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.48637858669043826,
      "learning_rate": 3.642509516478299e-05,
      "loss": 0.1621,
      "step": 1981
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6521297720642567,
      "learning_rate": 3.642103508509629e-05,
      "loss": 0.2048,
      "step": 1982
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5445678545389675,
      "learning_rate": 3.641697292771289e-05,
      "loss": 0.1785,
      "step": 1983
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4149623306252465,
      "learning_rate": 3.641290869314676e-05,
      "loss": 0.1299,
      "step": 1984
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7210705655256837,
      "learning_rate": 3.6408842381912144e-05,
      "loss": 0.226,
      "step": 1985
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.49010246903560933,
      "learning_rate": 3.640477399452354e-05,
      "loss": 0.1528,
      "step": 1986
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46981173117928565,
      "learning_rate": 3.640070353149569e-05,
      "loss": 0.1667,
      "step": 1987
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5661934288346968,
      "learning_rate": 3.639663099334363e-05,
      "loss": 0.1354,
      "step": 1988
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43644088886272214,
      "learning_rate": 3.639255638058264e-05,
      "loss": 0.1206,
      "step": 1989
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5508694409045998,
      "learning_rate": 3.6388479693728266e-05,
      "loss": 0.1247,
      "step": 1990
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5019408346964781,
      "learning_rate": 3.638440093329632e-05,
      "loss": 0.1689,
      "step": 1991
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5842317548407773,
      "learning_rate": 3.638032009980286e-05,
      "loss": 0.1647,
      "step": 1992
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7121271094595126,
      "learning_rate": 3.637623719376424e-05,
      "loss": 0.2153,
      "step": 1993
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4787442031190676,
      "learning_rate": 3.637215221569705e-05,
      "loss": 0.1456,
      "step": 1994
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5565819424933913,
      "learning_rate": 3.6368065166118136e-05,
      "loss": 0.1543,
      "step": 1995
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.500342609326308,
      "learning_rate": 3.636397604554463e-05,
      "loss": 0.1457,
      "step": 1996
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5064763316737215,
      "learning_rate": 3.635988485449391e-05,
      "loss": 0.1287,
      "step": 1997
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4277062951194384,
      "learning_rate": 3.635579159348362e-05,
      "loss": 0.1055,
      "step": 1998
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5991639710296278,
      "learning_rate": 3.635169626303168e-05,
      "loss": 0.1786,
      "step": 1999
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6257932873733754,
      "learning_rate": 3.634759886365623e-05,
      "loss": 0.1922,
      "step": 2000
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5257618666727696,
      "learning_rate": 3.634349939587573e-05,
      "loss": 0.1747,
      "step": 2001
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4432636107147846,
      "learning_rate": 3.633939786020884e-05,
      "loss": 0.1468,
      "step": 2002
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4857121898927471,
      "learning_rate": 3.633529425717454e-05,
      "loss": 0.1314,
      "step": 2003
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5365739407705447,
      "learning_rate": 3.633118858729203e-05,
      "loss": 0.1808,
      "step": 2004
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5266604765345163,
      "learning_rate": 3.63270808510808e-05,
      "loss": 0.159,
      "step": 2005
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5027939696658185,
      "learning_rate": 3.632297104906057e-05,
      "loss": 0.1739,
      "step": 2006
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4696130912911679,
      "learning_rate": 3.6318859181751346e-05,
      "loss": 0.1086,
      "step": 2007
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46110300551832906,
      "learning_rate": 3.63147452496734e-05,
      "loss": 0.1343,
      "step": 2008
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5443645895434426,
      "learning_rate": 3.631062925334723e-05,
      "loss": 0.1487,
      "step": 2009
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4930222864597726,
      "learning_rate": 3.6306511193293636e-05,
      "loss": 0.1421,
      "step": 2010
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4832259978398586,
      "learning_rate": 3.630239107003366e-05,
      "loss": 0.1283,
      "step": 2011
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6334396833505439,
      "learning_rate": 3.629826888408861e-05,
      "loss": 0.2047,
      "step": 2012
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7640047640527114,
      "learning_rate": 3.629414463598005e-05,
      "loss": 0.3396,
      "step": 2013
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5815493745875413,
      "learning_rate": 3.629001832622979e-05,
      "loss": 0.1834,
      "step": 2014
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5783466738922071,
      "learning_rate": 3.628588995535993e-05,
      "loss": 0.2054,
      "step": 2015
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4736537024226936,
      "learning_rate": 3.628175952389283e-05,
      "loss": 0.1625,
      "step": 2016
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5274918242358799,
      "learning_rate": 3.627762703235108e-05,
      "loss": 0.1682,
      "step": 2017
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46141825674875125,
      "learning_rate": 3.627349248125757e-05,
      "loss": 0.1439,
      "step": 2018
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4703071163034107,
      "learning_rate": 3.626935587113541e-05,
      "loss": 0.1588,
      "step": 2019
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3911449976716171,
      "learning_rate": 3.6265217202508006e-05,
      "loss": 0.1258,
      "step": 2020
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5918724590919682,
      "learning_rate": 3.6261076475899e-05,
      "loss": 0.1756,
      "step": 2021
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5198404639610035,
      "learning_rate": 3.625693369183231e-05,
      "loss": 0.1276,
      "step": 2022
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5160302390397821,
      "learning_rate": 3.62527888508321e-05,
      "loss": 0.1406,
      "step": 2023
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5973982409889064,
      "learning_rate": 3.624864195342281e-05,
      "loss": 0.1813,
      "step": 2024
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5722754676154685,
      "learning_rate": 3.624449300012914e-05,
      "loss": 0.163,
      "step": 2025
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6621744129567309,
      "learning_rate": 3.624034199147602e-05,
      "loss": 0.2156,
      "step": 2026
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5930819015199781,
      "learning_rate": 3.623618892798868e-05,
      "loss": 0.1678,
      "step": 2027
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5742386535516691,
      "learning_rate": 3.623203381019259e-05,
      "loss": 0.1775,
      "step": 2028
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4256503227588008,
      "learning_rate": 3.6227876638613484e-05,
      "loss": 0.1227,
      "step": 2029
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.497967104039795,
      "learning_rate": 3.6223717413777346e-05,
      "loss": 0.134,
      "step": 2030
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5522639021291449,
      "learning_rate": 3.6219556136210434e-05,
      "loss": 0.1778,
      "step": 2031
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5163330360949018,
      "learning_rate": 3.621539280643926e-05,
      "loss": 0.1758,
      "step": 2032
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4713782063741588,
      "learning_rate": 3.62112274249906e-05,
      "loss": 0.1292,
      "step": 2033
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5164288987633738,
      "learning_rate": 3.620705999239148e-05,
      "loss": 0.1958,
      "step": 2034
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4637345937652982,
      "learning_rate": 3.6202890509169195e-05,
      "loss": 0.1482,
      "step": 2035
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44388691777822264,
      "learning_rate": 3.619871897585129e-05,
      "loss": 0.1304,
      "step": 2036
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3920281956760082,
      "learning_rate": 3.619454539296557e-05,
      "loss": 0.1172,
      "step": 2037
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5284547476094049,
      "learning_rate": 3.6190369761040116e-05,
      "loss": 0.1413,
      "step": 2038
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5252998018290177,
      "learning_rate": 3.6186192080603256e-05,
      "loss": 0.1383,
      "step": 2039
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5471299530826533,
      "learning_rate": 3.618201235218356e-05,
      "loss": 0.1619,
      "step": 2040
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4663222066280349,
      "learning_rate": 3.61778305763099e-05,
      "loss": 0.1459,
      "step": 2041
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6423682243938496,
      "learning_rate": 3.617364675351136e-05,
      "loss": 0.2014,
      "step": 2042
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4189043536315817,
      "learning_rate": 3.616946088431732e-05,
      "loss": 0.1168,
      "step": 2043
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8351530868998507,
      "learning_rate": 3.61652729692574e-05,
      "loss": 0.1388,
      "step": 2044
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5602007870811945,
      "learning_rate": 3.6161083008861464e-05,
      "loss": 0.1645,
      "step": 2045
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4991469556974522,
      "learning_rate": 3.615689100365968e-05,
      "loss": 0.1936,
      "step": 2046
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6335008433154965,
      "learning_rate": 3.615269695418243e-05,
      "loss": 0.1996,
      "step": 2047
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4085568807097909,
      "learning_rate": 3.6148500860960386e-05,
      "loss": 0.1182,
      "step": 2048
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4847407344603472,
      "learning_rate": 3.6144302724524456e-05,
      "loss": 0.1196,
      "step": 2049
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4259377638101635,
      "learning_rate": 3.614010254540581e-05,
      "loss": 0.1464,
      "step": 2050
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5063057774557885,
      "learning_rate": 3.61359003241359e-05,
      "loss": 0.1807,
      "step": 2051
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49745151614196687,
      "learning_rate": 3.6131696061246405e-05,
      "loss": 0.1466,
      "step": 2052
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5721184603146693,
      "learning_rate": 3.612748975726928e-05,
      "loss": 0.194,
      "step": 2053
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5097667357918633,
      "learning_rate": 3.612328141273673e-05,
      "loss": 0.1703,
      "step": 2054
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.519969611838875,
      "learning_rate": 3.6119071028181225e-05,
      "loss": 0.179,
      "step": 2055
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.553502442196883,
      "learning_rate": 3.6114858604135496e-05,
      "loss": 0.1805,
      "step": 2056
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5507481612418532,
      "learning_rate": 3.6110644141132514e-05,
      "loss": 0.1616,
      "step": 2057
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5562239844421336,
      "learning_rate": 3.610642763970553e-05,
      "loss": 0.2282,
      "step": 2058
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4366349796756776,
      "learning_rate": 3.610220910038805e-05,
      "loss": 0.1399,
      "step": 2059
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49975418962685153,
      "learning_rate": 3.6097988523713816e-05,
      "loss": 0.13,
      "step": 2060
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.573067347623059,
      "learning_rate": 3.609376591021684e-05,
      "loss": 0.2138,
      "step": 2061
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5172269385475974,
      "learning_rate": 3.608954126043141e-05,
      "loss": 0.1805,
      "step": 2062
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5368863284211457,
      "learning_rate": 3.608531457489205e-05,
      "loss": 0.149,
      "step": 2063
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4474400675721666,
      "learning_rate": 3.608108585413356e-05,
      "loss": 0.1421,
      "step": 2064
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5168813081423282,
      "learning_rate": 3.607685509869096e-05,
      "loss": 0.1245,
      "step": 2065
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.47662445353554955,
      "learning_rate": 3.6072622309099566e-05,
      "loss": 0.1255,
      "step": 2066
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5527029668828085,
      "learning_rate": 3.606838748589495e-05,
      "loss": 0.1491,
      "step": 2067
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6733912060396514,
      "learning_rate": 3.60641506296129e-05,
      "loss": 0.1714,
      "step": 2068
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.590959743724629,
      "learning_rate": 3.605991174078952e-05,
      "loss": 0.1759,
      "step": 2069
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.48218091746378144,
      "learning_rate": 3.605567081996113e-05,
      "loss": 0.1377,
      "step": 2070
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5524899107240545,
      "learning_rate": 3.605142786766432e-05,
      "loss": 0.1638,
      "step": 2071
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5751770482944267,
      "learning_rate": 3.604718288443593e-05,
      "loss": 0.1772,
      "step": 2072
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6105105993973792,
      "learning_rate": 3.6042935870813075e-05,
      "loss": 0.2161,
      "step": 2073
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4335465964031393,
      "learning_rate": 3.60386868273331e-05,
      "loss": 0.1391,
      "step": 2074
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5847080811426361,
      "learning_rate": 3.603443575453364e-05,
      "loss": 0.216,
      "step": 2075
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5602100170938202,
      "learning_rate": 3.603018265295255e-05,
      "loss": 0.1591,
      "step": 2076
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5564367623708293,
      "learning_rate": 3.602592752312798e-05,
      "loss": 0.2086,
      "step": 2077
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49248547087214345,
      "learning_rate": 3.60216703655983e-05,
      "loss": 0.1588,
      "step": 2078
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5153313870088918,
      "learning_rate": 3.6017411180902157e-05,
      "loss": 0.1856,
      "step": 2079
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4188910673720334,
      "learning_rate": 3.601314996957845e-05,
      "loss": 0.12,
      "step": 2080
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5524775579045905,
      "learning_rate": 3.600888673216635e-05,
      "loss": 0.1693,
      "step": 2081
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5003919853054976,
      "learning_rate": 3.600462146920525e-05,
      "loss": 0.1415,
      "step": 2082
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5012279273997605,
      "learning_rate": 3.600035418123483e-05,
      "loss": 0.1361,
      "step": 2083
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5940832909202195,
      "learning_rate": 3.5996084868795015e-05,
      "loss": 0.1618,
      "step": 2084
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4627231504182507,
      "learning_rate": 3.5991813532425977e-05,
      "loss": 0.1374,
      "step": 2085
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6019412990747641,
      "learning_rate": 3.5987540172668164e-05,
      "loss": 0.1909,
      "step": 2086
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5094973855929547,
      "learning_rate": 3.598326479006226e-05,
      "loss": 0.1566,
      "step": 2087
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5033207826386441,
      "learning_rate": 3.597898738514923e-05,
      "loss": 0.1766,
      "step": 2088
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5586757466349173,
      "learning_rate": 3.597470795847026e-05,
      "loss": 0.1762,
      "step": 2089
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5388756027200612,
      "learning_rate": 3.5970426510566824e-05,
      "loss": 0.1844,
      "step": 2090
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4788717073861573,
      "learning_rate": 3.596614304198063e-05,
      "loss": 0.1141,
      "step": 2091
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49148072859430597,
      "learning_rate": 3.5961857553253665e-05,
      "loss": 0.1644,
      "step": 2092
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.501138463917304,
      "learning_rate": 3.595757004492814e-05,
      "loss": 0.1276,
      "step": 2093
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49792661523922327,
      "learning_rate": 3.595328051754654e-05,
      "loss": 0.1315,
      "step": 2094
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5218874627184253,
      "learning_rate": 3.5948988971651624e-05,
      "loss": 0.1289,
      "step": 2095
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.48924258393060116,
      "learning_rate": 3.594469540778637e-05,
      "loss": 0.1287,
      "step": 2096
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5518858634423857,
      "learning_rate": 3.594039982649402e-05,
      "loss": 0.1481,
      "step": 2097
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.584088344951233,
      "learning_rate": 3.593610222831809e-05,
      "loss": 0.1501,
      "step": 2098
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5855221356948225,
      "learning_rate": 3.593180261380235e-05,
      "loss": 0.1591,
      "step": 2099
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5031045850320036,
      "learning_rate": 3.59275009834908e-05,
      "loss": 0.1132,
      "step": 2100
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6306899938748599,
      "learning_rate": 3.592319733792772e-05,
      "loss": 0.1747,
      "step": 2101
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6248875278362477,
      "learning_rate": 3.591889167765762e-05,
      "loss": 0.1596,
      "step": 2102
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6393745127030023,
      "learning_rate": 3.59145840032253e-05,
      "loss": 0.1453,
      "step": 2103
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49098458203350615,
      "learning_rate": 3.591027431517577e-05,
      "loss": 0.1222,
      "step": 2104
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6470013997152715,
      "learning_rate": 3.590596261405435e-05,
      "loss": 0.2166,
      "step": 2105
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41923729740705373,
      "learning_rate": 3.590164890040657e-05,
      "loss": 0.1223,
      "step": 2106
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.570456578070775,
      "learning_rate": 3.5897333174778225e-05,
      "loss": 0.2212,
      "step": 2107
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5025193401429565,
      "learning_rate": 3.589301543771537e-05,
      "loss": 0.1512,
      "step": 2108
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4099237782292562,
      "learning_rate": 3.588869568976433e-05,
      "loss": 0.1581,
      "step": 2109
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5462918578462681,
      "learning_rate": 3.588437393147164e-05,
      "loss": 0.1854,
      "step": 2110
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5170585471562059,
      "learning_rate": 3.5880050163384137e-05,
      "loss": 0.134,
      "step": 2111
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5689140313843987,
      "learning_rate": 3.587572438604889e-05,
      "loss": 0.193,
      "step": 2112
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.618787165209569,
      "learning_rate": 3.587139660001322e-05,
      "loss": 0.1339,
      "step": 2113
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6374166209727526,
      "learning_rate": 3.586706680582471e-05,
      "loss": 0.2381,
      "step": 2114
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5573617914828713,
      "learning_rate": 3.5862735004031186e-05,
      "loss": 0.1656,
      "step": 2115
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4590089831958744,
      "learning_rate": 3.585840119518075e-05,
      "loss": 0.1065,
      "step": 2116
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5377403575736783,
      "learning_rate": 3.585406537982173e-05,
      "loss": 0.1729,
      "step": 2117
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4928683612491513,
      "learning_rate": 3.584972755850273e-05,
      "loss": 0.1023,
      "step": 2118
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5105135170532267,
      "learning_rate": 3.58453877317726e-05,
      "loss": 0.1256,
      "step": 2119
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5889843117303659,
      "learning_rate": 3.584104590018044e-05,
      "loss": 0.1798,
      "step": 2120
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5361342021875489,
      "learning_rate": 3.583670206427559e-05,
      "loss": 0.1746,
      "step": 2121
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49495203698553464,
      "learning_rate": 3.58323562246077e-05,
      "loss": 0.1449,
      "step": 2122
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4712080245966396,
      "learning_rate": 3.5828008381726604e-05,
      "loss": 0.1591,
      "step": 2123
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5619003755654761,
      "learning_rate": 3.5823658536182426e-05,
      "loss": 0.1988,
      "step": 2124
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.509945541727996,
      "learning_rate": 3.581930668852554e-05,
      "loss": 0.1764,
      "step": 2125
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4888556801770972,
      "learning_rate": 3.5814952839306574e-05,
      "loss": 0.1564,
      "step": 2126
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5436589636176077,
      "learning_rate": 3.581059698907639e-05,
      "loss": 0.2171,
      "step": 2127
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46314736179723454,
      "learning_rate": 3.580623913838613e-05,
      "loss": 0.1465,
      "step": 2128
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4723642724235566,
      "learning_rate": 3.5801879287787185e-05,
      "loss": 0.1449,
      "step": 2129
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5171827425550637,
      "learning_rate": 3.579751743783118e-05,
      "loss": 0.1469,
      "step": 2130
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3948596682734425,
      "learning_rate": 3.5793153589070005e-05,
      "loss": 0.1024,
      "step": 2131
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5111791080299523,
      "learning_rate": 3.578878774205581e-05,
      "loss": 0.1504,
      "step": 2132
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4992446856858184,
      "learning_rate": 3.578441989734097e-05,
      "loss": 0.1287,
      "step": 2133
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.568864001911079,
      "learning_rate": 3.578005005547817e-05,
      "loss": 0.1729,
      "step": 2134
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4987006141715803,
      "learning_rate": 3.577567821702028e-05,
      "loss": 0.1413,
      "step": 2135
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5919659313293283,
      "learning_rate": 3.577130438252046e-05,
      "loss": 0.173,
      "step": 2136
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5258159971729612,
      "learning_rate": 3.576692855253213e-05,
      "loss": 0.133,
      "step": 2137
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5210813758215879,
      "learning_rate": 3.576255072760893e-05,
      "loss": 0.1312,
      "step": 2138
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4247997415188903,
      "learning_rate": 3.575817090830479e-05,
      "loss": 0.1263,
      "step": 2139
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4656267063698448,
      "learning_rate": 3.575378909517385e-05,
      "loss": 0.143,
      "step": 2140
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5243593610084191,
      "learning_rate": 3.5749405288770554e-05,
      "loss": 0.1756,
      "step": 2141
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5194542332977483,
      "learning_rate": 3.574501948964954e-05,
      "loss": 0.1381,
      "step": 2142
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5325908009477068,
      "learning_rate": 3.574063169836575e-05,
      "loss": 0.1551,
      "step": 2143
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6515360188763195,
      "learning_rate": 3.5736241915474345e-05,
      "loss": 0.1705,
      "step": 2144
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4534291857766253,
      "learning_rate": 3.5731850141530755e-05,
      "loss": 0.1289,
      "step": 2145
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48019513038253836,
      "learning_rate": 3.572745637709065e-05,
      "loss": 0.1376,
      "step": 2146
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5842375445754202,
      "learning_rate": 3.572306062270997e-05,
      "loss": 0.1842,
      "step": 2147
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5098591276626897,
      "learning_rate": 3.5718662878944876e-05,
      "loss": 0.1588,
      "step": 2148
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48386490019139633,
      "learning_rate": 3.571426314635181e-05,
      "loss": 0.1302,
      "step": 2149
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5765781586544345,
      "learning_rate": 3.570986142548746e-05,
      "loss": 0.1264,
      "step": 2150
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5298212979553626,
      "learning_rate": 3.570545771690875e-05,
      "loss": 0.163,
      "step": 2151
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48545262226760266,
      "learning_rate": 3.5701052021172874e-05,
      "loss": 0.1347,
      "step": 2152
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.46175935182669003,
      "learning_rate": 3.569664433883726e-05,
      "loss": 0.1898,
      "step": 2153
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43663625944654316,
      "learning_rate": 3.5692234670459615e-05,
      "loss": 0.133,
      "step": 2154
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6176667103459438,
      "learning_rate": 3.568782301659786e-05,
      "loss": 0.1804,
      "step": 2155
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6520791802157258,
      "learning_rate": 3.5683409377810185e-05,
      "loss": 0.2391,
      "step": 2156
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39095976789741776,
      "learning_rate": 3.567899375465505e-05,
      "loss": 0.1262,
      "step": 2157
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.46209647063088194,
      "learning_rate": 3.567457614769113e-05,
      "loss": 0.1464,
      "step": 2158
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5520535646965972,
      "learning_rate": 3.567015655747739e-05,
      "loss": 0.1786,
      "step": 2159
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4781532005021844,
      "learning_rate": 3.566573498457301e-05,
      "loss": 0.1296,
      "step": 2160
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5174091567132771,
      "learning_rate": 3.5661311429537435e-05,
      "loss": 0.1858,
      "step": 2161
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.45674934822395147,
      "learning_rate": 3.5656885892930376e-05,
      "loss": 0.1317,
      "step": 2162
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5624280989575822,
      "learning_rate": 3.565245837531177e-05,
      "loss": 0.165,
      "step": 2163
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4039046923904342,
      "learning_rate": 3.564802887724181e-05,
      "loss": 0.1087,
      "step": 2164
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5439162734548321,
      "learning_rate": 3.564359739928097e-05,
      "loss": 0.1807,
      "step": 2165
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5429791236819561,
      "learning_rate": 3.563916394198991e-05,
      "loss": 0.1689,
      "step": 2166
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5109704056622522,
      "learning_rate": 3.563472850592962e-05,
      "loss": 0.1627,
      "step": 2167
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5563541143470068,
      "learning_rate": 3.5630291091661276e-05,
      "loss": 0.1573,
      "step": 2168
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4252159145202093,
      "learning_rate": 3.562585169974633e-05,
      "loss": 0.1332,
      "step": 2169
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43047371807796997,
      "learning_rate": 3.562141033074649e-05,
      "loss": 0.1112,
      "step": 2170
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5492191754226882,
      "learning_rate": 3.561696698522371e-05,
      "loss": 0.1705,
      "step": 2171
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4882957761511142,
      "learning_rate": 3.5612521663740183e-05,
      "loss": 0.1095,
      "step": 2172
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6301930218142273,
      "learning_rate": 3.560807436685837e-05,
      "loss": 0.1878,
      "step": 2173
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.49412684364853066,
      "learning_rate": 3.560362509514096e-05,
      "loss": 0.1493,
      "step": 2174
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5284920235359436,
      "learning_rate": 3.5599173849150905e-05,
      "loss": 0.1906,
      "step": 2175
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.44591722085417274,
      "learning_rate": 3.5594720629451414e-05,
      "loss": 0.1133,
      "step": 2176
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48880495596603685,
      "learning_rate": 3.559026543660593e-05,
      "loss": 0.1234,
      "step": 2177
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4255845901867083,
      "learning_rate": 3.558580827117817e-05,
      "loss": 0.1081,
      "step": 2178
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5065768246563851,
      "learning_rate": 3.558134913373206e-05,
      "loss": 0.1707,
      "step": 2179
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5155744238815785,
      "learning_rate": 3.557688802483181e-05,
      "loss": 0.1438,
      "step": 2180
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5612169573770273,
      "learning_rate": 3.557242494504187e-05,
      "loss": 0.1619,
      "step": 2181
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5686133168064943,
      "learning_rate": 3.556795989492694e-05,
      "loss": 0.1461,
      "step": 2182
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0100054871060726,
      "learning_rate": 3.556349287505196e-05,
      "loss": 0.114,
      "step": 2183
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6050875540504129,
      "learning_rate": 3.555902388598213e-05,
      "loss": 0.1534,
      "step": 2184
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6091933346436944,
      "learning_rate": 3.55545529282829e-05,
      "loss": 0.1575,
      "step": 2185
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4659817333959437,
      "learning_rate": 3.555008000251995e-05,
      "loss": 0.151,
      "step": 2186
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5740572124771376,
      "learning_rate": 3.554560510925924e-05,
      "loss": 0.1335,
      "step": 2187
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.509735778870272,
      "learning_rate": 3.554112824906696e-05,
      "loss": 0.133,
      "step": 2188
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5790508177269011,
      "learning_rate": 3.553664942250955e-05,
      "loss": 0.1705,
      "step": 2189
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5429820875546844,
      "learning_rate": 3.55321686301537e-05,
      "loss": 0.1609,
      "step": 2190
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5247871307968536,
      "learning_rate": 3.552768587256635e-05,
      "loss": 0.1532,
      "step": 2191
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6465658269089535,
      "learning_rate": 3.552320115031468e-05,
      "loss": 0.177,
      "step": 2192
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5163116506937354,
      "learning_rate": 3.551871446396613e-05,
      "loss": 0.1388,
      "step": 2193
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5481506727898416,
      "learning_rate": 3.55142258140884e-05,
      "loss": 0.1835,
      "step": 2194
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5506908261358118,
      "learning_rate": 3.550973520124941e-05,
      "loss": 0.1609,
      "step": 2195
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5165653324222624,
      "learning_rate": 3.5505242626017326e-05,
      "loss": 0.1719,
      "step": 2196
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.520231792846266,
      "learning_rate": 3.550074808896061e-05,
      "loss": 0.1632,
      "step": 2197
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5113968746865574,
      "learning_rate": 3.549625159064792e-05,
      "loss": 0.1425,
      "step": 2198
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.49042654640349476,
      "learning_rate": 3.549175313164819e-05,
      "loss": 0.1502,
      "step": 2199
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5543626451889639,
      "learning_rate": 3.5487252712530583e-05,
      "loss": 0.1438,
      "step": 2200
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.44781759926021647,
      "learning_rate": 3.548275033386453e-05,
      "loss": 0.1118,
      "step": 2201
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6032080005201187,
      "learning_rate": 3.547824599621971e-05,
      "loss": 0.184,
      "step": 2202
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4055394046795746,
      "learning_rate": 3.5473739700166025e-05,
      "loss": 0.109,
      "step": 2203
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.638074823287977,
      "learning_rate": 3.546923144627366e-05,
      "loss": 0.2261,
      "step": 2204
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5346847183181935,
      "learning_rate": 3.5464721235113006e-05,
      "loss": 0.1769,
      "step": 2205
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5362788211751282,
      "learning_rate": 3.546020906725474e-05,
      "loss": 0.1332,
      "step": 2206
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.571258778869819,
      "learning_rate": 3.545569494326977e-05,
      "loss": 0.1431,
      "step": 2207
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4222020188665076,
      "learning_rate": 3.5451178863729244e-05,
      "loss": 0.094,
      "step": 2208
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.44866693854335543,
      "learning_rate": 3.5446660829204575e-05,
      "loss": 0.1176,
      "step": 2209
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.45114148850933555,
      "learning_rate": 3.5442140840267404e-05,
      "loss": 0.1344,
      "step": 2210
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.479900861873437,
      "learning_rate": 3.5437618897489646e-05,
      "loss": 0.1524,
      "step": 2211
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6266895108379495,
      "learning_rate": 3.543309500144343e-05,
      "loss": 0.1924,
      "step": 2212
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5907835893804102,
      "learning_rate": 3.542856915270116e-05,
      "loss": 0.1727,
      "step": 2213
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42050965446497524,
      "learning_rate": 3.542404135183547e-05,
      "loss": 0.1441,
      "step": 2214
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6524131646600938,
      "learning_rate": 3.541951159941924e-05,
      "loss": 0.2036,
      "step": 2215
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.443467282510872,
      "learning_rate": 3.541497989602562e-05,
      "loss": 0.1049,
      "step": 2216
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.38469451402084853,
      "learning_rate": 3.5410446242227985e-05,
      "loss": 0.1078,
      "step": 2217
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4470458844378462,
      "learning_rate": 3.540591063859996e-05,
      "loss": 0.1542,
      "step": 2218
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.531859861263796,
      "learning_rate": 3.5401373085715413e-05,
      "loss": 0.1551,
      "step": 2219
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4349888532679271,
      "learning_rate": 3.539683358414848e-05,
      "loss": 0.1461,
      "step": 2220
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6628876110001641,
      "learning_rate": 3.5392292134473516e-05,
      "loss": 0.2032,
      "step": 2221
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5907690104760583,
      "learning_rate": 3.538774873726514e-05,
      "loss": 0.1643,
      "step": 2222
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4789184120712393,
      "learning_rate": 3.538320339309821e-05,
      "loss": 0.1368,
      "step": 2223
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.529772808440853,
      "learning_rate": 3.537865610254784e-05,
      "loss": 0.1474,
      "step": 2224
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4280752278375512,
      "learning_rate": 3.537410686618937e-05,
      "loss": 0.1128,
      "step": 2225
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4630702511219398,
      "learning_rate": 3.536955568459841e-05,
      "loss": 0.1283,
      "step": 2226
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.41466216083947766,
      "learning_rate": 3.53650025583508e-05,
      "loss": 0.1317,
      "step": 2227
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4446960666606069,
      "learning_rate": 3.536044748802263e-05,
      "loss": 0.1492,
      "step": 2228
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4525741319893434,
      "learning_rate": 3.5355890474190244e-05,
      "loss": 0.1265,
      "step": 2229
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48344273219763323,
      "learning_rate": 3.535133151743022e-05,
      "loss": 0.1292,
      "step": 2230
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6002288004616197,
      "learning_rate": 3.5346770618319384e-05,
      "loss": 0.1572,
      "step": 2231
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4727657601096312,
      "learning_rate": 3.534220777743482e-05,
      "loss": 0.1166,
      "step": 2232
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6247609158035904,
      "learning_rate": 3.533764299535384e-05,
      "loss": 0.1793,
      "step": 2233
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4440872334753855,
      "learning_rate": 3.5333076272654014e-05,
      "loss": 0.1142,
      "step": 2234
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4140559501849665,
      "learning_rate": 3.532850760991315e-05,
      "loss": 0.1235,
      "step": 2235
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.485755675419642,
      "learning_rate": 3.532393700770932e-05,
      "loss": 0.1403,
      "step": 2236
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5490501470592869,
      "learning_rate": 3.53193644666208e-05,
      "loss": 0.1485,
      "step": 2237
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5105878841213323,
      "learning_rate": 3.5314789987226156e-05,
      "loss": 0.1411,
      "step": 2238
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.448749795026387,
      "learning_rate": 3.531021357010419e-05,
      "loss": 0.1119,
      "step": 2239
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5074135019787672,
      "learning_rate": 3.5305635215833914e-05,
      "loss": 0.1473,
      "step": 2240
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6197653425594231,
      "learning_rate": 3.5301054924994626e-05,
      "loss": 0.1906,
      "step": 2241
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44412315220210885,
      "learning_rate": 3.5296472698165856e-05,
      "loss": 0.115,
      "step": 2242
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6136801327563189,
      "learning_rate": 3.5291888535927375e-05,
      "loss": 0.1504,
      "step": 2243
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4395960935515403,
      "learning_rate": 3.5287302438859204e-05,
      "loss": 0.1114,
      "step": 2244
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5274271016122571,
      "learning_rate": 3.5282714407541587e-05,
      "loss": 0.1554,
      "step": 2245
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3700567714529318,
      "learning_rate": 3.5278124442555066e-05,
      "loss": 0.1023,
      "step": 2246
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.46423611074653853,
      "learning_rate": 3.527353254448036e-05,
      "loss": 0.1501,
      "step": 2247
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4532808167469646,
      "learning_rate": 3.526893871389849e-05,
      "loss": 0.1125,
      "step": 2248
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4940976663243304,
      "learning_rate": 3.526434295139069e-05,
      "loss": 0.1261,
      "step": 2249
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5662019032142154,
      "learning_rate": 3.5259745257538443e-05,
      "loss": 0.1694,
      "step": 2250
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5210922710163226,
      "learning_rate": 3.525514563292347e-05,
      "loss": 0.156,
      "step": 2251
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6472290222095444,
      "learning_rate": 3.525054407812777e-05,
      "loss": 0.2152,
      "step": 2252
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4818578967907291,
      "learning_rate": 3.5245940593733535e-05,
      "loss": 0.1195,
      "step": 2253
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5370053277239778,
      "learning_rate": 3.524133518032325e-05,
      "loss": 0.1369,
      "step": 2254
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5434513554443757,
      "learning_rate": 3.52367278384796e-05,
      "loss": 0.176,
      "step": 2255
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.942033455878222,
      "learning_rate": 3.5232118568785565e-05,
      "loss": 0.3057,
      "step": 2256
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4373734495512156,
      "learning_rate": 3.522750737182431e-05,
      "loss": 0.1508,
      "step": 2257
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4600040715234357,
      "learning_rate": 3.52228942481793e-05,
      "loss": 0.1396,
      "step": 2258
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4726547682557643,
      "learning_rate": 3.5218279198434196e-05,
      "loss": 0.1186,
      "step": 2259
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6472880044773579,
      "learning_rate": 3.5213662223172935e-05,
      "loss": 0.1217,
      "step": 2260
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44111742005466137,
      "learning_rate": 3.5209043322979686e-05,
      "loss": 0.1383,
      "step": 2261
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.45533751597701205,
      "learning_rate": 3.520442249843887e-05,
      "loss": 0.119,
      "step": 2262
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4791494725777471,
      "learning_rate": 3.5199799750135114e-05,
      "loss": 0.1619,
      "step": 2263
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4201550922824717,
      "learning_rate": 3.5195175078653355e-05,
      "loss": 0.1379,
      "step": 2264
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.461100500995229,
      "learning_rate": 3.519054848457872e-05,
      "loss": 0.1209,
      "step": 2265
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.441000676134028,
      "learning_rate": 3.51859199684966e-05,
      "loss": 0.111,
      "step": 2266
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5076589792447682,
      "learning_rate": 3.518128953099261e-05,
      "loss": 0.1313,
      "step": 2267
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6128995643095091,
      "learning_rate": 3.517665717265265e-05,
      "loss": 0.1765,
      "step": 2268
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8309773679238123,
      "learning_rate": 3.517202289406281e-05,
      "loss": 0.1177,
      "step": 2269
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5300565384901884,
      "learning_rate": 3.516738669580947e-05,
      "loss": 0.1336,
      "step": 2270
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5162780845059695,
      "learning_rate": 3.516274857847922e-05,
      "loss": 0.1356,
      "step": 2271
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5635613389262888,
      "learning_rate": 3.5158108542658915e-05,
      "loss": 0.164,
      "step": 2272
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5276599077589897,
      "learning_rate": 3.515346658893562e-05,
      "loss": 0.1456,
      "step": 2273
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6588667062965825,
      "learning_rate": 3.5148822717896694e-05,
      "loss": 0.1691,
      "step": 2274
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42917307490900736,
      "learning_rate": 3.5144176930129694e-05,
      "loss": 0.1181,
      "step": 2275
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4395522620382328,
      "learning_rate": 3.513952922622243e-05,
      "loss": 0.1377,
      "step": 2276
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40315644119432026,
      "learning_rate": 3.513487960676298e-05,
      "loss": 0.1208,
      "step": 2277
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5077651955506216,
      "learning_rate": 3.513022807233964e-05,
      "loss": 0.137,
      "step": 2278
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4915112891079478,
      "learning_rate": 3.512557462354093e-05,
      "loss": 0.1273,
      "step": 2279
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.48581288325202493,
      "learning_rate": 3.5120919260955655e-05,
      "loss": 0.1621,
      "step": 2280
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5600576093111308,
      "learning_rate": 3.5116261985172835e-05,
      "loss": 0.1706,
      "step": 2281
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4828447041064188,
      "learning_rate": 3.511160279678174e-05,
      "loss": 0.1396,
      "step": 2282
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4689202952990553,
      "learning_rate": 3.510694169637188e-05,
      "loss": 0.1094,
      "step": 2283
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.47778822438637963,
      "learning_rate": 3.510227868453302e-05,
      "loss": 0.1322,
      "step": 2284
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4444224849713238,
      "learning_rate": 3.5097613761855134e-05,
      "loss": 0.1357,
      "step": 2285
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5777287542909266,
      "learning_rate": 3.509294692892847e-05,
      "loss": 0.1226,
      "step": 2286
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.47072056781534544,
      "learning_rate": 3.508827818634351e-05,
      "loss": 0.1214,
      "step": 2287
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5920025559350847,
      "learning_rate": 3.508360753469097e-05,
      "loss": 0.188,
      "step": 2288
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5112946769667809,
      "learning_rate": 3.507893497456181e-05,
      "loss": 0.1512,
      "step": 2289
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5717315525186366,
      "learning_rate": 3.5074260506547225e-05,
      "loss": 0.128,
      "step": 2290
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4356369814873788,
      "learning_rate": 3.506958413123867e-05,
      "loss": 0.1131,
      "step": 2291
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5316950112244213,
      "learning_rate": 3.506490584922784e-05,
      "loss": 0.1371,
      "step": 2292
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6285600360853553,
      "learning_rate": 3.5060225661106636e-05,
      "loss": 0.2104,
      "step": 2293
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44598772489222965,
      "learning_rate": 3.5055543567467244e-05,
      "loss": 0.1337,
      "step": 2294
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39710196890970895,
      "learning_rate": 3.5050859568902064e-05,
      "loss": 0.1111,
      "step": 2295
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42689801669309047,
      "learning_rate": 3.504617366600376e-05,
      "loss": 0.1443,
      "step": 2296
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4205638856971283,
      "learning_rate": 3.504148585936521e-05,
      "loss": 0.1156,
      "step": 2297
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5458887471275569,
      "learning_rate": 3.503679614957955e-05,
      "loss": 0.1394,
      "step": 2298
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6251908487758245,
      "learning_rate": 3.503210453724015e-05,
      "loss": 0.1811,
      "step": 2299
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5585225466716718,
      "learning_rate": 3.502741102294063e-05,
      "loss": 0.199,
      "step": 2300
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5455680488848759,
      "learning_rate": 3.5022715607274844e-05,
      "loss": 0.1433,
      "step": 2301
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5973739954829625,
      "learning_rate": 3.501801829083688e-05,
      "loss": 0.1675,
      "step": 2302
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.47042740466913274,
      "learning_rate": 3.501331907422107e-05,
      "loss": 0.1235,
      "step": 2303
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.49079710741970584,
      "learning_rate": 3.500861795802201e-05,
      "loss": 0.1172,
      "step": 2304
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5079860097392948,
      "learning_rate": 3.5003914942834485e-05,
      "loss": 0.1757,
      "step": 2305
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41771383622778796,
      "learning_rate": 3.499921002925357e-05,
      "loss": 0.1231,
      "step": 2306
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.471473870085316,
      "learning_rate": 3.499450321787457e-05,
      "loss": 0.1669,
      "step": 2307
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5325209617070864,
      "learning_rate": 3.4989794509293005e-05,
      "loss": 0.1229,
      "step": 2308
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.47373640362999636,
      "learning_rate": 3.4985083904104664e-05,
      "loss": 0.1193,
      "step": 2309
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41066476089416704,
      "learning_rate": 3.498037140290555e-05,
      "loss": 0.086,
      "step": 2310
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4641127087182747,
      "learning_rate": 3.4975657006291924e-05,
      "loss": 0.1478,
      "step": 2311
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.564468779561316,
      "learning_rate": 3.497094071486029e-05,
      "loss": 0.1677,
      "step": 2312
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4483025204813362,
      "learning_rate": 3.496622252920738e-05,
      "loss": 0.1222,
      "step": 2313
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5008888985895373,
      "learning_rate": 3.4961502449930165e-05,
      "loss": 0.1615,
      "step": 2314
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5723722182850036,
      "learning_rate": 3.495678047762586e-05,
      "loss": 0.1706,
      "step": 2315
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4692670831167042,
      "learning_rate": 3.495205661289193e-05,
      "loss": 0.1536,
      "step": 2316
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5784697406919613,
      "learning_rate": 3.494733085632606e-05,
      "loss": 0.1624,
      "step": 2317
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43607409603451874,
      "learning_rate": 3.494260320852619e-05,
      "loss": 0.1262,
      "step": 2318
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.444846076302698,
      "learning_rate": 3.493787367009049e-05,
      "loss": 0.1271,
      "step": 2319
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.48240519506367907,
      "learning_rate": 3.493314224161737e-05,
      "loss": 0.1148,
      "step": 2320
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4212320868421737,
      "learning_rate": 3.492840892370548e-05,
      "loss": 0.1318,
      "step": 2321
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.47440773404935044,
      "learning_rate": 3.4923673716953717e-05,
      "loss": 0.1303,
      "step": 2322
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4213940033001223,
      "learning_rate": 3.49189366219612e-05,
      "loss": 0.1235,
      "step": 2323
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.458923757892575,
      "learning_rate": 3.4914197639327306e-05,
      "loss": 0.1299,
      "step": 2324
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3922155993300642,
      "learning_rate": 3.4909456769651644e-05,
      "loss": 0.0959,
      "step": 2325
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5754029137360273,
      "learning_rate": 3.490471401353405e-05,
      "loss": 0.1093,
      "step": 2326
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6031911774138722,
      "learning_rate": 3.489996937157462e-05,
      "loss": 0.1866,
      "step": 2327
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.610369457745472,
      "learning_rate": 3.489522284437366e-05,
      "loss": 0.1819,
      "step": 2328
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5454656150987856,
      "learning_rate": 3.489047443253175e-05,
      "loss": 0.1658,
      "step": 2329
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.460045160896283,
      "learning_rate": 3.488572413664969e-05,
      "loss": 0.1574,
      "step": 2330
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5409176458883622,
      "learning_rate": 3.48809719573285e-05,
      "loss": 0.1423,
      "step": 2331
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41384444703099077,
      "learning_rate": 3.4876217895169474e-05,
      "loss": 0.1098,
      "step": 2332
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5495177582360932,
      "learning_rate": 3.487146195077412e-05,
      "loss": 0.1116,
      "step": 2333
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5059865843140341,
      "learning_rate": 3.4866704124744196e-05,
      "loss": 0.1388,
      "step": 2334
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5392883214779269,
      "learning_rate": 3.486194441768168e-05,
      "loss": 0.1202,
      "step": 2335
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4530227644455164,
      "learning_rate": 3.4857182830188816e-05,
      "loss": 0.1273,
      "step": 2336
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5598212939184222,
      "learning_rate": 3.485241936286807e-05,
      "loss": 0.1461,
      "step": 2337
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.519260077586708,
      "learning_rate": 3.484765401632214e-05,
      "loss": 0.1367,
      "step": 2338
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5711324623857723,
      "learning_rate": 3.484288679115397e-05,
      "loss": 0.1488,
      "step": 2339
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4193588740948005,
      "learning_rate": 3.483811768796674e-05,
      "loss": 0.1152,
      "step": 2340
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4506069461205856,
      "learning_rate": 3.483334670736388e-05,
      "loss": 0.1492,
      "step": 2341
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.45784512621656015,
      "learning_rate": 3.482857384994903e-05,
      "loss": 0.1149,
      "step": 2342
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.46683401494735116,
      "learning_rate": 3.4823799116326085e-05,
      "loss": 0.1272,
      "step": 2343
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5447042160401867,
      "learning_rate": 3.4819022507099184e-05,
      "loss": 0.159,
      "step": 2344
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5189526713534715,
      "learning_rate": 3.481424402287269e-05,
      "loss": 0.1381,
      "step": 2345
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40357091475042917,
      "learning_rate": 3.480946366425121e-05,
      "loss": 0.1189,
      "step": 2346
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5030026577354125,
      "learning_rate": 3.4804681431839586e-05,
      "loss": 0.1743,
      "step": 2347
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6876068465695868,
      "learning_rate": 3.4799897326242895e-05,
      "loss": 0.1863,
      "step": 2348
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4694536014782371,
      "learning_rate": 3.479511134806645e-05,
      "loss": 0.1161,
      "step": 2349
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.49525121797106664,
      "learning_rate": 3.479032349791581e-05,
      "loss": 0.1747,
      "step": 2350
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.48716213916397855,
      "learning_rate": 3.478553377639677e-05,
      "loss": 0.1594,
      "step": 2351
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5035681366450534,
      "learning_rate": 3.478074218411534e-05,
      "loss": 0.1385,
      "step": 2352
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41181360270482376,
      "learning_rate": 3.47759487216778e-05,
      "loss": 0.1071,
      "step": 2353
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.45572218935030345,
      "learning_rate": 3.477115338969065e-05,
      "loss": 0.1177,
      "step": 2354
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4684167949230917,
      "learning_rate": 3.4766356188760614e-05,
      "loss": 0.1071,
      "step": 2355
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4058304233890573,
      "learning_rate": 3.476155711949467e-05,
      "loss": 0.1147,
      "step": 2356
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5165177333147264,
      "learning_rate": 3.475675618250003e-05,
      "loss": 0.1406,
      "step": 2357
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6207859308165702,
      "learning_rate": 3.475195337838415e-05,
      "loss": 0.1986,
      "step": 2358
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4949982931126922,
      "learning_rate": 3.474714870775469e-05,
      "loss": 0.1286,
      "step": 2359
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.49226956969961394,
      "learning_rate": 3.474234217121959e-05,
      "loss": 0.1335,
      "step": 2360
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5040392441419242,
      "learning_rate": 3.4737533769386997e-05,
      "loss": 0.1175,
      "step": 2361
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.45955518575238313,
      "learning_rate": 3.473272350286529e-05,
      "loss": 0.1389,
      "step": 2362
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5275699836625878,
      "learning_rate": 3.4727911372263106e-05,
      "loss": 0.1518,
      "step": 2363
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5845572427141871,
      "learning_rate": 3.4723097378189306e-05,
      "loss": 0.1957,
      "step": 2364
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5162579957005713,
      "learning_rate": 3.4718281521252994e-05,
      "loss": 0.1458,
      "step": 2365
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44761637298698964,
      "learning_rate": 3.471346380206349e-05,
      "loss": 0.1201,
      "step": 2366
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3693440396586708,
      "learning_rate": 3.470864422123038e-05,
      "loss": 0.1405,
      "step": 2367
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5008741724032674,
      "learning_rate": 3.470382277936345e-05,
      "loss": 0.1185,
      "step": 2368
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38951496716072354,
      "learning_rate": 3.469899947707275e-05,
      "loss": 0.136,
      "step": 2369
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4024666348542523,
      "learning_rate": 3.4694174314968564e-05,
      "loss": 0.1248,
      "step": 2370
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6980626869225318,
      "learning_rate": 3.468934729366139e-05,
      "loss": 0.2104,
      "step": 2371
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.47399935352707895,
      "learning_rate": 3.468451841376198e-05,
      "loss": 0.1204,
      "step": 2372
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4058695625737495,
      "learning_rate": 3.467968767588131e-05,
      "loss": 0.1068,
      "step": 2373
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5491065408568233,
      "learning_rate": 3.467485508063061e-05,
      "loss": 0.1448,
      "step": 2374
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4023857094819941,
      "learning_rate": 3.4670020628621314e-05,
      "loss": 0.1159,
      "step": 2375
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4772763768176297,
      "learning_rate": 3.466518432046512e-05,
      "loss": 0.1196,
      "step": 2376
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.46904373502460245,
      "learning_rate": 3.466034615677395e-05,
      "loss": 0.1334,
      "step": 2377
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4962488817975877,
      "learning_rate": 3.4655506138159954e-05,
      "loss": 0.1223,
      "step": 2378
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4207307875525594,
      "learning_rate": 3.4650664265235525e-05,
      "loss": 0.0823,
      "step": 2379
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5438375040875051,
      "learning_rate": 3.464582053861329e-05,
      "loss": 0.159,
      "step": 2380
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44786654421090233,
      "learning_rate": 3.46409749589061e-05,
      "loss": 0.1324,
      "step": 2381
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5191653978394525,
      "learning_rate": 3.463612752672707e-05,
      "loss": 0.1447,
      "step": 2382
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5077798099266159,
      "learning_rate": 3.463127824268951e-05,
      "loss": 0.1347,
      "step": 2383
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.49486834952445025,
      "learning_rate": 3.462642710740699e-05,
      "loss": 0.1307,
      "step": 2384
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4767008564959547,
      "learning_rate": 3.4621574121493306e-05,
      "loss": 0.1332,
      "step": 2385
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.562184510236321,
      "learning_rate": 3.461671928556248e-05,
      "loss": 0.1334,
      "step": 2386
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4551552801256097,
      "learning_rate": 3.4611862600228806e-05,
      "loss": 0.1185,
      "step": 2387
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5190859729252469,
      "learning_rate": 3.4607004066106754e-05,
      "loss": 0.1438,
      "step": 2388
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4702150655807336,
      "learning_rate": 3.460214368381107e-05,
      "loss": 0.1352,
      "step": 2389
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41533386531436545,
      "learning_rate": 3.459728145395671e-05,
      "loss": 0.1364,
      "step": 2390
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.48752382660373395,
      "learning_rate": 3.45924173771589e-05,
      "loss": 0.1395,
      "step": 2391
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4973620624628419,
      "learning_rate": 3.458755145403306e-05,
      "loss": 0.1468,
      "step": 2392
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4156527839781709,
      "learning_rate": 3.458268368519485e-05,
      "loss": 0.0862,
      "step": 2393
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4459316400037335,
      "learning_rate": 3.457781407126018e-05,
      "loss": 0.1696,
      "step": 2394
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.45293708719556086,
      "learning_rate": 3.457294261284519e-05,
      "loss": 0.1253,
      "step": 2395
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3999450437225088,
      "learning_rate": 3.456806931056624e-05,
      "loss": 0.1176,
      "step": 2396
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4610868470655418,
      "learning_rate": 3.4563194165039936e-05,
      "loss": 0.1025,
      "step": 2397
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5221356671745733,
      "learning_rate": 3.4558317176883116e-05,
      "loss": 0.1768,
      "step": 2398
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5502811352761469,
      "learning_rate": 3.455343834671285e-05,
      "loss": 0.1496,
      "step": 2399
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5379780074360931,
      "learning_rate": 3.454855767514643e-05,
      "loss": 0.1541,
      "step": 2400
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4584004360264307,
      "learning_rate": 3.45436751628014e-05,
      "loss": 0.1067,
      "step": 2401
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39379693614412986,
      "learning_rate": 3.453879081029552e-05,
      "loss": 0.0946,
      "step": 2402
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5096344176061549,
      "learning_rate": 3.453390461824679e-05,
      "loss": 0.1382,
      "step": 2403
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4354407861256253,
      "learning_rate": 3.452901658727345e-05,
      "loss": 0.1362,
      "step": 2404
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5368259306255825,
      "learning_rate": 3.4524126717993964e-05,
      "loss": 0.1577,
      "step": 2405
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5147883245014098,
      "learning_rate": 3.451923501102703e-05,
      "loss": 0.1466,
      "step": 2406
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4669325882657561,
      "learning_rate": 3.451434146699157e-05,
      "loss": 0.1377,
      "step": 2407
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4435304152156242,
      "learning_rate": 3.450944608650677e-05,
      "loss": 0.1187,
      "step": 2408
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3853054616491007,
      "learning_rate": 3.450454887019199e-05,
      "loss": 0.1078,
      "step": 2409
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5312528610152453,
      "learning_rate": 3.449964981866689e-05,
      "loss": 0.1391,
      "step": 2410
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41947547783408645,
      "learning_rate": 3.449474893255131e-05,
      "loss": 0.1104,
      "step": 2411
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.47822334549682627,
      "learning_rate": 3.4489846212465356e-05,
      "loss": 0.1337,
      "step": 2412
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7251016430472258,
      "learning_rate": 3.448494165902935e-05,
      "loss": 0.2189,
      "step": 2413
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.541845628322965,
      "learning_rate": 3.448003527286383e-05,
      "loss": 0.1127,
      "step": 2414
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4373584075412242,
      "learning_rate": 3.4475127054589605e-05,
      "loss": 0.1131,
      "step": 2415
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.45598181851012226,
      "learning_rate": 3.447021700482769e-05,
      "loss": 0.1243,
      "step": 2416
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.46483971089122933,
      "learning_rate": 3.4465305124199334e-05,
      "loss": 0.12,
      "step": 2417
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.47634901674468855,
      "learning_rate": 3.446039141332602e-05,
      "loss": 0.0995,
      "step": 2418
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4334746188528998,
      "learning_rate": 3.4455475872829465e-05,
      "loss": 0.1259,
      "step": 2419
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.43772642202946466,
      "learning_rate": 3.4450558503331606e-05,
      "loss": 0.1142,
      "step": 2420
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6185378019041478,
      "learning_rate": 3.444563930545464e-05,
      "loss": 0.1765,
      "step": 2421
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6189667129548564,
      "learning_rate": 3.444071827982096e-05,
      "loss": 0.198,
      "step": 2422
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4302261963831156,
      "learning_rate": 3.443579542705321e-05,
      "loss": 0.1255,
      "step": 2423
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4765235071958492,
      "learning_rate": 3.4430870747774266e-05,
      "loss": 0.1215,
      "step": 2424
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39381946904576187,
      "learning_rate": 3.442594424260722e-05,
      "loss": 0.1182,
      "step": 2425
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4930452713617664,
      "learning_rate": 3.442101591217542e-05,
      "loss": 0.1228,
      "step": 2426
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.49824373431173125,
      "learning_rate": 3.441608575710242e-05,
      "loss": 0.1275,
      "step": 2427
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.47410511182412085,
      "learning_rate": 3.441115377801202e-05,
      "loss": 0.1294,
      "step": 2428
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4162248434592663,
      "learning_rate": 3.4406219975528246e-05,
      "loss": 0.0983,
      "step": 2429
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4711624057522236,
      "learning_rate": 3.440128435027536e-05,
      "loss": 0.1534,
      "step": 2430
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5580640829232395,
      "learning_rate": 3.439634690287784e-05,
      "loss": 0.1484,
      "step": 2431
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4435992495258715,
      "learning_rate": 3.43914076339604e-05,
      "loss": 0.1154,
      "step": 2432
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5904760445363514,
      "learning_rate": 3.4386466544148e-05,
      "loss": 0.158,
      "step": 2433
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4519467491023276,
      "learning_rate": 3.438152363406582e-05,
      "loss": 0.1322,
      "step": 2434
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42399426908601145,
      "learning_rate": 3.437657890433926e-05,
      "loss": 0.1439,
      "step": 2435
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.47979832217124674,
      "learning_rate": 3.437163235559396e-05,
      "loss": 0.1248,
      "step": 2436
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5027951847696335,
      "learning_rate": 3.43666839884558e-05,
      "loss": 0.1512,
      "step": 2437
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5739105331630803,
      "learning_rate": 3.4361733803550874e-05,
      "loss": 0.1732,
      "step": 2438
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5476255987843694,
      "learning_rate": 3.435678180150551e-05,
      "loss": 0.1475,
      "step": 2439
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4757794998713033,
      "learning_rate": 3.4351827982946274e-05,
      "loss": 0.0918,
      "step": 2440
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5858401917082804,
      "learning_rate": 3.434687234849995e-05,
      "loss": 0.1395,
      "step": 2441
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5510914247061958,
      "learning_rate": 3.434191489879355e-05,
      "loss": 0.151,
      "step": 2442
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5521287869389845,
      "learning_rate": 3.433695563445433e-05,
      "loss": 0.1487,
      "step": 2443
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4301850992291227,
      "learning_rate": 3.433199455610978e-05,
      "loss": 0.1298,
      "step": 2444
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.512626454406848,
      "learning_rate": 3.432703166438759e-05,
      "loss": 0.1552,
      "step": 2445
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5099205570896488,
      "learning_rate": 3.43220669599157e-05,
      "loss": 0.1541,
      "step": 2446
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.45604073546018137,
      "learning_rate": 3.431710044332229e-05,
      "loss": 0.1163,
      "step": 2447
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5146313543216957,
      "learning_rate": 3.431213211523574e-05,
      "loss": 0.1417,
      "step": 2448
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5792926646701452,
      "learning_rate": 3.4307161976284695e-05,
      "loss": 0.1584,
      "step": 2449
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4570558655459584,
      "learning_rate": 3.430219002709799e-05,
      "loss": 0.111,
      "step": 2450
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5079500452245315,
      "learning_rate": 3.429721626830472e-05,
      "loss": 0.1651,
      "step": 2451
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42599473203504645,
      "learning_rate": 3.429224070053419e-05,
      "loss": 0.1248,
      "step": 2452
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4124887898395745,
      "learning_rate": 3.428726332441595e-05,
      "loss": 0.1117,
      "step": 2453
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5736399477708762,
      "learning_rate": 3.428228414057975e-05,
      "loss": 0.157,
      "step": 2454
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5138786102962402,
      "learning_rate": 3.427730314965562e-05,
      "loss": 0.1634,
      "step": 2455
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.470029927274878,
      "learning_rate": 3.427232035227377e-05,
      "loss": 0.109,
      "step": 2456
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6285574994502313,
      "learning_rate": 3.4267335749064655e-05,
      "loss": 0.1512,
      "step": 2457
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.48596600758625846,
      "learning_rate": 3.426234934065896e-05,
      "loss": 0.1086,
      "step": 2458
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4557537744327087,
      "learning_rate": 3.42573611276876e-05,
      "loss": 0.1192,
      "step": 2459
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4629436092393336,
      "learning_rate": 3.4252371110781716e-05,
      "loss": 0.1605,
      "step": 2460
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.559382808353759,
      "learning_rate": 3.424737929057268e-05,
      "loss": 0.1447,
      "step": 2461
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5142982058483963,
      "learning_rate": 3.424238566769209e-05,
      "loss": 0.1351,
      "step": 2462
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42970532467024103,
      "learning_rate": 3.423739024277177e-05,
      "loss": 0.1268,
      "step": 2463
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.45780035660297774,
      "learning_rate": 3.423239301644377e-05,
      "loss": 0.1163,
      "step": 2464
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5660563473370789,
      "learning_rate": 3.4227393989340376e-05,
      "loss": 0.173,
      "step": 2465
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38702263115228835,
      "learning_rate": 3.42223931620941e-05,
      "loss": 0.1139,
      "step": 2466
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5000160333923749,
      "learning_rate": 3.421739053533767e-05,
      "loss": 0.1311,
      "step": 2467
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.47882141930155436,
      "learning_rate": 3.421238610970406e-05,
      "loss": 0.0989,
      "step": 2468
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4803223115394185,
      "learning_rate": 3.420737988582646e-05,
      "loss": 0.1513,
      "step": 2469
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4733201085960917,
      "learning_rate": 3.4202371864338295e-05,
      "loss": 0.1332,
      "step": 2470
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4788027933210143,
      "learning_rate": 3.41973620458732e-05,
      "loss": 0.1355,
      "step": 2471
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.46483507869550533,
      "learning_rate": 3.419235043106506e-05,
      "loss": 0.141,
      "step": 2472
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.46143146491208537,
      "learning_rate": 3.4187337020547974e-05,
      "loss": 0.1293,
      "step": 2473
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5768565102834681,
      "learning_rate": 3.4182321814956274e-05,
      "loss": 0.1671,
      "step": 2474
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.378530414241201,
      "learning_rate": 3.417730481492451e-05,
      "loss": 0.1048,
      "step": 2475
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5695373283658408,
      "learning_rate": 3.4172286021087475e-05,
      "loss": 0.1537,
      "step": 2476
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.44114690943394924,
      "learning_rate": 3.416726543408018e-05,
      "loss": 0.1093,
      "step": 2477
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.44971614335527815,
      "learning_rate": 3.416224305453785e-05,
      "loss": 0.114,
      "step": 2478
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.48868807819925997,
      "learning_rate": 3.4157218883095966e-05,
      "loss": 0.1281,
      "step": 2479
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5713469907946939,
      "learning_rate": 3.4152192920390195e-05,
      "loss": 0.1369,
      "step": 2480
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.501901675901334,
      "learning_rate": 3.414716516705648e-05,
      "loss": 0.1256,
      "step": 2481
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5675283432491127,
      "learning_rate": 3.4142135623730954e-05,
      "loss": 0.152,
      "step": 2482
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38015793984598484,
      "learning_rate": 3.413710429104999e-05,
      "loss": 0.099,
      "step": 2483
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39435350549948,
      "learning_rate": 3.413207116965018e-05,
      "loss": 0.0895,
      "step": 2484
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5166710732897631,
      "learning_rate": 3.4127036260168344e-05,
      "loss": 0.1407,
      "step": 2485
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.35974446254124093,
      "learning_rate": 3.412199956324155e-05,
      "loss": 0.0939,
      "step": 2486
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5665131303640929,
      "learning_rate": 3.411696107950706e-05,
      "loss": 0.1407,
      "step": 2487
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.45366777599828795,
      "learning_rate": 3.4111920809602374e-05,
      "loss": 0.12,
      "step": 2488
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4868389189478191,
      "learning_rate": 3.410687875416523e-05,
      "loss": 0.1439,
      "step": 2489
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4061955818995793,
      "learning_rate": 3.4101834913833576e-05,
      "loss": 0.1353,
      "step": 2490
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.395929258000338,
      "learning_rate": 3.409678928924558e-05,
      "loss": 0.1212,
      "step": 2491
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39322646773803677,
      "learning_rate": 3.4091741881039677e-05,
      "loss": 0.09,
      "step": 2492
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4876510520592031,
      "learning_rate": 3.408669268985447e-05,
      "loss": 0.1353,
      "step": 2493
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5247481150600097,
      "learning_rate": 3.4081641716328826e-05,
      "loss": 0.1392,
      "step": 2494
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42692365413770167,
      "learning_rate": 3.407658896110183e-05,
      "loss": 0.1282,
      "step": 2495
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4768050155182239,
      "learning_rate": 3.407153442481278e-05,
      "loss": 0.1276,
      "step": 2496
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4518767131546068,
      "learning_rate": 3.406647810810122e-05,
      "loss": 0.1293,
      "step": 2497
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.45255947997190943,
      "learning_rate": 3.4061420011606906e-05,
      "loss": 0.1459,
      "step": 2498
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5345606866094632,
      "learning_rate": 3.405636013596982e-05,
      "loss": 0.1599,
      "step": 2499
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5345141045731824,
      "learning_rate": 3.405129848183017e-05,
      "loss": 0.1333,
      "step": 2500
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.588238331317073,
      "learning_rate": 3.404623504982839e-05,
      "loss": 0.1604,
      "step": 2501
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5025613862115336,
      "learning_rate": 3.404116984060513e-05,
      "loss": 0.1408,
      "step": 2502
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4988122481983707,
      "learning_rate": 3.4036102854801286e-05,
      "loss": 0.1465,
      "step": 2503
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.43244140015778665,
      "learning_rate": 3.403103409305796e-05,
      "loss": 0.1288,
      "step": 2504
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5055626842353665,
      "learning_rate": 3.402596355601649e-05,
      "loss": 0.137,
      "step": 2505
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.371369028886359,
      "learning_rate": 3.402089124431843e-05,
      "loss": 0.1189,
      "step": 2506
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4596176810440495,
      "learning_rate": 3.401581715860556e-05,
      "loss": 0.1217,
      "step": 2507
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4652359694190133,
      "learning_rate": 3.4010741299519885e-05,
      "loss": 0.1208,
      "step": 2508
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5393702140144903,
      "learning_rate": 3.400566366770364e-05,
      "loss": 0.1122,
      "step": 2509
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.35453739350915175,
      "learning_rate": 3.400058426379929e-05,
      "loss": 0.0986,
      "step": 2510
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5010457725383107,
      "learning_rate": 3.399550308844949e-05,
      "loss": 0.1205,
      "step": 2511
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45002586171222564,
      "learning_rate": 3.3990420142297165e-05,
      "loss": 0.0938,
      "step": 2512
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4489480073483344,
      "learning_rate": 3.398533542598543e-05,
      "loss": 0.1305,
      "step": 2513
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5108483752991453,
      "learning_rate": 3.398024894015764e-05,
      "loss": 0.1266,
      "step": 2514
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4106827580897056,
      "learning_rate": 3.397516068545738e-05,
      "loss": 0.1051,
      "step": 2515
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6140333047386207,
      "learning_rate": 3.3970070662528436e-05,
      "loss": 0.1614,
      "step": 2516
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39032023939121047,
      "learning_rate": 3.396497887201484e-05,
      "loss": 0.1147,
      "step": 2517
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5848595431022527,
      "learning_rate": 3.395988531456083e-05,
      "loss": 0.1332,
      "step": 2518
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5156918106721885,
      "learning_rate": 3.395478999081088e-05,
      "loss": 0.1402,
      "step": 2519
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39500240482129434,
      "learning_rate": 3.394969290140969e-05,
      "loss": 0.1191,
      "step": 2520
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5060053728839906,
      "learning_rate": 3.3944594047002174e-05,
      "loss": 0.1544,
      "step": 2521
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5845301192740991,
      "learning_rate": 3.393949342823346e-05,
      "loss": 0.1808,
      "step": 2522
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38663861378832465,
      "learning_rate": 3.393439104574893e-05,
      "loss": 0.0869,
      "step": 2523
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43672018716265165,
      "learning_rate": 3.3929286900194154e-05,
      "loss": 0.1336,
      "step": 2524
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4646614743136001,
      "learning_rate": 3.392418099221495e-05,
      "loss": 0.127,
      "step": 2525
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.47530254717256076,
      "learning_rate": 3.3919073322457364e-05,
      "loss": 0.1102,
      "step": 2526
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5607044915497152,
      "learning_rate": 3.391396389156763e-05,
      "loss": 0.1363,
      "step": 2527
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3988736514299119,
      "learning_rate": 3.3908852700192236e-05,
      "loss": 0.1142,
      "step": 2528
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5249448395316002,
      "learning_rate": 3.3903739748977884e-05,
      "loss": 0.1259,
      "step": 2529
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.534279906305615,
      "learning_rate": 3.38986250385715e-05,
      "loss": 0.1523,
      "step": 2530
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.41890935143272784,
      "learning_rate": 3.3893508569620216e-05,
      "loss": 0.1267,
      "step": 2531
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44415084840521085,
      "learning_rate": 3.388839034277142e-05,
      "loss": 0.104,
      "step": 2532
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4814071485392782,
      "learning_rate": 3.38832703586727e-05,
      "loss": 0.1083,
      "step": 2533
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5115481376146486,
      "learning_rate": 3.387814861797186e-05,
      "loss": 0.122,
      "step": 2534
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5095427160952817,
      "learning_rate": 3.3873025121316945e-05,
      "loss": 0.1693,
      "step": 2535
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39141546379406983,
      "learning_rate": 3.386789986935621e-05,
      "loss": 0.0862,
      "step": 2536
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4862217767708159,
      "learning_rate": 3.386277286273814e-05,
      "loss": 0.1109,
      "step": 2537
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.439195090938755,
      "learning_rate": 3.385764410211143e-05,
      "loss": 0.1271,
      "step": 2538
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4824005616456806,
      "learning_rate": 3.3852513588125e-05,
      "loss": 0.1448,
      "step": 2539
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3998231355243317,
      "learning_rate": 3.3847381321428e-05,
      "loss": 0.0963,
      "step": 2540
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.528702405586636,
      "learning_rate": 3.384224730266982e-05,
      "loss": 0.1525,
      "step": 2541
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43344132433402593,
      "learning_rate": 3.383711153250002e-05,
      "loss": 0.082,
      "step": 2542
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4621537858738817,
      "learning_rate": 3.383197401156842e-05,
      "loss": 0.1171,
      "step": 2543
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5224876040489491,
      "learning_rate": 3.382683474052506e-05,
      "loss": 0.1669,
      "step": 2544
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4082463072792579,
      "learning_rate": 3.382169372002019e-05,
      "loss": 0.0986,
      "step": 2545
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.520631185085014,
      "learning_rate": 3.381655095070428e-05,
      "loss": 0.1199,
      "step": 2546
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4372131394945225,
      "learning_rate": 3.3811406433228034e-05,
      "loss": 0.0973,
      "step": 2547
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5215359780143387,
      "learning_rate": 3.3806260168242365e-05,
      "loss": 0.1199,
      "step": 2548
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45685334288672597,
      "learning_rate": 3.380111215639842e-05,
      "loss": 0.1436,
      "step": 2549
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5455295360376775,
      "learning_rate": 3.379596239834755e-05,
      "loss": 0.1672,
      "step": 2550
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5805738751802865,
      "learning_rate": 3.379081089474134e-05,
      "loss": 0.1487,
      "step": 2551
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38275840922087173,
      "learning_rate": 3.3785657646231596e-05,
      "loss": 0.0945,
      "step": 2552
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4860448663472422,
      "learning_rate": 3.378050265347033e-05,
      "loss": 0.1483,
      "step": 2553
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.41590030844948245,
      "learning_rate": 3.37753459171098e-05,
      "loss": 0.088,
      "step": 2554
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4933646370191088,
      "learning_rate": 3.377018743780245e-05,
      "loss": 0.1216,
      "step": 2555
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4483199872750863,
      "learning_rate": 3.376502721620098e-05,
      "loss": 0.1348,
      "step": 2556
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.57117660589376,
      "learning_rate": 3.37598652529583e-05,
      "loss": 0.1664,
      "step": 2557
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5086700241624675,
      "learning_rate": 3.375470154872751e-05,
      "loss": 0.1241,
      "step": 2558
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5585742226768746,
      "learning_rate": 3.3749536104161984e-05,
      "loss": 0.1862,
      "step": 2559
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5117998422913124,
      "learning_rate": 3.3744368919915275e-05,
      "loss": 0.1567,
      "step": 2560
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5150417728519117,
      "learning_rate": 3.373919999664117e-05,
      "loss": 0.1942,
      "step": 2561
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4999968558451024,
      "learning_rate": 3.3734029334993675e-05,
      "loss": 0.1266,
      "step": 2562
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4900600229693779,
      "learning_rate": 3.372885693562701e-05,
      "loss": 0.1373,
      "step": 2563
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38314127913920476,
      "learning_rate": 3.372368279919563e-05,
      "loss": 0.1067,
      "step": 2564
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5332637654569706,
      "learning_rate": 3.37185069263542e-05,
      "loss": 0.1791,
      "step": 2565
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5486049537742378,
      "learning_rate": 3.3713329317757594e-05,
      "loss": 0.1564,
      "step": 2566
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44130118140765495,
      "learning_rate": 3.370814997406093e-05,
      "loss": 0.1062,
      "step": 2567
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4813566919439541,
      "learning_rate": 3.370296889591953e-05,
      "loss": 0.1177,
      "step": 2568
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4349935121162049,
      "learning_rate": 3.369778608398894e-05,
      "loss": 0.1052,
      "step": 2569
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5049315084146065,
      "learning_rate": 3.369260153892491e-05,
      "loss": 0.1428,
      "step": 2570
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.46337019145395963,
      "learning_rate": 3.368741526138344e-05,
      "loss": 0.1215,
      "step": 2571
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.46036254455388864,
      "learning_rate": 3.3682227252020716e-05,
      "loss": 0.1277,
      "step": 2572
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4229826867041454,
      "learning_rate": 3.367703751149316e-05,
      "loss": 0.1204,
      "step": 2573
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5063872363256006,
      "learning_rate": 3.367184604045743e-05,
      "loss": 0.1262,
      "step": 2574
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5783439170196947,
      "learning_rate": 3.3666652839570366e-05,
      "loss": 0.1428,
      "step": 2575
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43727760111793185,
      "learning_rate": 3.3661457909489056e-05,
      "loss": 0.1148,
      "step": 2576
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4930275605392448,
      "learning_rate": 3.365626125087079e-05,
      "loss": 0.1169,
      "step": 2577
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.505109488921454,
      "learning_rate": 3.365106286437309e-05,
      "loss": 0.1634,
      "step": 2578
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.42353267303928016,
      "learning_rate": 3.3645862750653675e-05,
      "loss": 0.1051,
      "step": 2579
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4399823879921975,
      "learning_rate": 3.364066091037052e-05,
      "loss": 0.1144,
      "step": 2580
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7231892038130262,
      "learning_rate": 3.363545734418178e-05,
      "loss": 0.2078,
      "step": 2581
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.432307457539732,
      "learning_rate": 3.3630252052745844e-05,
      "loss": 0.1207,
      "step": 2582
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4000852583383531,
      "learning_rate": 3.362504503672133e-05,
      "loss": 0.1281,
      "step": 2583
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4081756728583051,
      "learning_rate": 3.361983629676705e-05,
      "loss": 0.1025,
      "step": 2584
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5165511685401845,
      "learning_rate": 3.361462583354207e-05,
      "loss": 0.1471,
      "step": 2585
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4667985349489968,
      "learning_rate": 3.360941364770562e-05,
      "loss": 0.1459,
      "step": 2586
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.40230317513025765,
      "learning_rate": 3.3604199739917205e-05,
      "loss": 0.1355,
      "step": 2587
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3959479249757505,
      "learning_rate": 3.359898411083652e-05,
      "loss": 0.1256,
      "step": 2588
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4677749665619581,
      "learning_rate": 3.359376676112347e-05,
      "loss": 0.147,
      "step": 2589
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5444453079947719,
      "learning_rate": 3.358854769143819e-05,
      "loss": 0.1601,
      "step": 2590
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.500436889987277,
      "learning_rate": 3.358332690244104e-05,
      "loss": 0.1044,
      "step": 2591
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6368952781436327,
      "learning_rate": 3.357810439479258e-05,
      "loss": 0.144,
      "step": 2592
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.47217379135439264,
      "learning_rate": 3.3572880169153595e-05,
      "loss": 0.1426,
      "step": 2593
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44072216328906727,
      "learning_rate": 3.356765422618509e-05,
      "loss": 0.1143,
      "step": 2594
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4589058666857585,
      "learning_rate": 3.35624265665483e-05,
      "loss": 0.161,
      "step": 2595
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5042011373994308,
      "learning_rate": 3.355719719090465e-05,
      "loss": 0.1374,
      "step": 2596
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5061150865570824,
      "learning_rate": 3.3551966099915776e-05,
      "loss": 0.1596,
      "step": 2597
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.368884265023208,
      "learning_rate": 3.3546733294243585e-05,
      "loss": 0.1116,
      "step": 2598
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5817271980944791,
      "learning_rate": 3.3541498774550134e-05,
      "loss": 0.1577,
      "step": 2599
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.46160100607101245,
      "learning_rate": 3.353626254149776e-05,
      "loss": 0.1329,
      "step": 2600
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4377371622051643,
      "learning_rate": 3.3531024595748955e-05,
      "loss": 0.1166,
      "step": 2601
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3762480115045868,
      "learning_rate": 3.3525784937966474e-05,
      "loss": 0.1212,
      "step": 2602
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4024979564366086,
      "learning_rate": 3.352054356881328e-05,
      "loss": 0.0978,
      "step": 2603
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4134229903727814,
      "learning_rate": 3.3515300488952534e-05,
      "loss": 0.1161,
      "step": 2604
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3793189171024969,
      "learning_rate": 3.3510055699047624e-05,
      "loss": 0.0986,
      "step": 2605
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3820031621862018,
      "learning_rate": 3.350480919976216e-05,
      "loss": 0.0968,
      "step": 2606
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4622382544556621,
      "learning_rate": 3.349956099175996e-05,
      "loss": 0.1629,
      "step": 2607
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4826096389408658,
      "learning_rate": 3.349431107570506e-05,
      "loss": 0.13,
      "step": 2608
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5134384727352723,
      "learning_rate": 3.348905945226173e-05,
      "loss": 0.1366,
      "step": 2609
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5132842831688942,
      "learning_rate": 3.348380612209441e-05,
      "loss": 0.1483,
      "step": 2610
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5784984877145398,
      "learning_rate": 3.347855108586781e-05,
      "loss": 0.1295,
      "step": 2611
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5369564601661964,
      "learning_rate": 3.347329434424683e-05,
      "loss": 0.1494,
      "step": 2612
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.536472178407048,
      "learning_rate": 3.346803589789657e-05,
      "loss": 0.1641,
      "step": 2613
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5125602907678098,
      "learning_rate": 3.346277574748238e-05,
      "loss": 0.0982,
      "step": 2614
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5289817801206103,
      "learning_rate": 3.34575138936698e-05,
      "loss": 0.1327,
      "step": 2615
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.44559800883512596,
      "learning_rate": 3.345225033712459e-05,
      "loss": 0.1007,
      "step": 2616
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5431117685898982,
      "learning_rate": 3.344698507851274e-05,
      "loss": 0.1235,
      "step": 2617
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9323867593073056,
      "learning_rate": 3.344171811850045e-05,
      "loss": 0.1646,
      "step": 2618
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4747742731418723,
      "learning_rate": 3.343644945775412e-05,
      "loss": 0.1595,
      "step": 2619
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42232267151456404,
      "learning_rate": 3.3431179096940375e-05,
      "loss": 0.13,
      "step": 2620
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4890635517828116,
      "learning_rate": 3.3425907036726055e-05,
      "loss": 0.1159,
      "step": 2621
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6026937820896767,
      "learning_rate": 3.3420633277778214e-05,
      "loss": 0.1649,
      "step": 2622
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5000208611904822,
      "learning_rate": 3.341535782076413e-05,
      "loss": 0.1136,
      "step": 2623
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.36418500893721073,
      "learning_rate": 3.341008066635129e-05,
      "loss": 0.1306,
      "step": 2624
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42959085591399254,
      "learning_rate": 3.3404801815207384e-05,
      "loss": 0.1276,
      "step": 2625
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5054793826934995,
      "learning_rate": 3.339952126800033e-05,
      "loss": 0.138,
      "step": 2626
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4707223722076287,
      "learning_rate": 3.339423902539826e-05,
      "loss": 0.1479,
      "step": 2627
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42210554958193724,
      "learning_rate": 3.3388955088069524e-05,
      "loss": 0.1442,
      "step": 2628
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.45822312554483813,
      "learning_rate": 3.3383669456682664e-05,
      "loss": 0.152,
      "step": 2629
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.48504328003701247,
      "learning_rate": 3.3378382131906465e-05,
      "loss": 0.1358,
      "step": 2630
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41985201587213494,
      "learning_rate": 3.337309311440991e-05,
      "loss": 0.0969,
      "step": 2631
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3560109457412073,
      "learning_rate": 3.33678024048622e-05,
      "loss": 0.1108,
      "step": 2632
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.744208265219096,
      "learning_rate": 3.336251000393275e-05,
      "loss": 0.1923,
      "step": 2633
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4882880858896491,
      "learning_rate": 3.335721591229119e-05,
      "loss": 0.1001,
      "step": 2634
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5291493903616434,
      "learning_rate": 3.335192013060737e-05,
      "loss": 0.1946,
      "step": 2635
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5928591771468422,
      "learning_rate": 3.334662265955133e-05,
      "loss": 0.1623,
      "step": 2636
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6567678678860573,
      "learning_rate": 3.334132349979336e-05,
      "loss": 0.2365,
      "step": 2637
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4130182954855353,
      "learning_rate": 3.3336022652003924e-05,
      "loss": 0.156,
      "step": 2638
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.49880110533805194,
      "learning_rate": 3.333072011685374e-05,
      "loss": 0.1391,
      "step": 2639
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4302024703130464,
      "learning_rate": 3.33254158950137e-05,
      "loss": 0.0964,
      "step": 2640
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3808579860564117,
      "learning_rate": 3.3320109987154956e-05,
      "loss": 0.137,
      "step": 2641
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41184524460488126,
      "learning_rate": 3.331480239394881e-05,
      "loss": 0.1043,
      "step": 2642
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4502693992440942,
      "learning_rate": 3.330949311606685e-05,
      "loss": 0.1234,
      "step": 2643
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3970998299918943,
      "learning_rate": 3.330418215418081e-05,
      "loss": 0.1239,
      "step": 2644
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.47855892374755765,
      "learning_rate": 3.3298869508962687e-05,
      "loss": 0.116,
      "step": 2645
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5117271364412973,
      "learning_rate": 3.329355518108466e-05,
      "loss": 0.1152,
      "step": 2646
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5207852023455704,
      "learning_rate": 3.328823917121914e-05,
      "loss": 0.1497,
      "step": 2647
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6219859882015374,
      "learning_rate": 3.328292148003875e-05,
      "loss": 0.2174,
      "step": 2648
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4191877823864674,
      "learning_rate": 3.32776021082163e-05,
      "loss": 0.101,
      "step": 2649
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5023916978037712,
      "learning_rate": 3.3272281056424854e-05,
      "loss": 0.1427,
      "step": 2650
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4312987030557723,
      "learning_rate": 3.326695832533764e-05,
      "loss": 0.0968,
      "step": 2651
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4996431239393354,
      "learning_rate": 3.326163391562814e-05,
      "loss": 0.1452,
      "step": 2652
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5041623019457745,
      "learning_rate": 3.325630782797004e-05,
      "loss": 0.1458,
      "step": 2653
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4463549842812147,
      "learning_rate": 3.325098006303722e-05,
      "loss": 0.1045,
      "step": 2654
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4017227342987408,
      "learning_rate": 3.324565062150379e-05,
      "loss": 0.0984,
      "step": 2655
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5222264414249359,
      "learning_rate": 3.324031950404406e-05,
      "loss": 0.1261,
      "step": 2656
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6623592749070012,
      "learning_rate": 3.3234986711332554e-05,
      "loss": 0.1706,
      "step": 2657
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4953809408081428,
      "learning_rate": 3.322965224404403e-05,
      "loss": 0.1365,
      "step": 2658
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4681862301883733,
      "learning_rate": 3.322431610285342e-05,
      "loss": 0.1148,
      "step": 2659
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.47687667121444377,
      "learning_rate": 3.3218978288435896e-05,
      "loss": 0.1349,
      "step": 2660
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.454294241768839,
      "learning_rate": 3.321363880146684e-05,
      "loss": 0.1112,
      "step": 2661
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41378591404655785,
      "learning_rate": 3.3208297642621824e-05,
      "loss": 0.0949,
      "step": 2662
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3859905746969514,
      "learning_rate": 3.320295481257665e-05,
      "loss": 0.0897,
      "step": 2663
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4841351684440104,
      "learning_rate": 3.319761031200735e-05,
      "loss": 0.1593,
      "step": 2664
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4632021342714253,
      "learning_rate": 3.319226414159011e-05,
      "loss": 0.127,
      "step": 2665
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41766667368361904,
      "learning_rate": 3.318691630200138e-05,
      "loss": 0.0876,
      "step": 2666
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41925842748575715,
      "learning_rate": 3.318156679391781e-05,
      "loss": 0.1093,
      "step": 2667
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3459959998629984,
      "learning_rate": 3.317621561801624e-05,
      "loss": 0.0824,
      "step": 2668
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4833338530581519,
      "learning_rate": 3.317086277497375e-05,
      "loss": 0.1287,
      "step": 2669
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4423587278837199,
      "learning_rate": 3.316550826546761e-05,
      "loss": 0.1021,
      "step": 2670
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.47057166578770115,
      "learning_rate": 3.31601520901753e-05,
      "loss": 0.1173,
      "step": 2671
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4638050881027339,
      "learning_rate": 3.315479424977453e-05,
      "loss": 0.1277,
      "step": 2672
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4100822381954122,
      "learning_rate": 3.314943474494321e-05,
      "loss": 0.1157,
      "step": 2673
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43098299495889725,
      "learning_rate": 3.3144073576359455e-05,
      "loss": 0.1492,
      "step": 2674
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4525672834649572,
      "learning_rate": 3.313871074470159e-05,
      "loss": 0.1066,
      "step": 2675
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.49358161396927436,
      "learning_rate": 3.313334625064816e-05,
      "loss": 0.1396,
      "step": 2676
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5213026951836351,
      "learning_rate": 3.312798009487791e-05,
      "loss": 0.1541,
      "step": 2677
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4292811292792686,
      "learning_rate": 3.312261227806982e-05,
      "loss": 0.1066,
      "step": 2678
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.49854560265725584,
      "learning_rate": 3.311724280090304e-05,
      "loss": 0.1099,
      "step": 2679
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3447681002135798,
      "learning_rate": 3.311187166405696e-05,
      "loss": 0.0769,
      "step": 2680
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5460788926708016,
      "learning_rate": 3.310649886821117e-05,
      "loss": 0.1254,
      "step": 2681
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43840882727574926,
      "learning_rate": 3.310112441404548e-05,
      "loss": 0.1102,
      "step": 2682
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.44350301625517363,
      "learning_rate": 3.309574830223988e-05,
      "loss": 0.1126,
      "step": 2683
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.44353024723205975,
      "learning_rate": 3.309037053347462e-05,
      "loss": 0.1455,
      "step": 2684
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.48658892521185115,
      "learning_rate": 3.3084991108430105e-05,
      "loss": 0.158,
      "step": 2685
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4352094771810096,
      "learning_rate": 3.3079610027786985e-05,
      "loss": 0.0968,
      "step": 2686
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.45618373180567723,
      "learning_rate": 3.307422729222611e-05,
      "loss": 0.1189,
      "step": 2687
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4506011920623656,
      "learning_rate": 3.306884290242854e-05,
      "loss": 0.1126,
      "step": 2688
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5525457377079442,
      "learning_rate": 3.306345685907553e-05,
      "loss": 0.1295,
      "step": 2689
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.359844771469462,
      "learning_rate": 3.3058069162848586e-05,
      "loss": 0.0952,
      "step": 2690
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4749128682129159,
      "learning_rate": 3.3052679814429374e-05,
      "loss": 0.1068,
      "step": 2691
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5393339385916132,
      "learning_rate": 3.3047288814499786e-05,
      "loss": 0.1034,
      "step": 2692
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6823146078509249,
      "learning_rate": 3.304189616374193e-05,
      "loss": 0.1689,
      "step": 2693
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6178597036957533,
      "learning_rate": 3.3036501862838125e-05,
      "loss": 0.2109,
      "step": 2694
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41554746119642,
      "learning_rate": 3.30311059124709e-05,
      "loss": 0.1319,
      "step": 2695
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4795717689993543,
      "learning_rate": 3.302570831332297e-05,
      "loss": 0.1187,
      "step": 2696
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4513382377898031,
      "learning_rate": 3.302030906607728e-05,
      "loss": 0.122,
      "step": 2697
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.47749757488993017,
      "learning_rate": 3.301490817141698e-05,
      "loss": 0.1422,
      "step": 2698
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4610132058252583,
      "learning_rate": 3.300950563002543e-05,
      "loss": 0.118,
      "step": 2699
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.45884733386225895,
      "learning_rate": 3.300410144258619e-05,
      "loss": 0.1575,
      "step": 2700
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33295770307284006,
      "learning_rate": 3.299869560978303e-05,
      "loss": 0.0839,
      "step": 2701
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.38639808615310256,
      "learning_rate": 3.2993288132299935e-05,
      "loss": 0.0913,
      "step": 2702
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5073165819549732,
      "learning_rate": 3.2987879010821096e-05,
      "loss": 0.1289,
      "step": 2703
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5183153480167265,
      "learning_rate": 3.298246824603091e-05,
      "loss": 0.1654,
      "step": 2704
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4461552525761872,
      "learning_rate": 3.2977055838613985e-05,
      "loss": 0.1398,
      "step": 2705
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48710769004499765,
      "learning_rate": 3.297164178925512e-05,
      "loss": 0.1248,
      "step": 2706
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4418211312216522,
      "learning_rate": 3.296622609863935e-05,
      "loss": 0.1339,
      "step": 2707
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.422403675418906,
      "learning_rate": 3.2960808767451905e-05,
      "loss": 0.1017,
      "step": 2708
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39050060198782655,
      "learning_rate": 3.295538979637821e-05,
      "loss": 0.0914,
      "step": 2709
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4273381074187671,
      "learning_rate": 3.294996918610393e-05,
      "loss": 0.1088,
      "step": 2710
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4908722151406998,
      "learning_rate": 3.294454693731488e-05,
      "loss": 0.1272,
      "step": 2711
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4371713357463505,
      "learning_rate": 3.293912305069715e-05,
      "loss": 0.1104,
      "step": 2712
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5580671268886105,
      "learning_rate": 3.293369752693699e-05,
      "loss": 0.1241,
      "step": 2713
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48085606478582876,
      "learning_rate": 3.292827036672089e-05,
      "loss": 0.1441,
      "step": 2714
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5133971433282902,
      "learning_rate": 3.2922841570735504e-05,
      "loss": 0.1328,
      "step": 2715
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.49036193039295145,
      "learning_rate": 3.291741113966773e-05,
      "loss": 0.1195,
      "step": 2716
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5488916000639453,
      "learning_rate": 3.2911979074204676e-05,
      "loss": 0.1432,
      "step": 2717
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5505345109764069,
      "learning_rate": 3.290654537503362e-05,
      "loss": 0.1289,
      "step": 2718
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7226473833519608,
      "learning_rate": 3.290111004284208e-05,
      "loss": 0.173,
      "step": 2719
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48289207370641346,
      "learning_rate": 3.2895673078317775e-05,
      "loss": 0.1001,
      "step": 2720
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4892949377658385,
      "learning_rate": 3.2890234482148615e-05,
      "loss": 0.1313,
      "step": 2721
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4395900597743192,
      "learning_rate": 3.288479425502273e-05,
      "loss": 0.1212,
      "step": 2722
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4441308690646111,
      "learning_rate": 3.2879352397628454e-05,
      "loss": 0.1519,
      "step": 2723
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4608396652619052,
      "learning_rate": 3.287390891065433e-05,
      "loss": 0.1355,
      "step": 2724
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5274939146640016,
      "learning_rate": 3.28684637947891e-05,
      "loss": 0.1455,
      "step": 2725
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4470603671786324,
      "learning_rate": 3.2863017050721715e-05,
      "loss": 0.1275,
      "step": 2726
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4137301280617195,
      "learning_rate": 3.285756867914133e-05,
      "loss": 0.1079,
      "step": 2727
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3900506183591493,
      "learning_rate": 3.2852118680737306e-05,
      "loss": 0.1016,
      "step": 2728
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4544293113005345,
      "learning_rate": 3.2846667056199235e-05,
      "loss": 0.1363,
      "step": 2729
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4558894248021203,
      "learning_rate": 3.2841213806216864e-05,
      "loss": 0.1121,
      "step": 2730
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.32132442476872114,
      "learning_rate": 3.283575893148019e-05,
      "loss": 0.0748,
      "step": 2731
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43447182214322044,
      "learning_rate": 3.283030243267939e-05,
      "loss": 0.1183,
      "step": 2732
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4582573509812947,
      "learning_rate": 3.282484431050486e-05,
      "loss": 0.1111,
      "step": 2733
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5683479439021748,
      "learning_rate": 3.281938456564721e-05,
      "loss": 0.136,
      "step": 2734
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4957322132422844,
      "learning_rate": 3.281392319879722e-05,
      "loss": 0.1446,
      "step": 2735
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40380508808796073,
      "learning_rate": 3.2808460210645906e-05,
      "loss": 0.1128,
      "step": 2736
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4764700002483128,
      "learning_rate": 3.280299560188449e-05,
      "loss": 0.1069,
      "step": 2737
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6165545392767173,
      "learning_rate": 3.2797529373204375e-05,
      "loss": 0.1615,
      "step": 2738
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.472093325753557,
      "learning_rate": 3.27920615252972e-05,
      "loss": 0.1227,
      "step": 2739
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5183952934279598,
      "learning_rate": 3.278659205885479e-05,
      "loss": 0.0944,
      "step": 2740
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5181295660206561,
      "learning_rate": 3.2781120974569165e-05,
      "loss": 0.1337,
      "step": 2741
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7172496311642858,
      "learning_rate": 3.2775648273132574e-05,
      "loss": 0.1739,
      "step": 2742
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4876381568277856,
      "learning_rate": 3.277017395523745e-05,
      "loss": 0.1425,
      "step": 2743
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4812705630083188,
      "learning_rate": 3.2764698021576446e-05,
      "loss": 0.1045,
      "step": 2744
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43638420871534683,
      "learning_rate": 3.275922047284241e-05,
      "loss": 0.1313,
      "step": 2745
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5629779586441177,
      "learning_rate": 3.27537413097284e-05,
      "loss": 0.1766,
      "step": 2746
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42660554661674854,
      "learning_rate": 3.2748260532927675e-05,
      "loss": 0.1017,
      "step": 2747
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5684302374327078,
      "learning_rate": 3.27427781431337e-05,
      "loss": 0.165,
      "step": 2748
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4390759863088853,
      "learning_rate": 3.2737294141040147e-05,
      "loss": 0.1216,
      "step": 2749
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43582255786256824,
      "learning_rate": 3.273180852734087e-05,
      "loss": 0.1053,
      "step": 2750
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.36589797825365517,
      "learning_rate": 3.2726321302729964e-05,
      "loss": 0.0894,
      "step": 2751
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3446400131204243,
      "learning_rate": 3.27208324679017e-05,
      "loss": 0.0767,
      "step": 2752
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43548492880647743,
      "learning_rate": 3.271534202355056e-05,
      "loss": 0.113,
      "step": 2753
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5204867704049548,
      "learning_rate": 3.270984997037123e-05,
      "loss": 0.124,
      "step": 2754
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4303420716418421,
      "learning_rate": 3.27043563090586e-05,
      "loss": 0.0956,
      "step": 2755
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.44570487287013144,
      "learning_rate": 3.269886104030778e-05,
      "loss": 0.1401,
      "step": 2756
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5551499333063893,
      "learning_rate": 3.2693364164814054e-05,
      "loss": 0.1553,
      "step": 2757
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3827095866541863,
      "learning_rate": 3.268786568327291e-05,
      "loss": 0.0979,
      "step": 2758
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4581304682812506,
      "learning_rate": 3.268236559638008e-05,
      "loss": 0.0926,
      "step": 2759
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5469102575653246,
      "learning_rate": 3.2676863904831444e-05,
      "loss": 0.1436,
      "step": 2760
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40900453811100485,
      "learning_rate": 3.2671360609323136e-05,
      "loss": 0.1054,
      "step": 2761
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4988024795380354,
      "learning_rate": 3.266585571055145e-05,
      "loss": 0.1202,
      "step": 2762
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5551967163969994,
      "learning_rate": 3.26603492092129e-05,
      "loss": 0.1639,
      "step": 2763
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5000273577835036,
      "learning_rate": 3.2654841106004225e-05,
      "loss": 0.1415,
      "step": 2764
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42071043980436845,
      "learning_rate": 3.264933140162233e-05,
      "loss": 0.1153,
      "step": 2765
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4066074339469922,
      "learning_rate": 3.264382009676435e-05,
      "loss": 0.1057,
      "step": 2766
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3972262502716515,
      "learning_rate": 3.2638307192127606e-05,
      "loss": 0.1395,
      "step": 2767
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5021946544885326,
      "learning_rate": 3.263279268840961e-05,
      "loss": 0.1328,
      "step": 2768
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4213494099779571,
      "learning_rate": 3.262727658630812e-05,
      "loss": 0.114,
      "step": 2769
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4496451965962864,
      "learning_rate": 3.262175888652106e-05,
      "loss": 0.114,
      "step": 2770
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4808302194168351,
      "learning_rate": 3.2616239589746555e-05,
      "loss": 0.1242,
      "step": 2771
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6925492402759219,
      "learning_rate": 3.261071869668296e-05,
      "loss": 0.1731,
      "step": 2772
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48603989972663947,
      "learning_rate": 3.26051962080288e-05,
      "loss": 0.1317,
      "step": 2773
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.538508227015408,
      "learning_rate": 3.259967212448282e-05,
      "loss": 0.1662,
      "step": 2774
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4637501137211498,
      "learning_rate": 3.2594146446743966e-05,
      "loss": 0.1212,
      "step": 2775
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.44628954664074705,
      "learning_rate": 3.2588619175511387e-05,
      "loss": 0.1086,
      "step": 2776
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3835377343449509,
      "learning_rate": 3.258309031148442e-05,
      "loss": 0.0974,
      "step": 2777
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38260678193925757,
      "learning_rate": 3.2577559855362614e-05,
      "loss": 0.0919,
      "step": 2778
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5223054173103331,
      "learning_rate": 3.257202780784572e-05,
      "loss": 0.1133,
      "step": 2779
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4268409770180373,
      "learning_rate": 3.2566494169633693e-05,
      "loss": 0.1224,
      "step": 2780
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5543464364280746,
      "learning_rate": 3.2560958941426686e-05,
      "loss": 0.1339,
      "step": 2781
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5053580609543851,
      "learning_rate": 3.255542212392505e-05,
      "loss": 0.1352,
      "step": 2782
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.47154714171298,
      "learning_rate": 3.254988371782933e-05,
      "loss": 0.1392,
      "step": 2783
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.47110212201576007,
      "learning_rate": 3.2544343723840296e-05,
      "loss": 0.1028,
      "step": 2784
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4021973575074634,
      "learning_rate": 3.2538802142658895e-05,
      "loss": 0.0959,
      "step": 2785
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41639631162882434,
      "learning_rate": 3.253325897498629e-05,
      "loss": 0.103,
      "step": 2786
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4860623104334949,
      "learning_rate": 3.252771422152383e-05,
      "loss": 0.1325,
      "step": 2787
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4068732982078391,
      "learning_rate": 3.2522167882973085e-05,
      "loss": 0.0974,
      "step": 2788
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4463915216875616,
      "learning_rate": 3.25166199600358e-05,
      "loss": 0.1066,
      "step": 2789
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4860026484542864,
      "learning_rate": 3.251107045341395e-05,
      "loss": 0.1031,
      "step": 2790
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6617598159459779,
      "learning_rate": 3.2505519363809685e-05,
      "loss": 0.2198,
      "step": 2791
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.45430772264933567,
      "learning_rate": 3.249996669192537e-05,
      "loss": 0.1146,
      "step": 2792
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38039694858574663,
      "learning_rate": 3.249441243846356e-05,
      "loss": 0.1055,
      "step": 2793
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44851697755764813,
      "learning_rate": 3.248885660412701e-05,
      "loss": 0.1335,
      "step": 2794
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4458850392397683,
      "learning_rate": 3.2483299189618705e-05,
      "loss": 0.1032,
      "step": 2795
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.46258618608540114,
      "learning_rate": 3.247774019564178e-05,
      "loss": 0.102,
      "step": 2796
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.523015093817182,
      "learning_rate": 3.24721796228996e-05,
      "loss": 0.1268,
      "step": 2797
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5075496286639818,
      "learning_rate": 3.2466617472095736e-05,
      "loss": 0.1162,
      "step": 2798
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35146832794302496,
      "learning_rate": 3.246105374393394e-05,
      "loss": 0.108,
      "step": 2799
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4337167994850244,
      "learning_rate": 3.245548843911817e-05,
      "loss": 0.1206,
      "step": 2800
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4613804386203868,
      "learning_rate": 3.244992155835259e-05,
      "loss": 0.1192,
      "step": 2801
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5545617619233006,
      "learning_rate": 3.244435310234156e-05,
      "loss": 0.1355,
      "step": 2802
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.49265364114183713,
      "learning_rate": 3.2438783071789626e-05,
      "loss": 0.0972,
      "step": 2803
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4492665472724663,
      "learning_rate": 3.243321146740155e-05,
      "loss": 0.1161,
      "step": 2804
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.40875608235538013,
      "learning_rate": 3.242763828988229e-05,
      "loss": 0.0981,
      "step": 2805
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6740897734902612,
      "learning_rate": 3.2422063539937006e-05,
      "loss": 0.1243,
      "step": 2806
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4105781927967041,
      "learning_rate": 3.241648721827104e-05,
      "loss": 0.0913,
      "step": 2807
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.48740071055850054,
      "learning_rate": 3.2410909325589954e-05,
      "loss": 0.1293,
      "step": 2808
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4675718442515754,
      "learning_rate": 3.2405329862599494e-05,
      "loss": 0.1245,
      "step": 2809
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4567423176174175,
      "learning_rate": 3.239974883000561e-05,
      "loss": 0.1248,
      "step": 2810
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.470729526396674,
      "learning_rate": 3.2394166228514455e-05,
      "loss": 0.1223,
      "step": 2811
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.43041775811614913,
      "learning_rate": 3.2388582058832375e-05,
      "loss": 0.1147,
      "step": 2812
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4887495008207538,
      "learning_rate": 3.238299632166591e-05,
      "loss": 0.1306,
      "step": 2813
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.402312083062914,
      "learning_rate": 3.237740901772181e-05,
      "loss": 0.0886,
      "step": 2814
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41691935347619574,
      "learning_rate": 3.2371820147707024e-05,
      "loss": 0.0975,
      "step": 2815
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37978861996755153,
      "learning_rate": 3.2366229712328675e-05,
      "loss": 0.0797,
      "step": 2816
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.45591320318229994,
      "learning_rate": 3.2360637712294106e-05,
      "loss": 0.102,
      "step": 2817
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4221487922910525,
      "learning_rate": 3.235504414831087e-05,
      "loss": 0.119,
      "step": 2818
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5546881984652567,
      "learning_rate": 3.234944902108669e-05,
      "loss": 0.1255,
      "step": 2819
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4819468425873679,
      "learning_rate": 3.234385233132949e-05,
      "loss": 0.1332,
      "step": 2820
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4753462796619483,
      "learning_rate": 3.233825407974741e-05,
      "loss": 0.1248,
      "step": 2821
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44142915446853137,
      "learning_rate": 3.233265426704877e-05,
      "loss": 0.0891,
      "step": 2822
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4527268798313024,
      "learning_rate": 3.2327052893942106e-05,
      "loss": 0.1097,
      "step": 2823
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35260125454367347,
      "learning_rate": 3.232144996113613e-05,
      "loss": 0.0854,
      "step": 2824
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3840752231237956,
      "learning_rate": 3.2315845469339766e-05,
      "loss": 0.0917,
      "step": 2825
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5943127525137295,
      "learning_rate": 3.231023941926213e-05,
      "loss": 0.1714,
      "step": 2826
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6812312254812225,
      "learning_rate": 3.230463181161254e-05,
      "loss": 0.1635,
      "step": 2827
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39237116582284726,
      "learning_rate": 3.22990226471005e-05,
      "loss": 0.0867,
      "step": 2828
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5052053870937787,
      "learning_rate": 3.229341192643572e-05,
      "loss": 0.1575,
      "step": 2829
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4376682230555415,
      "learning_rate": 3.2287799650328116e-05,
      "loss": 0.1048,
      "step": 2830
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6014293176434732,
      "learning_rate": 3.2282185819487774e-05,
      "loss": 0.157,
      "step": 2831
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42840982694173985,
      "learning_rate": 3.2276570434625e-05,
      "loss": 0.0969,
      "step": 2832
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4402574595751394,
      "learning_rate": 3.227095349645029e-05,
      "loss": 0.1243,
      "step": 2833
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4741245823583516,
      "learning_rate": 3.226533500567433e-05,
      "loss": 0.1467,
      "step": 2834
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36149475791765706,
      "learning_rate": 3.225971496300802e-05,
      "loss": 0.0948,
      "step": 2835
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6467126172575175,
      "learning_rate": 3.2254093369162425e-05,
      "loss": 0.1726,
      "step": 2836
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.45203656981654433,
      "learning_rate": 3.2248470224848846e-05,
      "loss": 0.1292,
      "step": 2837
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44120203205930697,
      "learning_rate": 3.2242845530778755e-05,
      "loss": 0.1327,
      "step": 2838
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36369861705234957,
      "learning_rate": 3.223721928766381e-05,
      "loss": 0.0896,
      "step": 2839
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37016563802965796,
      "learning_rate": 3.22315914962159e-05,
      "loss": 0.0971,
      "step": 2840
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38292713784327553,
      "learning_rate": 3.222596215714708e-05,
      "loss": 0.1093,
      "step": 2841
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3956013321689686,
      "learning_rate": 3.2220331271169614e-05,
      "loss": 0.1062,
      "step": 2842
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4573578279059922,
      "learning_rate": 3.2214698838995956e-05,
      "loss": 0.088,
      "step": 2843
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4162445513050773,
      "learning_rate": 3.220906486133876e-05,
      "loss": 0.1044,
      "step": 2844
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38693916897913666,
      "learning_rate": 3.2203429338910876e-05,
      "loss": 0.0889,
      "step": 2845
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5879779129875491,
      "learning_rate": 3.219779227242534e-05,
      "loss": 0.1585,
      "step": 2846
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4495205073022704,
      "learning_rate": 3.21921536625954e-05,
      "loss": 0.0807,
      "step": 2847
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.520493040179389,
      "learning_rate": 3.218651351013447e-05,
      "loss": 0.1081,
      "step": 2848
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5838184439909173,
      "learning_rate": 3.2180871815756204e-05,
      "loss": 0.137,
      "step": 2849
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.46657627271903684,
      "learning_rate": 3.217522858017442e-05,
      "loss": 0.0933,
      "step": 2850
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5860431830147785,
      "learning_rate": 3.216958380410313e-05,
      "loss": 0.1351,
      "step": 2851
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5241478043373128,
      "learning_rate": 3.216393748825654e-05,
      "loss": 0.1573,
      "step": 2852
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35970154729037884,
      "learning_rate": 3.2158289633349076e-05,
      "loss": 0.0754,
      "step": 2853
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42326465189719575,
      "learning_rate": 3.2152640240095335e-05,
      "loss": 0.0842,
      "step": 2854
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.46363169263800413,
      "learning_rate": 3.214698930921011e-05,
      "loss": 0.0894,
      "step": 2855
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41219970653237525,
      "learning_rate": 3.2141336841408406e-05,
      "loss": 0.0735,
      "step": 2856
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5205301992100385,
      "learning_rate": 3.2135682837405403e-05,
      "loss": 0.1313,
      "step": 2857
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4775367533934139,
      "learning_rate": 3.2130027297916476e-05,
      "loss": 0.1253,
      "step": 2858
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4645744633647556,
      "learning_rate": 3.2124370223657205e-05,
      "loss": 0.1098,
      "step": 2859
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41242256738250876,
      "learning_rate": 3.2118711615343366e-05,
      "loss": 0.085,
      "step": 2860
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5825409673438261,
      "learning_rate": 3.211305147369092e-05,
      "loss": 0.1389,
      "step": 2861
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.439102796151205,
      "learning_rate": 3.210738979941603e-05,
      "loss": 0.131,
      "step": 2862
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4764649807319358,
      "learning_rate": 3.210172659323504e-05,
      "loss": 0.1226,
      "step": 2863
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3781231919552791,
      "learning_rate": 3.2096061855864485e-05,
      "loss": 0.0779,
      "step": 2864
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5102179555685242,
      "learning_rate": 3.209039558802113e-05,
      "loss": 0.1239,
      "step": 2865
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.48636020416719494,
      "learning_rate": 3.2084727790421895e-05,
      "loss": 0.1267,
      "step": 2866
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.49665977447109855,
      "learning_rate": 3.2079058463783906e-05,
      "loss": 0.1205,
      "step": 2867
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.448465910619679,
      "learning_rate": 3.207338760882448e-05,
      "loss": 0.1105,
      "step": 2868
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4836145091492826,
      "learning_rate": 3.206771522626115e-05,
      "loss": 0.1269,
      "step": 2869
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.36375675050954004,
      "learning_rate": 3.20620413168116e-05,
      "loss": 0.0792,
      "step": 2870
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4618209680536433,
      "learning_rate": 3.205636588119373e-05,
      "loss": 0.1184,
      "step": 2871
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.46122777818116756,
      "learning_rate": 3.205068892012565e-05,
      "loss": 0.117,
      "step": 2872
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.49463767282111265,
      "learning_rate": 3.204501043432565e-05,
      "loss": 0.1104,
      "step": 2873
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.48935657349604333,
      "learning_rate": 3.203933042451218e-05,
      "loss": 0.1182,
      "step": 2874
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42880548638069527,
      "learning_rate": 3.203364889140393e-05,
      "loss": 0.1087,
      "step": 2875
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4579786767212388,
      "learning_rate": 3.202796583571977e-05,
      "loss": 0.1207,
      "step": 2876
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5125150527139145,
      "learning_rate": 3.2022281258178754e-05,
      "loss": 0.1431,
      "step": 2877
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5530907119209748,
      "learning_rate": 3.2016595159500127e-05,
      "loss": 0.1435,
      "step": 2878
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5459456039552176,
      "learning_rate": 3.2010907540403335e-05,
      "loss": 0.1314,
      "step": 2879
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.49508681947171745,
      "learning_rate": 3.2005218401608006e-05,
      "loss": 0.138,
      "step": 2880
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42446159451237536,
      "learning_rate": 3.1999527743833975e-05,
      "loss": 0.1169,
      "step": 2881
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3389935216748857,
      "learning_rate": 3.1993835567801266e-05,
      "loss": 0.0604,
      "step": 2882
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4630287704674285,
      "learning_rate": 3.198814187423007e-05,
      "loss": 0.1177,
      "step": 2883
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4593882630178724,
      "learning_rate": 3.19824466638408e-05,
      "loss": 0.1328,
      "step": 2884
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4489455843780021,
      "learning_rate": 3.197674993735407e-05,
      "loss": 0.1213,
      "step": 2885
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39465177224796905,
      "learning_rate": 3.1971051695490644e-05,
      "loss": 0.1138,
      "step": 2886
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6270922212403857,
      "learning_rate": 3.196535193897151e-05,
      "loss": 0.1533,
      "step": 2887
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.33597799544998475,
      "learning_rate": 3.195965066851784e-05,
      "loss": 0.0938,
      "step": 2888
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4547840199488066,
      "learning_rate": 3.195394788485099e-05,
      "loss": 0.1197,
      "step": 2889
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4167497095165373,
      "learning_rate": 3.194824358869252e-05,
      "loss": 0.0918,
      "step": 2890
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4186681261729406,
      "learning_rate": 3.194253778076417e-05,
      "loss": 0.1231,
      "step": 2891
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3788548267510137,
      "learning_rate": 3.1936830461787866e-05,
      "loss": 0.1055,
      "step": 2892
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.45524410773564244,
      "learning_rate": 3.193112163248576e-05,
      "loss": 0.1068,
      "step": 2893
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4242718391147415,
      "learning_rate": 3.192541129358014e-05,
      "loss": 0.1014,
      "step": 2894
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4096755162340635,
      "learning_rate": 3.191969944579355e-05,
      "loss": 0.0952,
      "step": 2895
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.409126697104361,
      "learning_rate": 3.191398608984867e-05,
      "loss": 0.1122,
      "step": 2896
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38579875676111497,
      "learning_rate": 3.1908271226468395e-05,
      "loss": 0.1146,
      "step": 2897
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38031963347967596,
      "learning_rate": 3.19025548563758e-05,
      "loss": 0.1034,
      "step": 2898
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.471395877095604,
      "learning_rate": 3.189683698029416e-05,
      "loss": 0.1116,
      "step": 2899
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5635761986081148,
      "learning_rate": 3.189111759894695e-05,
      "loss": 0.1578,
      "step": 2900
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.46752206184227274,
      "learning_rate": 3.188539671305781e-05,
      "loss": 0.0992,
      "step": 2901
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4336949479513155,
      "learning_rate": 3.1879674323350594e-05,
      "loss": 0.0952,
      "step": 2902
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.45388483597738255,
      "learning_rate": 3.187395043054932e-05,
      "loss": 0.1088,
      "step": 2903
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4341001903739148,
      "learning_rate": 3.186822503537823e-05,
      "loss": 0.1232,
      "step": 2904
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4210084209438096,
      "learning_rate": 3.186249813856173e-05,
      "loss": 0.1028,
      "step": 2905
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4711448527844678,
      "learning_rate": 3.1856769740824426e-05,
      "loss": 0.1076,
      "step": 2906
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5267324790195119,
      "learning_rate": 3.1851039842891116e-05,
      "loss": 0.0979,
      "step": 2907
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4197890315513478,
      "learning_rate": 3.184530844548678e-05,
      "loss": 0.1,
      "step": 2908
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5446062712251911,
      "learning_rate": 3.183957554933659e-05,
      "loss": 0.1078,
      "step": 2909
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44408810589364317,
      "learning_rate": 3.183384115516591e-05,
      "loss": 0.1134,
      "step": 2910
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.43406895209090485,
      "learning_rate": 3.18281052637003e-05,
      "loss": 0.0888,
      "step": 2911
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6769660090008992,
      "learning_rate": 3.182236787566549e-05,
      "loss": 0.1779,
      "step": 2912
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40729635541644726,
      "learning_rate": 3.181662899178743e-05,
      "loss": 0.0862,
      "step": 2913
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42780672619798754,
      "learning_rate": 3.181088861279222e-05,
      "loss": 0.1087,
      "step": 2914
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5031914779625248,
      "learning_rate": 3.180514673940619e-05,
      "loss": 0.1434,
      "step": 2915
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.416027194296565,
      "learning_rate": 3.179940337235582e-05,
      "loss": 0.1107,
      "step": 2916
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4858639811745089,
      "learning_rate": 3.179365851236781e-05,
      "loss": 0.1286,
      "step": 2917
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40715823984007565,
      "learning_rate": 3.178791216016904e-05,
      "loss": 0.1262,
      "step": 2918
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37182254501447454,
      "learning_rate": 3.1782164316486566e-05,
      "loss": 0.1141,
      "step": 2919
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3794245763245099,
      "learning_rate": 3.177641498204765e-05,
      "loss": 0.1016,
      "step": 2920
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.48080410922519184,
      "learning_rate": 3.177066415757974e-05,
      "loss": 0.1414,
      "step": 2921
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5072801587355023,
      "learning_rate": 3.1764911843810456e-05,
      "loss": 0.179,
      "step": 2922
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.43148695683022004,
      "learning_rate": 3.175915804146763e-05,
      "loss": 0.1051,
      "step": 2923
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4481569263093252,
      "learning_rate": 3.175340275127925e-05,
      "loss": 0.1208,
      "step": 2924
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3930988559004642,
      "learning_rate": 3.174764597397354e-05,
      "loss": 0.096,
      "step": 2925
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4133310154624265,
      "learning_rate": 3.1741887710278874e-05,
      "loss": 0.083,
      "step": 2926
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4863014060651003,
      "learning_rate": 3.1736127960923815e-05,
      "loss": 0.0997,
      "step": 2927
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4427860603165456,
      "learning_rate": 3.173036672663714e-05,
      "loss": 0.1048,
      "step": 2928
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4855828607164218,
      "learning_rate": 3.172460400814778e-05,
      "loss": 0.1169,
      "step": 2929
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.45067012018664165,
      "learning_rate": 3.17188398061849e-05,
      "loss": 0.1268,
      "step": 2930
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5605033299349742,
      "learning_rate": 3.1713074121477794e-05,
      "loss": 0.1664,
      "step": 2931
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5654894545201731,
      "learning_rate": 3.170730695475599e-05,
      "loss": 0.1256,
      "step": 2932
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.46496657343668496,
      "learning_rate": 3.1701538306749186e-05,
      "loss": 0.1145,
      "step": 2933
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4540724222000948,
      "learning_rate": 3.1695768178187267e-05,
      "loss": 0.1169,
      "step": 2934
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.462436727113538,
      "learning_rate": 3.168999656980031e-05,
      "loss": 0.1154,
      "step": 2935
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4836513589738762,
      "learning_rate": 3.168422348231857e-05,
      "loss": 0.1071,
      "step": 2936
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5527361273316552,
      "learning_rate": 3.167844891647251e-05,
      "loss": 0.1331,
      "step": 2937
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4452489506400164,
      "learning_rate": 3.1672672872992755e-05,
      "loss": 0.1034,
      "step": 2938
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.49691749127157675,
      "learning_rate": 3.1666895352610126e-05,
      "loss": 0.1136,
      "step": 2939
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4144521355504801,
      "learning_rate": 3.166111635605564e-05,
      "loss": 0.1044,
      "step": 2940
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.49526216991324307,
      "learning_rate": 3.165533588406049e-05,
      "loss": 0.1517,
      "step": 2941
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.43031213614920066,
      "learning_rate": 3.164955393735605e-05,
      "loss": 0.0859,
      "step": 2942
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4337079696739261,
      "learning_rate": 3.16437705166739e-05,
      "loss": 0.1198,
      "step": 2943
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4739767182805181,
      "learning_rate": 3.1637985622745795e-05,
      "loss": 0.0891,
      "step": 2944
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38762596913117836,
      "learning_rate": 3.163219925630368e-05,
      "loss": 0.0831,
      "step": 2945
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.401134674219508,
      "learning_rate": 3.1626411418079684e-05,
      "loss": 0.1198,
      "step": 2946
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39292806714341305,
      "learning_rate": 3.162062210880611e-05,
      "loss": 0.1199,
      "step": 2947
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4449422200334777,
      "learning_rate": 3.1614831329215475e-05,
      "loss": 0.0911,
      "step": 2948
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5414378312493097,
      "learning_rate": 3.1609039080040455e-05,
      "loss": 0.1068,
      "step": 2949
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.35962108810943744,
      "learning_rate": 3.160324536201393e-05,
      "loss": 0.1037,
      "step": 2950
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.48035682366989074,
      "learning_rate": 3.159745017586897e-05,
      "loss": 0.106,
      "step": 2951
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.571358961736077,
      "learning_rate": 3.159165352233879e-05,
      "loss": 0.1354,
      "step": 2952
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.47375772932888566,
      "learning_rate": 3.158585540215684e-05,
      "loss": 0.1019,
      "step": 2953
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.46424789065044625,
      "learning_rate": 3.158005581605673e-05,
      "loss": 0.0886,
      "step": 2954
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4505474343651107,
      "learning_rate": 3.157425476477227e-05,
      "loss": 0.1106,
      "step": 2955
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5112918208488875,
      "learning_rate": 3.156845224903745e-05,
      "loss": 0.1304,
      "step": 2956
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4139795040226315,
      "learning_rate": 3.156264826958642e-05,
      "loss": 0.1017,
      "step": 2957
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4014890907962274,
      "learning_rate": 3.1556842827153556e-05,
      "loss": 0.096,
      "step": 2958
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5446101291614608,
      "learning_rate": 3.15510359224734e-05,
      "loss": 0.1051,
      "step": 2959
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41049852168134116,
      "learning_rate": 3.154522755628067e-05,
      "loss": 0.1093,
      "step": 2960
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5497611318244457,
      "learning_rate": 3.153941772931029e-05,
      "loss": 0.1161,
      "step": 2961
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40641058902264804,
      "learning_rate": 3.153360644229735e-05,
      "loss": 0.0898,
      "step": 2962
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40112083652697644,
      "learning_rate": 3.1527793695977124e-05,
      "loss": 0.109,
      "step": 2963
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5210666769564253,
      "learning_rate": 3.1521979491085095e-05,
      "loss": 0.1151,
      "step": 2964
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5178861744968977,
      "learning_rate": 3.1516163828356915e-05,
      "loss": 0.1594,
      "step": 2965
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4173312450322156,
      "learning_rate": 3.15103467085284e-05,
      "loss": 0.1113,
      "step": 2966
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4244110212642795,
      "learning_rate": 3.150452813233558e-05,
      "loss": 0.1138,
      "step": 2967
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38501936393195807,
      "learning_rate": 3.149870810051467e-05,
      "loss": 0.0866,
      "step": 2968
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5662222991045568,
      "learning_rate": 3.1492886613802045e-05,
      "loss": 0.1209,
      "step": 2969
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.504351072110167,
      "learning_rate": 3.148706367293428e-05,
      "loss": 0.1275,
      "step": 2970
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5069101385879063,
      "learning_rate": 3.148123927864814e-05,
      "loss": 0.1491,
      "step": 2971
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5331641661555634,
      "learning_rate": 3.147541343168055e-05,
      "loss": 0.149,
      "step": 2972
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4708691537644511,
      "learning_rate": 3.1469586132768656e-05,
      "loss": 0.1124,
      "step": 2973
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5286898915640014,
      "learning_rate": 3.146375738264975e-05,
      "loss": 0.1011,
      "step": 2974
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4905430707132425,
      "learning_rate": 3.145792718206132e-05,
      "loss": 0.1567,
      "step": 2975
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.49901026878226007,
      "learning_rate": 3.145209553174105e-05,
      "loss": 0.1483,
      "step": 2976
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4107647151818684,
      "learning_rate": 3.14462624324268e-05,
      "loss": 0.1185,
      "step": 2977
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42616440218379914,
      "learning_rate": 3.14404278848566e-05,
      "loss": 0.0877,
      "step": 2978
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.49872181356227585,
      "learning_rate": 3.143459188976869e-05,
      "loss": 0.102,
      "step": 2979
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.44638180760277624,
      "learning_rate": 3.142875444790147e-05,
      "loss": 0.1224,
      "step": 2980
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38642494523558324,
      "learning_rate": 3.1422915559993534e-05,
      "loss": 0.0754,
      "step": 2981
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4505590264626607,
      "learning_rate": 3.141707522678365e-05,
      "loss": 0.1216,
      "step": 2982
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42332581669537106,
      "learning_rate": 3.141123344901079e-05,
      "loss": 0.0955,
      "step": 2983
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3898175095038572,
      "learning_rate": 3.140539022741408e-05,
      "loss": 0.1052,
      "step": 2984
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4274599422961613,
      "learning_rate": 3.139954556273284e-05,
      "loss": 0.0855,
      "step": 2985
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3745110025610411,
      "learning_rate": 3.139369945570659e-05,
      "loss": 0.1003,
      "step": 2986
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5244706618810534,
      "learning_rate": 3.138785190707501e-05,
      "loss": 0.1236,
      "step": 2987
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3993838235838536,
      "learning_rate": 3.138200291757797e-05,
      "loss": 0.1152,
      "step": 2988
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.565500811647934,
      "learning_rate": 3.1376152487955516e-05,
      "loss": 0.1234,
      "step": 2989
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.39855446688085017,
      "learning_rate": 3.137030061894789e-05,
      "loss": 0.1068,
      "step": 2990
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43042078737549466,
      "learning_rate": 3.1364447311295514e-05,
      "loss": 0.0912,
      "step": 2991
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.45558722510814526,
      "learning_rate": 3.135859256573898e-05,
      "loss": 0.1272,
      "step": 2992
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5314258677028182,
      "learning_rate": 3.135273638301906e-05,
      "loss": 0.1414,
      "step": 2993
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37565773661006496,
      "learning_rate": 3.134687876387673e-05,
      "loss": 0.1,
      "step": 2994
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.46104462235519333,
      "learning_rate": 3.134101970905313e-05,
      "loss": 0.1144,
      "step": 2995
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.49527339239396323,
      "learning_rate": 3.1335159219289585e-05,
      "loss": 0.111,
      "step": 2996
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40770607213618226,
      "learning_rate": 3.132929729532761e-05,
      "loss": 0.0986,
      "step": 2997
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43585775596261883,
      "learning_rate": 3.132343393790887e-05,
      "loss": 0.1067,
      "step": 2998
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38173871375776935,
      "learning_rate": 3.131756914777527e-05,
      "loss": 0.1107,
      "step": 2999
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.49243572575733324,
      "learning_rate": 3.131170292566883e-05,
      "loss": 0.109,
      "step": 3000
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3982948346999536,
      "learning_rate": 3.130583527233181e-05,
      "loss": 0.0887,
      "step": 3001
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4616214397922201,
      "learning_rate": 3.12999661885066e-05,
      "loss": 0.1206,
      "step": 3002
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.49812576566706723,
      "learning_rate": 3.129409567493581e-05,
      "loss": 0.1277,
      "step": 3003
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43991502011648514,
      "learning_rate": 3.12882237323622e-05,
      "loss": 0.1096,
      "step": 3004
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47991013035786234,
      "learning_rate": 3.128235036152875e-05,
      "loss": 0.1153,
      "step": 3005
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.361143033845957,
      "learning_rate": 3.127647556317858e-05,
      "loss": 0.0807,
      "step": 3006
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.44795305821345316,
      "learning_rate": 3.127059933805502e-05,
      "loss": 0.1059,
      "step": 3007
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3714933352187029,
      "learning_rate": 3.126472168690156e-05,
      "loss": 0.0833,
      "step": 3008
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40408548182061005,
      "learning_rate": 3.125884261046188e-05,
      "loss": 0.0964,
      "step": 3009
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3900624229530096,
      "learning_rate": 3.125296210947983e-05,
      "loss": 0.0998,
      "step": 3010
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47855207343349315,
      "learning_rate": 3.124708018469947e-05,
      "loss": 0.1206,
      "step": 3011
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4298627496012635,
      "learning_rate": 3.1241196836865e-05,
      "loss": 0.1036,
      "step": 3012
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3456469021353894,
      "learning_rate": 3.123531206672083e-05,
      "loss": 0.1003,
      "step": 3013
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4107678530987156,
      "learning_rate": 3.1229425875011534e-05,
      "loss": 0.1049,
      "step": 3014
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.45278337359952275,
      "learning_rate": 3.122353826248188e-05,
      "loss": 0.0936,
      "step": 3015
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.45648093388592437,
      "learning_rate": 3.12176492298768e-05,
      "loss": 0.0872,
      "step": 3016
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3808449766909785,
      "learning_rate": 3.1211758777941415e-05,
      "loss": 0.0757,
      "step": 3017
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.33018857473631386,
      "learning_rate": 3.120586690742102e-05,
      "loss": 0.0721,
      "step": 3018
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.46929928068532223,
      "learning_rate": 3.11999736190611e-05,
      "loss": 0.114,
      "step": 3019
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42938155639840064,
      "learning_rate": 3.119407891360732e-05,
      "loss": 0.1022,
      "step": 3020
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.32270341671587993,
      "learning_rate": 3.118818279180549e-05,
      "loss": 0.0838,
      "step": 3021
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.548832496272697,
      "learning_rate": 3.118228525440165e-05,
      "loss": 0.1125,
      "step": 3022
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4936748163709445,
      "learning_rate": 3.117638630214198e-05,
      "loss": 0.1083,
      "step": 3023
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3778902176150338,
      "learning_rate": 3.1170485935772864e-05,
      "loss": 0.09,
      "step": 3024
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43850738485026913,
      "learning_rate": 3.116458415604085e-05,
      "loss": 0.1021,
      "step": 3025
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3726730071004992,
      "learning_rate": 3.1158680963692676e-05,
      "loss": 0.0871,
      "step": 3026
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41160556315924834,
      "learning_rate": 3.115277635947524e-05,
      "loss": 0.0982,
      "step": 3027
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40699768953112764,
      "learning_rate": 3.114687034413564e-05,
      "loss": 0.086,
      "step": 3028
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.45263826602477925,
      "learning_rate": 3.114096291842114e-05,
      "loss": 0.0912,
      "step": 3029
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6396457118784367,
      "learning_rate": 3.1135054083079194e-05,
      "loss": 0.1593,
      "step": 3030
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.412874659262969,
      "learning_rate": 3.1129143838857417e-05,
      "loss": 0.1139,
      "step": 3031
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4598995094829215,
      "learning_rate": 3.112323218650362e-05,
      "loss": 0.1226,
      "step": 3032
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.31994202751805884,
      "learning_rate": 3.111731912676577e-05,
      "loss": 0.0882,
      "step": 3033
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.433226833950231,
      "learning_rate": 3.111140466039205e-05,
      "loss": 0.1066,
      "step": 3034
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4237915058401346,
      "learning_rate": 3.1105488788130774e-05,
      "loss": 0.0937,
      "step": 3035
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4376426191839178,
      "learning_rate": 3.1099571510730466e-05,
      "loss": 0.1243,
      "step": 3036
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3749582148160103,
      "learning_rate": 3.109365282893982e-05,
      "loss": 0.089,
      "step": 3037
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4552989963998003,
      "learning_rate": 3.1087732743507704e-05,
      "loss": 0.1071,
      "step": 3038
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4040797843983112,
      "learning_rate": 3.108181125518316e-05,
      "loss": 0.1025,
      "step": 3039
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3628885751590384,
      "learning_rate": 3.107588836471542e-05,
      "loss": 0.0799,
      "step": 3040
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40287186859125884,
      "learning_rate": 3.106996407285389e-05,
      "loss": 0.0883,
      "step": 3041
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6091119981188723,
      "learning_rate": 3.106403838034815e-05,
      "loss": 0.1403,
      "step": 3042
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.425885616425448,
      "learning_rate": 3.1058111287947946e-05,
      "loss": 0.11,
      "step": 3043
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5320074627861954,
      "learning_rate": 3.1052182796403225e-05,
      "loss": 0.1341,
      "step": 3044
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.408036960201621,
      "learning_rate": 3.104625290646409e-05,
      "loss": 0.1063,
      "step": 3045
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3801160552337564,
      "learning_rate": 3.104032161888084e-05,
      "loss": 0.0777,
      "step": 3046
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42475136608602376,
      "learning_rate": 3.1034388934403924e-05,
      "loss": 0.088,
      "step": 3047
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4019793554705773,
      "learning_rate": 3.1028454853784e-05,
      "loss": 0.1036,
      "step": 3048
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4516689320304525,
      "learning_rate": 3.1022519377771885e-05,
      "loss": 0.1136,
      "step": 3049
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3740455282276215,
      "learning_rate": 3.101658250711856e-05,
      "loss": 0.0722,
      "step": 3050
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4856231359497586,
      "learning_rate": 3.101064424257521e-05,
      "loss": 0.1162,
      "step": 3051
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.361839674640322,
      "learning_rate": 3.100470458489318e-05,
      "loss": 0.09,
      "step": 3052
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5044663148301474,
      "learning_rate": 3.099876353482399e-05,
      "loss": 0.1242,
      "step": 3053
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4003546721901271,
      "learning_rate": 3.099282109311934e-05,
      "loss": 0.1077,
      "step": 3054
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4444386457853068,
      "learning_rate": 3.098687726053112e-05,
      "loss": 0.1225,
      "step": 3055
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33634263382829,
      "learning_rate": 3.098093203781137e-05,
      "loss": 0.1123,
      "step": 3056
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5061390225311463,
      "learning_rate": 3.097498542571232e-05,
      "loss": 0.1259,
      "step": 3057
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3792954834776004,
      "learning_rate": 3.0969037424986376e-05,
      "loss": 0.0788,
      "step": 3058
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.38489028924950525,
      "learning_rate": 3.096308803638612e-05,
      "loss": 0.0891,
      "step": 3059
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4740600546646167,
      "learning_rate": 3.09571372606643e-05,
      "loss": 0.0922,
      "step": 3060
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.361128489013011,
      "learning_rate": 3.095118509857386e-05,
      "loss": 0.0843,
      "step": 3061
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5383975032180267,
      "learning_rate": 3.09452315508679e-05,
      "loss": 0.1501,
      "step": 3062
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4743842073534603,
      "learning_rate": 3.093927661829969e-05,
      "loss": 0.0984,
      "step": 3063
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.49085683933388163,
      "learning_rate": 3.09333203016227e-05,
      "loss": 0.1112,
      "step": 3064
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42029817266046576,
      "learning_rate": 3.0927362601590555e-05,
      "loss": 0.1206,
      "step": 3065
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4430981509275775,
      "learning_rate": 3.0921403518957076e-05,
      "loss": 0.1025,
      "step": 3066
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46659690372381546,
      "learning_rate": 3.0915443054476236e-05,
      "loss": 0.1134,
      "step": 3067
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5489216789251119,
      "learning_rate": 3.0909481208902185e-05,
      "loss": 0.1639,
      "step": 3068
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5491942681055818,
      "learning_rate": 3.090351798298926e-05,
      "loss": 0.1231,
      "step": 3069
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.49797293918309016,
      "learning_rate": 3.089755337749198e-05,
      "loss": 0.1029,
      "step": 3070
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4489617317348411,
      "learning_rate": 3.089158739316501e-05,
      "loss": 0.0974,
      "step": 3071
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6138681409915191,
      "learning_rate": 3.08856200307632e-05,
      "loss": 0.1553,
      "step": 3072
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.49109413168804644,
      "learning_rate": 3.08796512910416e-05,
      "loss": 0.1046,
      "step": 3073
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4891476992656243,
      "learning_rate": 3.08736811747554e-05,
      "loss": 0.1377,
      "step": 3074
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41818725364836296,
      "learning_rate": 3.0867709682659984e-05,
      "loss": 0.0909,
      "step": 3075
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36086557010107234,
      "learning_rate": 3.08617368155109e-05,
      "loss": 0.0724,
      "step": 3076
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.49923156757645265,
      "learning_rate": 3.0855762574063874e-05,
      "loss": 0.112,
      "step": 3077
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46968722426558046,
      "learning_rate": 3.084978695907482e-05,
      "loss": 0.1132,
      "step": 3078
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4245450684422398,
      "learning_rate": 3.0843809971299784e-05,
      "loss": 0.1032,
      "step": 3079
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.47929073538108424,
      "learning_rate": 3.0837831611495036e-05,
      "loss": 0.0958,
      "step": 3080
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46562686638170014,
      "learning_rate": 3.0831851880416994e-05,
      "loss": 0.1084,
      "step": 3081
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4238469168017291,
      "learning_rate": 3.082587077882225e-05,
      "loss": 0.0758,
      "step": 3082
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5357730123710422,
      "learning_rate": 3.081988830746756e-05,
      "loss": 0.115,
      "step": 3083
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4500577472350527,
      "learning_rate": 3.081390446710989e-05,
      "loss": 0.1171,
      "step": 3084
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4064815118334347,
      "learning_rate": 3.0807919258506336e-05,
      "loss": 0.1025,
      "step": 3085
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4635172276715917,
      "learning_rate": 3.080193268241419e-05,
      "loss": 0.1061,
      "step": 3086
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4442429329927746,
      "learning_rate": 3.079594473959091e-05,
      "loss": 0.1242,
      "step": 3087
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45337500745585574,
      "learning_rate": 3.0789955430794145e-05,
      "loss": 0.1172,
      "step": 3088
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.48484490274285524,
      "learning_rate": 3.0783964756781685e-05,
      "loss": 0.0994,
      "step": 3089
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46188765726568143,
      "learning_rate": 3.077797271831152e-05,
      "loss": 0.118,
      "step": 3090
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4415617981005447,
      "learning_rate": 3.077197931614178e-05,
      "loss": 0.0875,
      "step": 3091
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45212355477268457,
      "learning_rate": 3.076598455103081e-05,
      "loss": 0.1173,
      "step": 3092
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3798810081144888,
      "learning_rate": 3.075998842373711e-05,
      "loss": 0.1105,
      "step": 3093
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4444268437549615,
      "learning_rate": 3.0753990935019345e-05,
      "loss": 0.0947,
      "step": 3094
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3525911118397617,
      "learning_rate": 3.074799208563635e-05,
      "loss": 0.094,
      "step": 3095
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40570005460150627,
      "learning_rate": 3.074199187634713e-05,
      "loss": 0.0962,
      "step": 3096
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36367741384908137,
      "learning_rate": 3.0735990307910894e-05,
      "loss": 0.0805,
      "step": 3097
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36832007788127663,
      "learning_rate": 3.072998738108699e-05,
      "loss": 0.0955,
      "step": 3098
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4871920682926601,
      "learning_rate": 3.072398309663494e-05,
      "loss": 0.1355,
      "step": 3099
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35801060452285616,
      "learning_rate": 3.071797745531445e-05,
      "loss": 0.0859,
      "step": 3100
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4586708783824543,
      "learning_rate": 3.07119704578854e-05,
      "loss": 0.1162,
      "step": 3101
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5060537546576939,
      "learning_rate": 3.070596210510783e-05,
      "loss": 0.1317,
      "step": 3102
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42745424268350307,
      "learning_rate": 3.069995239774195e-05,
      "loss": 0.116,
      "step": 3103
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5242596617423675,
      "learning_rate": 3.069394133654815e-05,
      "loss": 0.1405,
      "step": 3104
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43990031906804894,
      "learning_rate": 3.068792892228699e-05,
      "loss": 0.0921,
      "step": 3105
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41211794442040073,
      "learning_rate": 3.068191515571921e-05,
      "loss": 0.1005,
      "step": 3106
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4081394200603503,
      "learning_rate": 3.06759000376057e-05,
      "loss": 0.0964,
      "step": 3107
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4125515161481568,
      "learning_rate": 3.066988356870752e-05,
      "loss": 0.1102,
      "step": 3108
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4089077066082897,
      "learning_rate": 3.066386574978594e-05,
      "loss": 0.126,
      "step": 3109
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45905942911947795,
      "learning_rate": 3.0657846581602355e-05,
      "loss": 0.1488,
      "step": 3110
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4605613078540536,
      "learning_rate": 3.0651826064918355e-05,
      "loss": 0.0951,
      "step": 3111
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42187291603986515,
      "learning_rate": 3.06458042004957e-05,
      "loss": 0.0849,
      "step": 3112
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36034887462740356,
      "learning_rate": 3.063978098909631e-05,
      "loss": 0.0829,
      "step": 3113
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3632977184541642,
      "learning_rate": 3.063375643148228e-05,
      "loss": 0.069,
      "step": 3114
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4851895528226818,
      "learning_rate": 3.062773052841588e-05,
      "loss": 0.1626,
      "step": 3115
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3872482835843667,
      "learning_rate": 3.062170328065954e-05,
      "loss": 0.1065,
      "step": 3116
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4971942859428728,
      "learning_rate": 3.061567468897588e-05,
      "loss": 0.1151,
      "step": 3117
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.39652990446208886,
      "learning_rate": 3.060964475412766e-05,
      "loss": 0.0838,
      "step": 3118
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3570625291474109,
      "learning_rate": 3.060361347687785e-05,
      "loss": 0.0836,
      "step": 3119
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.47844947824863304,
      "learning_rate": 3.059758085798954e-05,
      "loss": 0.1116,
      "step": 3120
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41583393051331374,
      "learning_rate": 3.059154689822603e-05,
      "loss": 0.0684,
      "step": 3121
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.48391631158130655,
      "learning_rate": 3.058551159835078e-05,
      "loss": 0.1186,
      "step": 3122
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4795059854732528,
      "learning_rate": 3.057947495912741e-05,
      "loss": 0.1028,
      "step": 3123
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.523734392830372,
      "learning_rate": 3.057343698131971e-05,
      "loss": 0.1262,
      "step": 3124
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5734783771299878,
      "learning_rate": 3.056739766569165e-05,
      "loss": 0.1342,
      "step": 3125
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35129553937460845,
      "learning_rate": 3.056135701300736e-05,
      "loss": 0.0854,
      "step": 3126
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5021104974112424,
      "learning_rate": 3.0555315024031155e-05,
      "loss": 0.114,
      "step": 3127
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43072038831456666,
      "learning_rate": 3.054927169952749e-05,
      "loss": 0.1291,
      "step": 3128
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4552643521783998,
      "learning_rate": 3.054322704026102e-05,
      "loss": 0.1403,
      "step": 3129
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4395759749888948,
      "learning_rate": 3.053718104699654e-05,
      "loss": 0.1353,
      "step": 3130
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.37793177719918475,
      "learning_rate": 3.0531133720499046e-05,
      "loss": 0.0792,
      "step": 3131
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.368637630345871,
      "learning_rate": 3.052508506153368e-05,
      "loss": 0.1311,
      "step": 3132
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3813160628005562,
      "learning_rate": 3.051903507086575e-05,
      "loss": 0.0945,
      "step": 3133
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.37838045980309515,
      "learning_rate": 3.051298374926074e-05,
      "loss": 0.1224,
      "step": 3134
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4566265825212334,
      "learning_rate": 3.050693109748431e-05,
      "loss": 0.1201,
      "step": 3135
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3721509831000016,
      "learning_rate": 3.0500877116302284e-05,
      "loss": 0.0853,
      "step": 3136
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40926806239802366,
      "learning_rate": 3.049482180648064e-05,
      "loss": 0.1013,
      "step": 3137
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4610618003445871,
      "learning_rate": 3.0488765168785544e-05,
      "loss": 0.1359,
      "step": 3138
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4495561245058242,
      "learning_rate": 3.048270720398332e-05,
      "loss": 0.1351,
      "step": 3139
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4992390743410643,
      "learning_rate": 3.047664791284046e-05,
      "loss": 0.1277,
      "step": 3140
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4582018573349948,
      "learning_rate": 3.0470587296123617e-05,
      "loss": 0.1085,
      "step": 3141
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.45195916253411794,
      "learning_rate": 3.046452535459963e-05,
      "loss": 0.0975,
      "step": 3142
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5637618851519003,
      "learning_rate": 3.0458462089035498e-05,
      "loss": 0.1244,
      "step": 3143
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4296957708776518,
      "learning_rate": 3.045239750019839e-05,
      "loss": 0.0858,
      "step": 3144
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.49455340487976396,
      "learning_rate": 3.0446331588855613e-05,
      "loss": 0.1107,
      "step": 3145
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4946187235771653,
      "learning_rate": 3.044026435577469e-05,
      "loss": 0.1298,
      "step": 3146
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.34252976453535766,
      "learning_rate": 3.0434195801723275e-05,
      "loss": 0.1011,
      "step": 3147
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4744166230045854,
      "learning_rate": 3.0428125927469198e-05,
      "loss": 0.1035,
      "step": 3148
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.49002332057732234,
      "learning_rate": 3.0422054733780474e-05,
      "loss": 0.1191,
      "step": 3149
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4361156632792582,
      "learning_rate": 3.0415982221425257e-05,
      "loss": 0.1162,
      "step": 3150
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42436254884771873,
      "learning_rate": 3.0409908391171884e-05,
      "loss": 0.1133,
      "step": 3151
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4287328345196262,
      "learning_rate": 3.040383324378885e-05,
      "loss": 0.0778,
      "step": 3152
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0794419800298358,
      "learning_rate": 3.0397756780044837e-05,
      "loss": 0.138,
      "step": 3153
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39912295750574145,
      "learning_rate": 3.0391679000708673e-05,
      "loss": 0.0969,
      "step": 3154
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46869888026919865,
      "learning_rate": 3.0385599906549355e-05,
      "loss": 0.1192,
      "step": 3155
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.454829217514146,
      "learning_rate": 3.0379519498336054e-05,
      "loss": 0.0839,
      "step": 3156
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.35027192911091515,
      "learning_rate": 3.037343777683809e-05,
      "loss": 0.0771,
      "step": 3157
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4620675281657158,
      "learning_rate": 3.036735474282498e-05,
      "loss": 0.12,
      "step": 3158
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.45333623895111336,
      "learning_rate": 3.0361270397066377e-05,
      "loss": 0.126,
      "step": 3159
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5962545625444104,
      "learning_rate": 3.035518474033212e-05,
      "loss": 0.1808,
      "step": 3160
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.41072979743761384,
      "learning_rate": 3.0349097773392207e-05,
      "loss": 0.095,
      "step": 3161
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4411565023465971,
      "learning_rate": 3.0343009497016787e-05,
      "loss": 0.1163,
      "step": 3162
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5257631137208278,
      "learning_rate": 3.0336919911976203e-05,
      "loss": 0.1497,
      "step": 3163
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4825417064766318,
      "learning_rate": 3.0330829019040945e-05,
      "loss": 0.1358,
      "step": 3164
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4884254242715508,
      "learning_rate": 3.0324736818981672e-05,
      "loss": 0.1333,
      "step": 3165
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3891675462583503,
      "learning_rate": 3.0318643312569204e-05,
      "loss": 0.1102,
      "step": 3166
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4384664011841985,
      "learning_rate": 3.0312548500574547e-05,
      "loss": 0.0736,
      "step": 3167
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5031724658634823,
      "learning_rate": 3.0306452383768833e-05,
      "loss": 0.1085,
      "step": 3168
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4123828215769342,
      "learning_rate": 3.03003549629234e-05,
      "loss": 0.1108,
      "step": 3169
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.44261685342517704,
      "learning_rate": 3.0294256238809727e-05,
      "loss": 0.1066,
      "step": 3170
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42945234626531475,
      "learning_rate": 3.0288156212199472e-05,
      "loss": 0.0761,
      "step": 3171
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.34760929176287386,
      "learning_rate": 3.0282054883864434e-05,
      "loss": 0.0874,
      "step": 3172
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37169043103966387,
      "learning_rate": 3.0275952254576614e-05,
      "loss": 0.0797,
      "step": 3173
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4751407176436066,
      "learning_rate": 3.026984832510814e-05,
      "loss": 0.1075,
      "step": 3174
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5237182604383818,
      "learning_rate": 3.026374309623133e-05,
      "loss": 0.1276,
      "step": 3175
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.557450571905389,
      "learning_rate": 3.025763656871865e-05,
      "loss": 0.1191,
      "step": 3176
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5787788508332388,
      "learning_rate": 3.0251528743342738e-05,
      "loss": 0.1531,
      "step": 3177
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5211203515441835,
      "learning_rate": 3.024541962087641e-05,
      "loss": 0.1358,
      "step": 3178
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3587496333502514,
      "learning_rate": 3.023930920209262e-05,
      "loss": 0.0987,
      "step": 3179
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37807603708210863,
      "learning_rate": 3.0233197487764494e-05,
      "loss": 0.0783,
      "step": 3180
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7034061293722425,
      "learning_rate": 3.022708447866534e-05,
      "loss": 0.1014,
      "step": 3181
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3480642903056366,
      "learning_rate": 3.0220970175568604e-05,
      "loss": 0.091,
      "step": 3182
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4264683560173701,
      "learning_rate": 3.0214854579247915e-05,
      "loss": 0.117,
      "step": 3183
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4338549608984276,
      "learning_rate": 3.020873769047705e-05,
      "loss": 0.1249,
      "step": 3184
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4383588263250387,
      "learning_rate": 3.0202619510029962e-05,
      "loss": 0.0943,
      "step": 3185
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43014782610154034,
      "learning_rate": 3.019650003868077e-05,
      "loss": 0.101,
      "step": 3186
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4398822469447675,
      "learning_rate": 3.019037927720374e-05,
      "loss": 0.105,
      "step": 3187
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4319008435523064,
      "learning_rate": 3.0184257226373317e-05,
      "loss": 0.1116,
      "step": 3188
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4481603510332299,
      "learning_rate": 3.0178133886964098e-05,
      "loss": 0.1167,
      "step": 3189
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38827553865772274,
      "learning_rate": 3.0172009259750852e-05,
      "loss": 0.0763,
      "step": 3190
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.40908766920439577,
      "learning_rate": 3.01658833455085e-05,
      "loss": 0.0878,
      "step": 3191
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5326026077988799,
      "learning_rate": 3.015975614501214e-05,
      "loss": 0.142,
      "step": 3192
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4350477874475564,
      "learning_rate": 3.0153627659037023e-05,
      "loss": 0.1156,
      "step": 3193
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.49758283651513785,
      "learning_rate": 3.0147497888358564e-05,
      "loss": 0.1409,
      "step": 3194
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3567095887273041,
      "learning_rate": 3.0141366833752344e-05,
      "loss": 0.0911,
      "step": 3195
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46343247774249074,
      "learning_rate": 3.0135234495994107e-05,
      "loss": 0.1138,
      "step": 3196
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4412212828241887,
      "learning_rate": 3.0129100875859744e-05,
      "loss": 0.085,
      "step": 3197
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4876661164356047,
      "learning_rate": 3.0122965974125335e-05,
      "loss": 0.1191,
      "step": 3198
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6069640369099728,
      "learning_rate": 3.01168297915671e-05,
      "loss": 0.1763,
      "step": 3199
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42108344948366455,
      "learning_rate": 3.0110692328961435e-05,
      "loss": 0.082,
      "step": 3200
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37898324400466255,
      "learning_rate": 3.0104553587084883e-05,
      "loss": 0.0949,
      "step": 3201
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5577067555445726,
      "learning_rate": 3.0098413566714165e-05,
      "loss": 0.1525,
      "step": 3202
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5217010390847702,
      "learning_rate": 3.0092272268626145e-05,
      "loss": 0.114,
      "step": 3203
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42151197072675173,
      "learning_rate": 3.008612969359788e-05,
      "loss": 0.1121,
      "step": 3204
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4122075149327242,
      "learning_rate": 3.0079985842406554e-05,
      "loss": 0.0909,
      "step": 3205
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.34812381463388364,
      "learning_rate": 3.0073840715829532e-05,
      "loss": 0.0837,
      "step": 3206
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.402323879751254,
      "learning_rate": 3.0067694314644334e-05,
      "loss": 0.1124,
      "step": 3207
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3650611017635027,
      "learning_rate": 3.006154663962865e-05,
      "loss": 0.0686,
      "step": 3208
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.35124720103957596,
      "learning_rate": 3.0055397691560303e-05,
      "loss": 0.0703,
      "step": 3209
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.40179917600574755,
      "learning_rate": 3.0049247471217326e-05,
      "loss": 0.0888,
      "step": 3210
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.44993540777119245,
      "learning_rate": 3.0043095979377864e-05,
      "loss": 0.1441,
      "step": 3211
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43885701757241913,
      "learning_rate": 3.0036943216820256e-05,
      "loss": 0.0976,
      "step": 3212
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3822749704184947,
      "learning_rate": 3.0030789184322982e-05,
      "loss": 0.0715,
      "step": 3213
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.44569746744374844,
      "learning_rate": 3.00246338826647e-05,
      "loss": 0.1217,
      "step": 3214
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.34571687487828434,
      "learning_rate": 3.001847731262421e-05,
      "loss": 0.0678,
      "step": 3215
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42773914769963217,
      "learning_rate": 3.001231947498048e-05,
      "loss": 0.0818,
      "step": 3216
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.515581764951698,
      "learning_rate": 3.000616037051265e-05,
      "loss": 0.1402,
      "step": 3217
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4564655095448477,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 0.1015,
      "step": 3218
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.49162276733399707,
      "learning_rate": 2.9993838364221993e-05,
      "loss": 0.1034,
      "step": 3219
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5381377207352398,
      "learning_rate": 2.998767546395822e-05,
      "loss": 0.1287,
      "step": 3220
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.45231658247828566,
      "learning_rate": 2.998151129998847e-05,
      "loss": 0.1105,
      "step": 3221
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3484451793457346,
      "learning_rate": 2.9975345873092662e-05,
      "loss": 0.0991,
      "step": 3222
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43040865289971103,
      "learning_rate": 2.9969179184050895e-05,
      "loss": 0.0889,
      "step": 3223
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4963507849974048,
      "learning_rate": 2.996301123364341e-05,
      "loss": 0.0876,
      "step": 3224
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5361396775214365,
      "learning_rate": 2.9956842022650622e-05,
      "loss": 0.1217,
      "step": 3225
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37727243606508093,
      "learning_rate": 2.9950671551853094e-05,
      "loss": 0.0954,
      "step": 3226
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46502643489555767,
      "learning_rate": 2.994449982203157e-05,
      "loss": 0.1189,
      "step": 3227
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4049779895111427,
      "learning_rate": 2.9938326833966914e-05,
      "loss": 0.1145,
      "step": 3228
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.34258256254725755,
      "learning_rate": 2.9932152588440185e-05,
      "loss": 0.0669,
      "step": 3229
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5165877457557599,
      "learning_rate": 2.992597708623259e-05,
      "loss": 0.1478,
      "step": 3230
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.33504843055745764,
      "learning_rate": 2.9919800328125498e-05,
      "loss": 0.0849,
      "step": 3231
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38863907488724414,
      "learning_rate": 2.991362231490042e-05,
      "loss": 0.1085,
      "step": 3232
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.417900186779823,
      "learning_rate": 2.9907443047339054e-05,
      "loss": 0.1147,
      "step": 3233
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4307403323914625,
      "learning_rate": 2.990126252622323e-05,
      "loss": 0.0932,
      "step": 3234
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5270524244581487,
      "learning_rate": 2.989508075233495e-05,
      "loss": 0.1367,
      "step": 3235
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4390436086968612,
      "learning_rate": 2.9888897726456374e-05,
      "loss": 0.1057,
      "step": 3236
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.41602180368912456,
      "learning_rate": 2.988271344936982e-05,
      "loss": 0.1133,
      "step": 3237
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44824996642300363,
      "learning_rate": 2.9876527921857756e-05,
      "loss": 0.12,
      "step": 3238
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4070965493040737,
      "learning_rate": 2.9870341144702833e-05,
      "loss": 0.0934,
      "step": 3239
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.48243086339322283,
      "learning_rate": 2.986415311868782e-05,
      "loss": 0.1096,
      "step": 3240
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3567004818944347,
      "learning_rate": 2.9857963844595684e-05,
      "loss": 0.072,
      "step": 3241
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37221330112865947,
      "learning_rate": 2.985177332320952e-05,
      "loss": 0.0752,
      "step": 3242
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44665488977824974,
      "learning_rate": 2.9845581555312604e-05,
      "loss": 0.0915,
      "step": 3243
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4897877760061545,
      "learning_rate": 2.9839388541688352e-05,
      "loss": 0.1076,
      "step": 3244
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.49107164717334983,
      "learning_rate": 2.9833194283120347e-05,
      "loss": 0.1221,
      "step": 3245
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3936081464253936,
      "learning_rate": 2.9826998780392324e-05,
      "loss": 0.1034,
      "step": 3246
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.49539637171394946,
      "learning_rate": 2.9820802034288178e-05,
      "loss": 0.09,
      "step": 3247
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.463286041826809,
      "learning_rate": 2.9814604045591974e-05,
      "loss": 0.0959,
      "step": 3248
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3553555486496809,
      "learning_rate": 2.9808404815087904e-05,
      "loss": 0.0872,
      "step": 3249
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3909009721555061,
      "learning_rate": 2.980220434356035e-05,
      "loss": 0.0857,
      "step": 3250
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.33218698776076433,
      "learning_rate": 2.9796002631793825e-05,
      "loss": 0.0937,
      "step": 3251
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4403915219571039,
      "learning_rate": 2.9789799680573014e-05,
      "loss": 0.1005,
      "step": 3252
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37249729357126665,
      "learning_rate": 2.9783595490682765e-05,
      "loss": 0.0647,
      "step": 3253
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32111672878415914,
      "learning_rate": 2.9777390062908056e-05,
      "loss": 0.059,
      "step": 3254
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40766659753839923,
      "learning_rate": 2.9771183398034047e-05,
      "loss": 0.1047,
      "step": 3255
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3676373992380219,
      "learning_rate": 2.976497549684605e-05,
      "loss": 0.0932,
      "step": 3256
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4815949832176938,
      "learning_rate": 2.975876636012952e-05,
      "loss": 0.107,
      "step": 3257
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4409996286604739,
      "learning_rate": 2.9752555988670084e-05,
      "loss": 0.0667,
      "step": 3258
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.451647157224258,
      "learning_rate": 2.9746344383253515e-05,
      "loss": 0.0925,
      "step": 3259
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4505782247014681,
      "learning_rate": 2.9740131544665748e-05,
      "loss": 0.1021,
      "step": 3260
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5363340851187944,
      "learning_rate": 2.9733917473692866e-05,
      "loss": 0.1397,
      "step": 3261
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39002028388069626,
      "learning_rate": 2.9727702171121125e-05,
      "loss": 0.0842,
      "step": 3262
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4836490020248306,
      "learning_rate": 2.972148563773692e-05,
      "loss": 0.126,
      "step": 3263
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3555231367152643,
      "learning_rate": 2.9715267874326805e-05,
      "loss": 0.0798,
      "step": 3264
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4027143086744787,
      "learning_rate": 2.9709048881677494e-05,
      "loss": 0.0923,
      "step": 3265
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44944412548670526,
      "learning_rate": 2.970282866057586e-05,
      "loss": 0.0982,
      "step": 3266
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4294087632965411,
      "learning_rate": 2.9696607211808915e-05,
      "loss": 0.1189,
      "step": 3267
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34902067007238874,
      "learning_rate": 2.969038453616385e-05,
      "loss": 0.0883,
      "step": 3268
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3579566790020374,
      "learning_rate": 2.9684160634427982e-05,
      "loss": 0.0775,
      "step": 3269
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.445023426125086,
      "learning_rate": 2.9677935507388817e-05,
      "loss": 0.0954,
      "step": 3270
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4022606513863688,
      "learning_rate": 2.967170915583399e-05,
      "loss": 0.1033,
      "step": 3271
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4481409827112859,
      "learning_rate": 2.96654815805513e-05,
      "loss": 0.1215,
      "step": 3272
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4549180596578309,
      "learning_rate": 2.96592527823287e-05,
      "loss": 0.1107,
      "step": 3273
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32150388915674455,
      "learning_rate": 2.965302276195431e-05,
      "loss": 0.0656,
      "step": 3274
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3153099683200609,
      "learning_rate": 2.9646791520216375e-05,
      "loss": 0.0781,
      "step": 3275
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38994854529552486,
      "learning_rate": 2.9640559057903325e-05,
      "loss": 0.0818,
      "step": 3276
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38004797554209646,
      "learning_rate": 2.963432537580372e-05,
      "loss": 0.0805,
      "step": 3277
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4639890330052402,
      "learning_rate": 2.9628090474706304e-05,
      "loss": 0.121,
      "step": 3278
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3185861049294222,
      "learning_rate": 2.9621854355399937e-05,
      "loss": 0.0663,
      "step": 3279
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39458920270360465,
      "learning_rate": 2.9615617018673663e-05,
      "loss": 0.1031,
      "step": 3280
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4040666560514684,
      "learning_rate": 2.960937846531668e-05,
      "loss": 0.0836,
      "step": 3281
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42137389975771644,
      "learning_rate": 2.9603138696118315e-05,
      "loss": 0.0966,
      "step": 3282
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3757150111833766,
      "learning_rate": 2.959689771186807e-05,
      "loss": 0.083,
      "step": 3283
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.339311292229142,
      "learning_rate": 2.9590655513355598e-05,
      "loss": 0.0807,
      "step": 3284
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3154760391617572,
      "learning_rate": 2.9584412101370708e-05,
      "loss": 0.0545,
      "step": 3285
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4455571924493586,
      "learning_rate": 2.957816747670334e-05,
      "loss": 0.0927,
      "step": 3286
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3700563083794951,
      "learning_rate": 2.9571921640143623e-05,
      "loss": 0.0876,
      "step": 3287
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3887715235881641,
      "learning_rate": 2.956567459248181e-05,
      "loss": 0.0744,
      "step": 3288
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4117860293927746,
      "learning_rate": 2.9559426334508315e-05,
      "loss": 0.0952,
      "step": 3289
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.658661951264724,
      "learning_rate": 2.9553176867013714e-05,
      "loss": 0.15,
      "step": 3290
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5343304966916523,
      "learning_rate": 2.954692619078874e-05,
      "loss": 0.1122,
      "step": 3291
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3622286117760669,
      "learning_rate": 2.9540674306624262e-05,
      "loss": 0.1008,
      "step": 3292
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4900858680749494,
      "learning_rate": 2.9534421215311303e-05,
      "loss": 0.1009,
      "step": 3293
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.616686146553049,
      "learning_rate": 2.9528166917641048e-05,
      "loss": 0.1114,
      "step": 3294
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4106133048832169,
      "learning_rate": 2.952191141440484e-05,
      "loss": 0.1133,
      "step": 3295
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.46079471768954333,
      "learning_rate": 2.951565470639415e-05,
      "loss": 0.1237,
      "step": 3296
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5117802473781548,
      "learning_rate": 2.9509396794400635e-05,
      "loss": 0.151,
      "step": 3297
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3537842378122041,
      "learning_rate": 2.9503137679216073e-05,
      "loss": 0.0599,
      "step": 3298
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4160583187636401,
      "learning_rate": 2.9496877361632422e-05,
      "loss": 0.1081,
      "step": 3299
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3779150592821644,
      "learning_rate": 2.9490615842441764e-05,
      "loss": 0.0961,
      "step": 3300
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3393794099335051,
      "learning_rate": 2.948435312243636e-05,
      "loss": 0.0766,
      "step": 3301
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.33970985844366197,
      "learning_rate": 2.94780892024086e-05,
      "loss": 0.0824,
      "step": 3302
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.35619211479526935,
      "learning_rate": 2.9471824083151043e-05,
      "loss": 0.111,
      "step": 3303
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3984910143649249,
      "learning_rate": 2.9465557765456387e-05,
      "loss": 0.0965,
      "step": 3304
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38307900785906207,
      "learning_rate": 2.9459290250117493e-05,
      "loss": 0.0838,
      "step": 3305
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3307757257244414,
      "learning_rate": 2.9453021537927363e-05,
      "loss": 0.0938,
      "step": 3306
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3836677883818397,
      "learning_rate": 2.944675162967916e-05,
      "loss": 0.0746,
      "step": 3307
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39931193867248543,
      "learning_rate": 2.9440480526166193e-05,
      "loss": 0.0873,
      "step": 3308
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4865081943986252,
      "learning_rate": 2.9434208228181923e-05,
      "loss": 0.103,
      "step": 3309
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4499228225859652,
      "learning_rate": 2.9427934736519962e-05,
      "loss": 0.1078,
      "step": 3310
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4496546413459929,
      "learning_rate": 2.9421660051974067e-05,
      "loss": 0.1068,
      "step": 3311
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39819425750013676,
      "learning_rate": 2.9415384175338154e-05,
      "loss": 0.0787,
      "step": 3312
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.500960559849647,
      "learning_rate": 2.9409107107406297e-05,
      "loss": 0.1206,
      "step": 3313
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4375505588473,
      "learning_rate": 2.9402828848972706e-05,
      "loss": 0.0837,
      "step": 3314
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40711258132424494,
      "learning_rate": 2.9396549400831745e-05,
      "loss": 0.0905,
      "step": 3315
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4505658062958951,
      "learning_rate": 2.9390268763777938e-05,
      "loss": 0.0951,
      "step": 3316
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4489243911280175,
      "learning_rate": 2.9383986938605944e-05,
      "loss": 0.0983,
      "step": 3317
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.490512358740168,
      "learning_rate": 2.937770392611058e-05,
      "loss": 0.122,
      "step": 3318
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.48373053346446265,
      "learning_rate": 2.9371419727086824e-05,
      "loss": 0.1085,
      "step": 3319
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.43288110471623803,
      "learning_rate": 2.9365134342329783e-05,
      "loss": 0.1177,
      "step": 3320
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.418275471028594,
      "learning_rate": 2.9358847772634736e-05,
      "loss": 0.121,
      "step": 3321
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34610886121013157,
      "learning_rate": 2.935256001879709e-05,
      "loss": 0.0714,
      "step": 3322
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40236443401612426,
      "learning_rate": 2.934627108161241e-05,
      "loss": 0.1028,
      "step": 3323
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.44570458869125384,
      "learning_rate": 2.9339980961876434e-05,
      "loss": 0.1304,
      "step": 3324
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5316031348178363,
      "learning_rate": 2.933368966038501e-05,
      "loss": 0.1206,
      "step": 3325
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4796466151405658,
      "learning_rate": 2.932739717793416e-05,
      "loss": 0.1003,
      "step": 3326
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3974991388731452,
      "learning_rate": 2.9321103515320047e-05,
      "loss": 0.0874,
      "step": 3327
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4008349860382639,
      "learning_rate": 2.9314808673338997e-05,
      "loss": 0.0672,
      "step": 3328
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40050698427118214,
      "learning_rate": 2.930851265278746e-05,
      "loss": 0.0858,
      "step": 3329
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4615775045591722,
      "learning_rate": 2.9302215454462063e-05,
      "loss": 0.1141,
      "step": 3330
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39894977996993175,
      "learning_rate": 2.9295917079159557e-05,
      "loss": 0.0937,
      "step": 3331
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4644302324022367,
      "learning_rate": 2.928961752767686e-05,
      "loss": 0.1062,
      "step": 3332
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3965266726627453,
      "learning_rate": 2.9283316800811032e-05,
      "loss": 0.0953,
      "step": 3333
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.47918679015431803,
      "learning_rate": 2.9277014899359284e-05,
      "loss": 0.1098,
      "step": 3334
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.457606092669368,
      "learning_rate": 2.9270711824118972e-05,
      "loss": 0.1212,
      "step": 3335
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4985575432752776,
      "learning_rate": 2.92644075758876e-05,
      "loss": 0.1026,
      "step": 3336
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39988032130152645,
      "learning_rate": 2.9258102155462824e-05,
      "loss": 0.1108,
      "step": 3337
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.46838151435607894,
      "learning_rate": 2.9251795563642445e-05,
      "loss": 0.0847,
      "step": 3338
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.47027761767225,
      "learning_rate": 2.924548780122442e-05,
      "loss": 0.106,
      "step": 3339
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4413420031217952,
      "learning_rate": 2.923917886900685e-05,
      "loss": 0.0684,
      "step": 3340
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3158040733149297,
      "learning_rate": 2.923286876778797e-05,
      "loss": 0.0642,
      "step": 3341
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4667887986254614,
      "learning_rate": 2.922655749836618e-05,
      "loss": 0.0957,
      "step": 3342
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3847996265859636,
      "learning_rate": 2.922024506154004e-05,
      "loss": 0.0785,
      "step": 3343
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3976016533625361,
      "learning_rate": 2.921393145810821e-05,
      "loss": 0.0838,
      "step": 3344
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4201546666919501,
      "learning_rate": 2.9207616688869556e-05,
      "loss": 0.1122,
      "step": 3345
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40566798847853486,
      "learning_rate": 2.9201300754623046e-05,
      "loss": 0.1104,
      "step": 3346
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4747499799261345,
      "learning_rate": 2.9194983656167823e-05,
      "loss": 0.0905,
      "step": 3347
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.46798120077697836,
      "learning_rate": 2.9188665394303163e-05,
      "loss": 0.1233,
      "step": 3348
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42366889740131286,
      "learning_rate": 2.9182345969828496e-05,
      "loss": 0.0918,
      "step": 3349
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4382036545006077,
      "learning_rate": 2.9176025383543395e-05,
      "loss": 0.0879,
      "step": 3350
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39428761957566644,
      "learning_rate": 2.9169703636247587e-05,
      "loss": 0.0956,
      "step": 3351
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42348531403800055,
      "learning_rate": 2.916338072874093e-05,
      "loss": 0.1153,
      "step": 3352
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6052951933159613,
      "learning_rate": 2.915705666182346e-05,
      "loss": 0.1387,
      "step": 3353
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4401405214120454,
      "learning_rate": 2.915073143629531e-05,
      "loss": 0.0984,
      "step": 3354
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5342718738742374,
      "learning_rate": 2.914440505295682e-05,
      "loss": 0.1223,
      "step": 3355
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.34057531869432783,
      "learning_rate": 2.9138077512608417e-05,
      "loss": 0.0911,
      "step": 3356
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.28605574002737116,
      "learning_rate": 2.9131748816050724e-05,
      "loss": 0.0959,
      "step": 3357
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4227325412905755,
      "learning_rate": 2.9125418964084474e-05,
      "loss": 0.1035,
      "step": 3358
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39024853685435873,
      "learning_rate": 2.9119087957510572e-05,
      "loss": 0.0846,
      "step": 3359
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.47840982941405963,
      "learning_rate": 2.9112755797130052e-05,
      "loss": 0.1368,
      "step": 3360
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37818614134373457,
      "learning_rate": 2.9106422483744107e-05,
      "loss": 0.0806,
      "step": 3361
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38189468517231107,
      "learning_rate": 2.910008801815406e-05,
      "loss": 0.0872,
      "step": 3362
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37712449047647206,
      "learning_rate": 2.9093752401161405e-05,
      "loss": 0.101,
      "step": 3363
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4254584882705601,
      "learning_rate": 2.908741563356774e-05,
      "loss": 0.0801,
      "step": 3364
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4003226990185956,
      "learning_rate": 2.908107771617486e-05,
      "loss": 0.0892,
      "step": 3365
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40462977759905516,
      "learning_rate": 2.9074738649784665e-05,
      "loss": 0.0911,
      "step": 3366
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37542434366289107,
      "learning_rate": 2.9068398435199215e-05,
      "loss": 0.0867,
      "step": 3367
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3312993003937427,
      "learning_rate": 2.9062057073220723e-05,
      "loss": 0.0873,
      "step": 3368
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3830628841321665,
      "learning_rate": 2.905571456465153e-05,
      "loss": 0.091,
      "step": 3369
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43911866069836175,
      "learning_rate": 2.9049370910294143e-05,
      "loss": 0.0949,
      "step": 3370
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.44958149734890546,
      "learning_rate": 2.90430261109512e-05,
      "loss": 0.0919,
      "step": 3371
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3955619869099262,
      "learning_rate": 2.9036680167425476e-05,
      "loss": 0.0699,
      "step": 3372
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3847020479142803,
      "learning_rate": 2.9030333080519913e-05,
      "loss": 0.0878,
      "step": 3373
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5157528920945456,
      "learning_rate": 2.902398485103758e-05,
      "loss": 0.1434,
      "step": 3374
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36910051804095106,
      "learning_rate": 2.90176354797817e-05,
      "loss": 0.0844,
      "step": 3375
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.34696542489149695,
      "learning_rate": 2.901128496755564e-05,
      "loss": 0.0783,
      "step": 3376
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40574100593004825,
      "learning_rate": 2.9004933315162892e-05,
      "loss": 0.0803,
      "step": 3377
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42794517652719505,
      "learning_rate": 2.899858052340713e-05,
      "loss": 0.0961,
      "step": 3378
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4214190032013644,
      "learning_rate": 2.8992226593092135e-05,
      "loss": 0.1034,
      "step": 3379
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5084984474721068,
      "learning_rate": 2.8985871525021857e-05,
      "loss": 0.0966,
      "step": 3380
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3727625817120575,
      "learning_rate": 2.8979515320000374e-05,
      "loss": 0.0975,
      "step": 3381
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4605135341803006,
      "learning_rate": 2.897315797883192e-05,
      "loss": 0.1503,
      "step": 3382
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3570303309656247,
      "learning_rate": 2.8966799502320864e-05,
      "loss": 0.0766,
      "step": 3383
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4221254770786254,
      "learning_rate": 2.896043989127172e-05,
      "loss": 0.1072,
      "step": 3384
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4293559442481154,
      "learning_rate": 2.8954079146489155e-05,
      "loss": 0.1187,
      "step": 3385
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3774202052154274,
      "learning_rate": 2.8947717268777968e-05,
      "loss": 0.0841,
      "step": 3386
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.45600114789931223,
      "learning_rate": 2.8941354258943106e-05,
      "loss": 0.107,
      "step": 3387
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.49775256268441753,
      "learning_rate": 2.8934990117789658e-05,
      "loss": 0.1456,
      "step": 3388
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.45716027940818854,
      "learning_rate": 2.8928624846122853e-05,
      "loss": 0.0761,
      "step": 3389
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43560224446282425,
      "learning_rate": 2.8922258444748074e-05,
      "loss": 0.0959,
      "step": 3390
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43425794368960496,
      "learning_rate": 2.8915890914470832e-05,
      "loss": 0.0985,
      "step": 3391
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.41410210258103936,
      "learning_rate": 2.8909522256096795e-05,
      "loss": 0.096,
      "step": 3392
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4260304273072613,
      "learning_rate": 2.8903152470431762e-05,
      "loss": 0.0918,
      "step": 3393
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43745003142422706,
      "learning_rate": 2.8896781558281688e-05,
      "loss": 0.0948,
      "step": 3394
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3717033799638932,
      "learning_rate": 2.8890409520452657e-05,
      "loss": 0.076,
      "step": 3395
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4132785573485837,
      "learning_rate": 2.88840363577509e-05,
      "loss": 0.0864,
      "step": 3396
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.41931848867259014,
      "learning_rate": 2.8877662070982783e-05,
      "loss": 0.0983,
      "step": 3397
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4257400737765631,
      "learning_rate": 2.8871286660954846e-05,
      "loss": 0.0826,
      "step": 3398
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3647279951923982,
      "learning_rate": 2.8864910128473725e-05,
      "loss": 0.0924,
      "step": 3399
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42229278505337503,
      "learning_rate": 2.8858532474346232e-05,
      "loss": 0.0712,
      "step": 3400
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4102022780976816,
      "learning_rate": 2.8852153699379305e-05,
      "loss": 0.0871,
      "step": 3401
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3721410128486554,
      "learning_rate": 2.8845773804380028e-05,
      "loss": 0.0822,
      "step": 3402
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.45634594914325,
      "learning_rate": 2.8839392790155632e-05,
      "loss": 0.1305,
      "step": 3403
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4275111482516275,
      "learning_rate": 2.883301065751348e-05,
      "loss": 0.0877,
      "step": 3404
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.41723550670313014,
      "learning_rate": 2.882662740726108e-05,
      "loss": 0.0802,
      "step": 3405
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37908156753473154,
      "learning_rate": 2.882024304020609e-05,
      "loss": 0.0671,
      "step": 3406
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39287973081997424,
      "learning_rate": 2.881385755715629e-05,
      "loss": 0.0918,
      "step": 3407
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4492151840731854,
      "learning_rate": 2.8807470958919626e-05,
      "loss": 0.0772,
      "step": 3408
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37803053188137514,
      "learning_rate": 2.880108324630417e-05,
      "loss": 0.0816,
      "step": 3409
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43195961259981386,
      "learning_rate": 2.8794694420118117e-05,
      "loss": 0.1074,
      "step": 3410
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3557425482017068,
      "learning_rate": 2.8788304481169854e-05,
      "loss": 0.075,
      "step": 3411
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43805074086940304,
      "learning_rate": 2.8781913430267857e-05,
      "loss": 0.094,
      "step": 3412
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3953387617351781,
      "learning_rate": 2.8775521268220772e-05,
      "loss": 0.0777,
      "step": 3413
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4246914065800988,
      "learning_rate": 2.876912799583737e-05,
      "loss": 0.0843,
      "step": 3414
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43593821029878677,
      "learning_rate": 2.8762733613926582e-05,
      "loss": 0.0955,
      "step": 3415
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.608713402206091,
      "learning_rate": 2.8756338123297455e-05,
      "loss": 0.1669,
      "step": 3416
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3879497855594798,
      "learning_rate": 2.8749941524759194e-05,
      "loss": 0.0721,
      "step": 3417
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.33576957808382996,
      "learning_rate": 2.8743543819121132e-05,
      "loss": 0.0916,
      "step": 3418
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3442734937040848,
      "learning_rate": 2.8737145007192756e-05,
      "loss": 0.0729,
      "step": 3419
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5710367542219306,
      "learning_rate": 2.8730745089783686e-05,
      "loss": 0.1457,
      "step": 3420
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4547104721558514,
      "learning_rate": 2.872434406770368e-05,
      "loss": 0.1238,
      "step": 3421
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.41386860680258014,
      "learning_rate": 2.871794194176263e-05,
      "loss": 0.1044,
      "step": 3422
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35018108017739,
      "learning_rate": 2.8711538712770583e-05,
      "loss": 0.0761,
      "step": 3423
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3672851676099066,
      "learning_rate": 2.8705134381537718e-05,
      "loss": 0.077,
      "step": 3424
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.346085482427561,
      "learning_rate": 2.8698728948874344e-05,
      "loss": 0.0721,
      "step": 3425
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4026717913422045,
      "learning_rate": 2.869232241559093e-05,
      "loss": 0.0994,
      "step": 3426
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.381926269773148,
      "learning_rate": 2.868591478249806e-05,
      "loss": 0.068,
      "step": 3427
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4089908756005457,
      "learning_rate": 2.8679506050406475e-05,
      "loss": 0.1011,
      "step": 3428
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5281927832719631,
      "learning_rate": 2.8673096220127062e-05,
      "loss": 0.138,
      "step": 3429
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4583565659125651,
      "learning_rate": 2.8666685292470814e-05,
      "loss": 0.1121,
      "step": 3430
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43365182573569055,
      "learning_rate": 2.8660273268248894e-05,
      "loss": 0.1265,
      "step": 3431
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4293034484471088,
      "learning_rate": 2.8653860148272596e-05,
      "loss": 0.0943,
      "step": 3432
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5120178396909366,
      "learning_rate": 2.8647445933353342e-05,
      "loss": 0.1,
      "step": 3433
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4431122582706101,
      "learning_rate": 2.8641030624302704e-05,
      "loss": 0.1247,
      "step": 3434
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.48175600561744814,
      "learning_rate": 2.8634614221932394e-05,
      "loss": 0.1187,
      "step": 3435
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40393398418428117,
      "learning_rate": 2.8628196727054244e-05,
      "loss": 0.0883,
      "step": 3436
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3767657827401431,
      "learning_rate": 2.8621778140480252e-05,
      "loss": 0.0675,
      "step": 3437
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4296220556185453,
      "learning_rate": 2.8615358463022533e-05,
      "loss": 0.1052,
      "step": 3438
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3682775550174626,
      "learning_rate": 2.8608937695493347e-05,
      "loss": 0.1028,
      "step": 3439
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3807212383455408,
      "learning_rate": 2.860251583870509e-05,
      "loss": 0.1011,
      "step": 3440
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42134723500450993,
      "learning_rate": 2.8596092893470296e-05,
      "loss": 0.0813,
      "step": 3441
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.32172565884464516,
      "learning_rate": 2.8589668860601643e-05,
      "loss": 0.0878,
      "step": 3442
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4967773113546929,
      "learning_rate": 2.858324374091194e-05,
      "loss": 0.1284,
      "step": 3443
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3414099001307428,
      "learning_rate": 2.857681753521413e-05,
      "loss": 0.0778,
      "step": 3444
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4045005322253636,
      "learning_rate": 2.8570390244321303e-05,
      "loss": 0.0757,
      "step": 3445
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.46888621258530566,
      "learning_rate": 2.856396186904669e-05,
      "loss": 0.0696,
      "step": 3446
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.38386437839432436,
      "learning_rate": 2.8557532410203633e-05,
      "loss": 0.0714,
      "step": 3447
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3398009635036262,
      "learning_rate": 2.8551101868605644e-05,
      "loss": 0.0772,
      "step": 3448
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.39659506093713975,
      "learning_rate": 2.8544670245066353e-05,
      "loss": 0.0751,
      "step": 3449
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4322757449914007,
      "learning_rate": 2.8538237540399528e-05,
      "loss": 0.1132,
      "step": 3450
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36379845011414,
      "learning_rate": 2.8531803755419082e-05,
      "loss": 0.0995,
      "step": 3451
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5417590490447135,
      "learning_rate": 2.8525368890939055e-05,
      "loss": 0.1388,
      "step": 3452
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3858068484422771,
      "learning_rate": 2.8518932947773625e-05,
      "loss": 0.0839,
      "step": 3453
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3646439047860147,
      "learning_rate": 2.851249592673712e-05,
      "loss": 0.0626,
      "step": 3454
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4154183483504243,
      "learning_rate": 2.850605782864398e-05,
      "loss": 0.0875,
      "step": 3455
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4361004753783307,
      "learning_rate": 2.8499618654308815e-05,
      "loss": 0.1194,
      "step": 3456
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5199713313012686,
      "learning_rate": 2.8493178404546334e-05,
      "loss": 0.131,
      "step": 3457
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3846459565712655,
      "learning_rate": 2.8486737080171405e-05,
      "loss": 0.0732,
      "step": 3458
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.38434917161583537,
      "learning_rate": 2.8480294681999026e-05,
      "loss": 0.0728,
      "step": 3459
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.31545425177668673,
      "learning_rate": 2.847385121084434e-05,
      "loss": 0.0712,
      "step": 3460
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37580723107630243,
      "learning_rate": 2.8467406667522597e-05,
      "loss": 0.095,
      "step": 3461
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36407093644088173,
      "learning_rate": 2.8460961052849222e-05,
      "loss": 0.085,
      "step": 3462
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4195142319848757,
      "learning_rate": 2.845451436763975e-05,
      "loss": 0.0899,
      "step": 3463
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3311234124341084,
      "learning_rate": 2.8448066612709854e-05,
      "loss": 0.0795,
      "step": 3464
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36759958072841836,
      "learning_rate": 2.8441617788875352e-05,
      "loss": 0.0738,
      "step": 3465
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4867884437542287,
      "learning_rate": 2.843516789695219e-05,
      "loss": 0.0874,
      "step": 3466
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.49147631717274276,
      "learning_rate": 2.842871693775645e-05,
      "loss": 0.1025,
      "step": 3467
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37212141191715153,
      "learning_rate": 2.842226491210434e-05,
      "loss": 0.0721,
      "step": 3468
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.555623782127599,
      "learning_rate": 2.8415811820812226e-05,
      "loss": 0.1224,
      "step": 3469
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42344359784116864,
      "learning_rate": 2.8409357664696585e-05,
      "loss": 0.0955,
      "step": 3470
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.46745134704804603,
      "learning_rate": 2.8402902444574052e-05,
      "loss": 0.0928,
      "step": 3471
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5694713660469333,
      "learning_rate": 2.8396446161261372e-05,
      "loss": 0.1153,
      "step": 3472
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.531193225295172,
      "learning_rate": 2.8389988815575443e-05,
      "loss": 0.1352,
      "step": 3473
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3936939797589597,
      "learning_rate": 2.8383530408333285e-05,
      "loss": 0.0901,
      "step": 3474
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4029800050782596,
      "learning_rate": 2.8377070940352066e-05,
      "loss": 0.0945,
      "step": 3475
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37695077292543594,
      "learning_rate": 2.8370610412449066e-05,
      "loss": 0.0938,
      "step": 3476
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3364952691101305,
      "learning_rate": 2.8364148825441725e-05,
      "loss": 0.0674,
      "step": 3477
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4415806451134829,
      "learning_rate": 2.8357686180147604e-05,
      "loss": 0.107,
      "step": 3478
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4453857093998799,
      "learning_rate": 2.83512224773844e-05,
      "loss": 0.0994,
      "step": 3479
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.342291577506136,
      "learning_rate": 2.834475771796993e-05,
      "loss": 0.067,
      "step": 3480
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.535475670374912,
      "learning_rate": 2.833829190272218e-05,
      "loss": 0.1448,
      "step": 3481
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3911197009375313,
      "learning_rate": 2.8331825032459228e-05,
      "loss": 0.0995,
      "step": 3482
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36371226019911224,
      "learning_rate": 2.832535710799931e-05,
      "loss": 0.094,
      "step": 3483
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4007534523111423,
      "learning_rate": 2.8318888130160796e-05,
      "loss": 0.0827,
      "step": 3484
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3752636777200926,
      "learning_rate": 2.831241809976218e-05,
      "loss": 0.0828,
      "step": 3485
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3747609687978747,
      "learning_rate": 2.830594701762209e-05,
      "loss": 0.0797,
      "step": 3486
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3972239619700844,
      "learning_rate": 2.829947488455929e-05,
      "loss": 0.086,
      "step": 3487
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4071037418295652,
      "learning_rate": 2.8293001701392677e-05,
      "loss": 0.1054,
      "step": 3488
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4251344825642706,
      "learning_rate": 2.828652746894129e-05,
      "loss": 0.0913,
      "step": 3489
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3226936272499918,
      "learning_rate": 2.828005218802427e-05,
      "loss": 0.0739,
      "step": 3490
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.33636715493043123,
      "learning_rate": 2.8273575859460935e-05,
      "loss": 0.0833,
      "step": 3491
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.44709308077996146,
      "learning_rate": 2.8267098484070693e-05,
      "loss": 0.1168,
      "step": 3492
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4165290028925522,
      "learning_rate": 2.826062006267312e-05,
      "loss": 0.0822,
      "step": 3493
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5093707312656639,
      "learning_rate": 2.8254140596087897e-05,
      "loss": 0.1448,
      "step": 3494
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42914811004716735,
      "learning_rate": 2.8247660085134856e-05,
      "loss": 0.114,
      "step": 3495
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4143769311931827,
      "learning_rate": 2.8241178530633947e-05,
      "loss": 0.1122,
      "step": 3496
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5156118217864979,
      "learning_rate": 2.8234695933405255e-05,
      "loss": 0.1163,
      "step": 3497
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4423480493895119,
      "learning_rate": 2.822821229426902e-05,
      "loss": 0.1216,
      "step": 3498
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4392958635172358,
      "learning_rate": 2.8221727614045576e-05,
      "loss": 0.1143,
      "step": 3499
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4309899963038868,
      "learning_rate": 2.8215241893555415e-05,
      "loss": 0.089,
      "step": 3500
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5637227913746374,
      "learning_rate": 2.8208755133619146e-05,
      "loss": 0.1337,
      "step": 3501
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3888968770118791,
      "learning_rate": 2.8202267335057522e-05,
      "loss": 0.0901,
      "step": 3502
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3692297247677406,
      "learning_rate": 2.8195778498691422e-05,
      "loss": 0.0698,
      "step": 3503
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.41266208123128156,
      "learning_rate": 2.818928862534185e-05,
      "loss": 0.0679,
      "step": 3504
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.44760630264050744,
      "learning_rate": 2.8182797715829955e-05,
      "loss": 0.1241,
      "step": 3505
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4358505934907392,
      "learning_rate": 2.817630577097701e-05,
      "loss": 0.1135,
      "step": 3506
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4025473033463134,
      "learning_rate": 2.816981279160441e-05,
      "loss": 0.0857,
      "step": 3507
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.47481516705289056,
      "learning_rate": 2.8163318778533692e-05,
      "loss": 0.1109,
      "step": 3508
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3885089973883152,
      "learning_rate": 2.8156823732586525e-05,
      "loss": 0.0927,
      "step": 3509
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4640449263716323,
      "learning_rate": 2.81503276545847e-05,
      "loss": 0.1114,
      "step": 3510
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3766046405496704,
      "learning_rate": 2.8143830545350147e-05,
      "loss": 0.0893,
      "step": 3511
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4300228890714434,
      "learning_rate": 2.8137332405704922e-05,
      "loss": 0.0919,
      "step": 3512
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36432360752886406,
      "learning_rate": 2.8130833236471208e-05,
      "loss": 0.0949,
      "step": 3513
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.35784830389005967,
      "learning_rate": 2.812433303847133e-05,
      "loss": 0.0772,
      "step": 3514
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3346223704940486,
      "learning_rate": 2.8117831812527724e-05,
      "loss": 0.0732,
      "step": 3515
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.46472517472664593,
      "learning_rate": 2.811132955946298e-05,
      "loss": 0.099,
      "step": 3516
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4325094670158144,
      "learning_rate": 2.8104826280099796e-05,
      "loss": 0.1041,
      "step": 3517
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42947699852592836,
      "learning_rate": 2.8098321975261026e-05,
      "loss": 0.075,
      "step": 3518
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4099711772626729,
      "learning_rate": 2.809181664576961e-05,
      "loss": 0.0883,
      "step": 3519
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44197976992126387,
      "learning_rate": 2.8085310292448666e-05,
      "loss": 0.1066,
      "step": 3520
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.48493875490447325,
      "learning_rate": 2.807880291612141e-05,
      "loss": 0.1141,
      "step": 3521
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.48675934697264356,
      "learning_rate": 2.8072294517611208e-05,
      "loss": 0.0963,
      "step": 3522
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39768477002228564,
      "learning_rate": 2.8065785097741538e-05,
      "loss": 0.0885,
      "step": 3523
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3712355694069428,
      "learning_rate": 2.805927465733601e-05,
      "loss": 0.0895,
      "step": 3524
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.31483897827924573,
      "learning_rate": 2.8052763197218376e-05,
      "loss": 0.0643,
      "step": 3525
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3578252548543086,
      "learning_rate": 2.8046250718212507e-05,
      "loss": 0.0712,
      "step": 3526
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4630772017447761,
      "learning_rate": 2.8039737221142403e-05,
      "loss": 0.1017,
      "step": 3527
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3342574269376214,
      "learning_rate": 2.8033222706832187e-05,
      "loss": 0.0738,
      "step": 3528
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4917352960528285,
      "learning_rate": 2.802670717610613e-05,
      "loss": 0.1216,
      "step": 3529
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.40694144900278795,
      "learning_rate": 2.802019062978861e-05,
      "loss": 0.0919,
      "step": 3530
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3887436383576269,
      "learning_rate": 2.8013673068704156e-05,
      "loss": 0.0717,
      "step": 3531
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4490751617494209,
      "learning_rate": 2.80071544936774e-05,
      "loss": 0.0947,
      "step": 3532
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42147189767945803,
      "learning_rate": 2.8000634905533116e-05,
      "loss": 0.0826,
      "step": 3533
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3544186812765945,
      "learning_rate": 2.7994114305096208e-05,
      "loss": 0.087,
      "step": 3534
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36138494921103986,
      "learning_rate": 2.7987592693191707e-05,
      "loss": 0.0795,
      "step": 3535
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4038858397471632,
      "learning_rate": 2.7981070070644764e-05,
      "loss": 0.1127,
      "step": 3536
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4199915273697348,
      "learning_rate": 2.797454643828067e-05,
      "loss": 0.0908,
      "step": 3537
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.46442943028100175,
      "learning_rate": 2.7968021796924834e-05,
      "loss": 0.1366,
      "step": 3538
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4939540767231532,
      "learning_rate": 2.796149614740279e-05,
      "loss": 0.125,
      "step": 3539
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39145042946743575,
      "learning_rate": 2.7954969490540223e-05,
      "loss": 0.0788,
      "step": 3540
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4113374685474274,
      "learning_rate": 2.7948441827162917e-05,
      "loss": 0.0945,
      "step": 3541
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.47296766000602547,
      "learning_rate": 2.7941913158096792e-05,
      "loss": 0.09,
      "step": 3542
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.31737148558138323,
      "learning_rate": 2.7935383484167906e-05,
      "loss": 0.054,
      "step": 3543
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3961445521894762,
      "learning_rate": 2.7928852806202424e-05,
      "loss": 0.0755,
      "step": 3544
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3797827149928469,
      "learning_rate": 2.7922321125026667e-05,
      "loss": 0.0722,
      "step": 3545
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.48141437609090587,
      "learning_rate": 2.7915788441467052e-05,
      "loss": 0.0998,
      "step": 3546
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.45212200573643446,
      "learning_rate": 2.7909254756350143e-05,
      "loss": 0.0851,
      "step": 3547
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4762996667960886,
      "learning_rate": 2.790272007050262e-05,
      "loss": 0.0996,
      "step": 3548
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3814486712939259,
      "learning_rate": 2.78961843847513e-05,
      "loss": 0.0875,
      "step": 3549
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4990424285081931,
      "learning_rate": 2.7889647699923114e-05,
      "loss": 0.1049,
      "step": 3550
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3949896538260701,
      "learning_rate": 2.788311001684514e-05,
      "loss": 0.0889,
      "step": 3551
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3208288125343036,
      "learning_rate": 2.7876571336344546e-05,
      "loss": 0.053,
      "step": 3552
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3813343119026568,
      "learning_rate": 2.7870031659248674e-05,
      "loss": 0.0964,
      "step": 3553
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4067905571031011,
      "learning_rate": 2.7863490986384945e-05,
      "loss": 0.0893,
      "step": 3554
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3837801905569949,
      "learning_rate": 2.7856949318580938e-05,
      "loss": 0.0824,
      "step": 3555
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6595342742048063,
      "learning_rate": 2.7850406656664346e-05,
      "loss": 0.1266,
      "step": 3556
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4127501437315757,
      "learning_rate": 2.784386300146299e-05,
      "loss": 0.0852,
      "step": 3557
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.49301509306845137,
      "learning_rate": 2.783731835380482e-05,
      "loss": 0.1132,
      "step": 3558
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3952460659128091,
      "learning_rate": 2.7830772714517897e-05,
      "loss": 0.0827,
      "step": 3559
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3830156176684609,
      "learning_rate": 2.782422608443043e-05,
      "loss": 0.0728,
      "step": 3560
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3346835620053629,
      "learning_rate": 2.7817678464370725e-05,
      "loss": 0.0653,
      "step": 3561
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42750299196144387,
      "learning_rate": 2.781112985516725e-05,
      "loss": 0.0948,
      "step": 3562
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4892119121530535,
      "learning_rate": 2.7804580257648564e-05,
      "loss": 0.117,
      "step": 3563
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4595612816464375,
      "learning_rate": 2.7798029672643375e-05,
      "loss": 0.0974,
      "step": 3564
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4478804680886803,
      "learning_rate": 2.7791478100980486e-05,
      "loss": 0.1058,
      "step": 3565
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37482116329046383,
      "learning_rate": 2.778492554348887e-05,
      "loss": 0.0794,
      "step": 3566
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.45620239937049717,
      "learning_rate": 2.7778372000997583e-05,
      "loss": 0.0849,
      "step": 3567
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4994403568129309,
      "learning_rate": 2.7771817474335835e-05,
      "loss": 0.1245,
      "step": 3568
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4417694083258809,
      "learning_rate": 2.7765261964332926e-05,
      "loss": 0.1028,
      "step": 3569
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3957806435033321,
      "learning_rate": 2.7758705471818327e-05,
      "loss": 0.0684,
      "step": 3570
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38470979469454475,
      "learning_rate": 2.775214799762159e-05,
      "loss": 0.076,
      "step": 3571
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3069180005512638,
      "learning_rate": 2.7745589542572424e-05,
      "loss": 0.0798,
      "step": 3572
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3998921152001471,
      "learning_rate": 2.773903010750063e-05,
      "loss": 0.0858,
      "step": 3573
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3444286280101821,
      "learning_rate": 2.7732469693236166e-05,
      "loss": 0.0809,
      "step": 3574
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4927372964917862,
      "learning_rate": 2.772590830060909e-05,
      "loss": 0.123,
      "step": 3575
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4993873956070473,
      "learning_rate": 2.77193459304496e-05,
      "loss": 0.1088,
      "step": 3576
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4169589884851231,
      "learning_rate": 2.7712782583588002e-05,
      "loss": 0.0635,
      "step": 3577
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3985956663461733,
      "learning_rate": 2.7706218260854738e-05,
      "loss": 0.0815,
      "step": 3578
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4176896491434237,
      "learning_rate": 2.7699652963080365e-05,
      "loss": 0.1026,
      "step": 3579
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4340718528832963,
      "learning_rate": 2.7693086691095573e-05,
      "loss": 0.0828,
      "step": 3580
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44624847628229125,
      "learning_rate": 2.7686519445731165e-05,
      "loss": 0.1099,
      "step": 3581
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38683835501590985,
      "learning_rate": 2.767995122781807e-05,
      "loss": 0.0821,
      "step": 3582
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3828269828425894,
      "learning_rate": 2.7673382038187355e-05,
      "loss": 0.1727,
      "step": 3583
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4187759021320502,
      "learning_rate": 2.7666811877670177e-05,
      "loss": 0.0807,
      "step": 3584
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4394913211753281,
      "learning_rate": 2.7660240747097858e-05,
      "loss": 0.0591,
      "step": 3585
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3907763760036637,
      "learning_rate": 2.7653668647301797e-05,
      "loss": 0.0878,
      "step": 3586
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37151942684418104,
      "learning_rate": 2.7647095579113554e-05,
      "loss": 0.0751,
      "step": 3587
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3795652263285278,
      "learning_rate": 2.7640521543364797e-05,
      "loss": 0.0962,
      "step": 3588
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.45918853764217504,
      "learning_rate": 2.7633946540887307e-05,
      "loss": 0.1022,
      "step": 3589
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3847690717668627,
      "learning_rate": 2.7627370572513005e-05,
      "loss": 0.0742,
      "step": 3590
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.450988980168159,
      "learning_rate": 2.7620793639073917e-05,
      "loss": 0.0949,
      "step": 3591
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4345922228829612,
      "learning_rate": 2.7614215741402204e-05,
      "loss": 0.1069,
      "step": 3592
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.453537227303409,
      "learning_rate": 2.760763688033015e-05,
      "loss": 0.086,
      "step": 3593
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3166807302794481,
      "learning_rate": 2.7601057056690148e-05,
      "loss": 0.0603,
      "step": 3594
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42196853801657275,
      "learning_rate": 2.7594476271314725e-05,
      "loss": 0.0799,
      "step": 3595
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4255819816808162,
      "learning_rate": 2.7587894525036517e-05,
      "loss": 0.0964,
      "step": 3596
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35040990719000464,
      "learning_rate": 2.75813118186883e-05,
      "loss": 0.0596,
      "step": 3597
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43195369638317926,
      "learning_rate": 2.7574728153102956e-05,
      "loss": 0.097,
      "step": 3598
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37971084742720335,
      "learning_rate": 2.756814352911349e-05,
      "loss": 0.0845,
      "step": 3599
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39909323792756657,
      "learning_rate": 2.7561557947553037e-05,
      "loss": 0.0783,
      "step": 3600
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36411122908711735,
      "learning_rate": 2.7554971409254848e-05,
      "loss": 0.0633,
      "step": 3601
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4875586388851573,
      "learning_rate": 2.7548383915052287e-05,
      "loss": 0.1089,
      "step": 3602
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.47123748079664823,
      "learning_rate": 2.7541795465778865e-05,
      "loss": 0.1105,
      "step": 3603
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3795529971089713,
      "learning_rate": 2.7535206062268174e-05,
      "loss": 0.0914,
      "step": 3604
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4340650728944526,
      "learning_rate": 2.752861570535396e-05,
      "loss": 0.0842,
      "step": 3605
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4889310861328,
      "learning_rate": 2.7522024395870075e-05,
      "loss": 0.1004,
      "step": 3606
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.48759925516555136,
      "learning_rate": 2.7515432134650502e-05,
      "loss": 0.0932,
      "step": 3607
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4937229124920279,
      "learning_rate": 2.7508838922529316e-05,
      "loss": 0.1019,
      "step": 3608
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3846958116428314,
      "learning_rate": 2.750224476034076e-05,
      "loss": 0.0907,
      "step": 3609
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38047751821953774,
      "learning_rate": 2.7495649648919153e-05,
      "loss": 0.0889,
      "step": 3610
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.49451134085692566,
      "learning_rate": 2.7489053589098966e-05,
      "loss": 0.1178,
      "step": 3611
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43259050962974316,
      "learning_rate": 2.7482456581714757e-05,
      "loss": 0.1086,
      "step": 3612
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.394322519513368,
      "learning_rate": 2.747585862760124e-05,
      "loss": 0.0902,
      "step": 3613
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5170476246025132,
      "learning_rate": 2.7469259727593213e-05,
      "loss": 0.113,
      "step": 3614
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39153336765941504,
      "learning_rate": 2.746265988252563e-05,
      "loss": 0.0924,
      "step": 3615
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3767012510661694,
      "learning_rate": 2.7456059093233537e-05,
      "loss": 0.0764,
      "step": 3616
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44122173875317494,
      "learning_rate": 2.7449457360552108e-05,
      "loss": 0.1075,
      "step": 3617
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39906721281108154,
      "learning_rate": 2.7442854685316643e-05,
      "loss": 0.1019,
      "step": 3618
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38799180385201243,
      "learning_rate": 2.7436251068362555e-05,
      "loss": 0.066,
      "step": 3619
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33345881598640764,
      "learning_rate": 2.7429646510525373e-05,
      "loss": 0.0645,
      "step": 3620
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.367687959709963,
      "learning_rate": 2.7423041012640744e-05,
      "loss": 0.0933,
      "step": 3621
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3662568940890612,
      "learning_rate": 2.7416434575544455e-05,
      "loss": 0.0625,
      "step": 3622
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.448969664139347,
      "learning_rate": 2.7409827200072377e-05,
      "loss": 0.1064,
      "step": 3623
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42573111352676835,
      "learning_rate": 2.7403218887060538e-05,
      "loss": 0.0862,
      "step": 3624
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43705566867206613,
      "learning_rate": 2.7396609637345038e-05,
      "loss": 0.0893,
      "step": 3625
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36062537234599495,
      "learning_rate": 2.738999945176215e-05,
      "loss": 0.074,
      "step": 3626
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4499766549041885,
      "learning_rate": 2.7383388331148225e-05,
      "loss": 0.0996,
      "step": 3627
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44231468164461946,
      "learning_rate": 2.7376776276339745e-05,
      "loss": 0.093,
      "step": 3628
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3355116584780354,
      "learning_rate": 2.7370163288173314e-05,
      "loss": 0.0829,
      "step": 3629
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3392438636420186,
      "learning_rate": 2.7363549367485648e-05,
      "loss": 0.0623,
      "step": 3630
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39994563985041476,
      "learning_rate": 2.7356934515113582e-05,
      "loss": 0.0855,
      "step": 3631
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.47382107168861803,
      "learning_rate": 2.7350318731894075e-05,
      "loss": 0.113,
      "step": 3632
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35320911502800384,
      "learning_rate": 2.7343702018664194e-05,
      "loss": 0.0683,
      "step": 3633
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.34911172494121734,
      "learning_rate": 2.7337084376261135e-05,
      "loss": 0.0835,
      "step": 3634
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44526797624162595,
      "learning_rate": 2.7330465805522196e-05,
      "loss": 0.0881,
      "step": 3635
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39307455682300707,
      "learning_rate": 2.7323846307284814e-05,
      "loss": 0.0761,
      "step": 3636
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35096955629381305,
      "learning_rate": 2.731722588238652e-05,
      "loss": 0.0749,
      "step": 3637
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3916129208104275,
      "learning_rate": 2.7310604531664983e-05,
      "loss": 0.0915,
      "step": 3638
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35182089316771875,
      "learning_rate": 2.7303982255957967e-05,
      "loss": 0.0781,
      "step": 3639
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3938062067635411,
      "learning_rate": 2.7297359056103378e-05,
      "loss": 0.0937,
      "step": 3640
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3947547996874248,
      "learning_rate": 2.729073493293922e-05,
      "loss": 0.0682,
      "step": 3641
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37419170449148553,
      "learning_rate": 2.7284109887303628e-05,
      "loss": 0.0972,
      "step": 3642
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5598878342404536,
      "learning_rate": 2.7277483920034832e-05,
      "loss": 0.1632,
      "step": 3643
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39484498797072787,
      "learning_rate": 2.7270857031971203e-05,
      "loss": 0.075,
      "step": 3644
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43548780305843615,
      "learning_rate": 2.7264229223951217e-05,
      "loss": 0.0971,
      "step": 3645
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5114433479315054,
      "learning_rate": 2.7257600496813475e-05,
      "loss": 0.1145,
      "step": 3646
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.477165918892703,
      "learning_rate": 2.725097085139667e-05,
      "loss": 0.1083,
      "step": 3647
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3972147148466966,
      "learning_rate": 2.7244340288539638e-05,
      "loss": 0.0736,
      "step": 3648
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36628622679839784,
      "learning_rate": 2.723770880908132e-05,
      "loss": 0.0664,
      "step": 3649
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4688244283715075,
      "learning_rate": 2.7231076413860774e-05,
      "loss": 0.0961,
      "step": 3650
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3720494462217135,
      "learning_rate": 2.7224443103717173e-05,
      "loss": 0.0569,
      "step": 3651
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33484646590035255,
      "learning_rate": 2.721780887948981e-05,
      "loss": 0.0658,
      "step": 3652
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42390671454072687,
      "learning_rate": 2.7211173742018088e-05,
      "loss": 0.0882,
      "step": 3653
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3600978630219675,
      "learning_rate": 2.7204537692141526e-05,
      "loss": 0.0981,
      "step": 3654
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39346598342554273,
      "learning_rate": 2.7197900730699768e-05,
      "loss": 0.0791,
      "step": 3655
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4118791814137419,
      "learning_rate": 2.7191262858532552e-05,
      "loss": 0.0897,
      "step": 3656
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5253493282155162,
      "learning_rate": 2.7184624076479763e-05,
      "loss": 0.1185,
      "step": 3657
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3058321798226607,
      "learning_rate": 2.7177984385381366e-05,
      "loss": 0.0671,
      "step": 3658
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.41925439348233295,
      "learning_rate": 2.7171343786077468e-05,
      "loss": 0.0879,
      "step": 3659
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44979976729212706,
      "learning_rate": 2.7164702279408275e-05,
      "loss": 0.0756,
      "step": 3660
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6790933752786962,
      "learning_rate": 2.7158059866214118e-05,
      "loss": 0.08,
      "step": 3661
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.34682940621622094,
      "learning_rate": 2.715141654733544e-05,
      "loss": 0.0672,
      "step": 3662
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4167402520369813,
      "learning_rate": 2.7144772323612798e-05,
      "loss": 0.1162,
      "step": 3663
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4755558889558451,
      "learning_rate": 2.7138127195886856e-05,
      "loss": 0.1089,
      "step": 3664
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4554168030313123,
      "learning_rate": 2.7131481164998407e-05,
      "loss": 0.0773,
      "step": 3665
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3734506748381317,
      "learning_rate": 2.712483423178834e-05,
      "loss": 0.0866,
      "step": 3666
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33750662179031643,
      "learning_rate": 2.711818639709768e-05,
      "loss": 0.065,
      "step": 3667
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33147065577728335,
      "learning_rate": 2.7111537661767537e-05,
      "loss": 0.0691,
      "step": 3668
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4368118937459918,
      "learning_rate": 2.7104888026639175e-05,
      "loss": 0.0777,
      "step": 3669
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4076485402908683,
      "learning_rate": 2.7098237492553937e-05,
      "loss": 0.0809,
      "step": 3670
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4115515273419336,
      "learning_rate": 2.7091586060353298e-05,
      "loss": 0.1022,
      "step": 3671
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3655827946226697,
      "learning_rate": 2.7084933730878824e-05,
      "loss": 0.0868,
      "step": 3672
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3714646142623628,
      "learning_rate": 2.7078280504972237e-05,
      "loss": 0.074,
      "step": 3673
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43147825406312484,
      "learning_rate": 2.7071626383475327e-05,
      "loss": 0.1109,
      "step": 3674
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3604714212372367,
      "learning_rate": 2.7064971367230023e-05,
      "loss": 0.08,
      "step": 3675
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3760598622971818,
      "learning_rate": 2.7058315457078358e-05,
      "loss": 0.088,
      "step": 3676
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3429940863876655,
      "learning_rate": 2.7051658653862488e-05,
      "loss": 0.0647,
      "step": 3677
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38157090455092857,
      "learning_rate": 2.7045000958424674e-05,
      "loss": 0.0799,
      "step": 3678
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.41507291767940063,
      "learning_rate": 2.7038342371607282e-05,
      "loss": 0.1151,
      "step": 3679
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.48991045946299666,
      "learning_rate": 2.7031682894252816e-05,
      "loss": 0.1091,
      "step": 3680
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.402193856326176,
      "learning_rate": 2.702502252720386e-05,
      "loss": 0.0845,
      "step": 3681
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33391212217230803,
      "learning_rate": 2.701836127130314e-05,
      "loss": 0.0732,
      "step": 3682
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3626760548546362,
      "learning_rate": 2.7011699127393476e-05,
      "loss": 0.0528,
      "step": 3683
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4059666232145197,
      "learning_rate": 2.7005036096317802e-05,
      "loss": 0.1133,
      "step": 3684
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3472857161589751,
      "learning_rate": 2.6998372178919175e-05,
      "loss": 0.0752,
      "step": 3685
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3599597071907598,
      "learning_rate": 2.6991707376040755e-05,
      "loss": 0.0814,
      "step": 3686
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.48472505656375875,
      "learning_rate": 2.6985041688525814e-05,
      "loss": 0.107,
      "step": 3687
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41989417761013553,
      "learning_rate": 2.6978375117217743e-05,
      "loss": 0.078,
      "step": 3688
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3536485243045561,
      "learning_rate": 2.6971707662960037e-05,
      "loss": 0.0756,
      "step": 3689
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4151581896189963,
      "learning_rate": 2.696503932659631e-05,
      "loss": 0.089,
      "step": 3690
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.395437804254397,
      "learning_rate": 2.6958370108970274e-05,
      "loss": 0.0812,
      "step": 3691
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2876587108318242,
      "learning_rate": 2.6951700010925774e-05,
      "loss": 0.0573,
      "step": 3692
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42486186026456246,
      "learning_rate": 2.6945029033306745e-05,
      "loss": 0.0846,
      "step": 3693
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7010630573638799,
      "learning_rate": 2.6938357176957243e-05,
      "loss": 0.1485,
      "step": 3694
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.44829336302258704,
      "learning_rate": 2.6931684442721443e-05,
      "loss": 0.0744,
      "step": 3695
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3989750097522774,
      "learning_rate": 2.692501083144362e-05,
      "loss": 0.0732,
      "step": 3696
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34813477233648726,
      "learning_rate": 2.6918336343968158e-05,
      "loss": 0.0653,
      "step": 3697
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4174504497956514,
      "learning_rate": 2.6911660981139563e-05,
      "loss": 0.0825,
      "step": 3698
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4243495388286499,
      "learning_rate": 2.6904984743802437e-05,
      "loss": 0.0782,
      "step": 3699
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41008554484972526,
      "learning_rate": 2.6898307632801515e-05,
      "loss": 0.0785,
      "step": 3700
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3082796985569781,
      "learning_rate": 2.6891629648981612e-05,
      "loss": 0.0539,
      "step": 3701
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.36285927578599037,
      "learning_rate": 2.6884950793187684e-05,
      "loss": 0.0773,
      "step": 3702
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4554967959947677,
      "learning_rate": 2.6878271066264777e-05,
      "loss": 0.1007,
      "step": 3703
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42337043062125784,
      "learning_rate": 2.6871590469058052e-05,
      "loss": 0.0798,
      "step": 3704
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4658530361749296,
      "learning_rate": 2.6864909002412782e-05,
      "loss": 0.0829,
      "step": 3705
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4741350480293947,
      "learning_rate": 2.6858226667174362e-05,
      "loss": 0.1136,
      "step": 3706
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.47693753728615235,
      "learning_rate": 2.6851543464188265e-05,
      "loss": 0.0843,
      "step": 3707
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4464780711932984,
      "learning_rate": 2.684485939430011e-05,
      "loss": 0.0832,
      "step": 3708
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.32607685920900675,
      "learning_rate": 2.68381744583556e-05,
      "loss": 0.0843,
      "step": 3709
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7015489504722764,
      "learning_rate": 2.683148865720056e-05,
      "loss": 0.1591,
      "step": 3710
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4136532074171647,
      "learning_rate": 2.682480199168092e-05,
      "loss": 0.0766,
      "step": 3711
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37340939467901163,
      "learning_rate": 2.6818114462642726e-05,
      "loss": 0.0596,
      "step": 3712
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.404133196883607,
      "learning_rate": 2.6811426070932118e-05,
      "loss": 0.0749,
      "step": 3713
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4182063167342618,
      "learning_rate": 2.6804736817395362e-05,
      "loss": 0.0812,
      "step": 3714
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.55612289712465,
      "learning_rate": 2.679804670287883e-05,
      "loss": 0.1685,
      "step": 3715
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.385284983417369,
      "learning_rate": 2.6791355728228986e-05,
      "loss": 0.073,
      "step": 3716
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.36111051863026716,
      "learning_rate": 2.6784663894292432e-05,
      "loss": 0.083,
      "step": 3717
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4024285532642165,
      "learning_rate": 2.6777971201915843e-05,
      "loss": 0.0825,
      "step": 3718
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4460402949385001,
      "learning_rate": 2.6771277651946043e-05,
      "loss": 0.099,
      "step": 3719
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3694163710441471,
      "learning_rate": 2.676458324522992e-05,
      "loss": 0.0856,
      "step": 3720
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43341904634605877,
      "learning_rate": 2.675788798261452e-05,
      "loss": 0.1087,
      "step": 3721
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43076734969411545,
      "learning_rate": 2.675119186494696e-05,
      "loss": 0.1132,
      "step": 3722
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43265528112918133,
      "learning_rate": 2.674449489307447e-05,
      "loss": 0.1019,
      "step": 3723
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41506430156485086,
      "learning_rate": 2.6737797067844403e-05,
      "loss": 0.0764,
      "step": 3724
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.442401927678175,
      "learning_rate": 2.6731098390104212e-05,
      "loss": 0.1004,
      "step": 3725
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.47862447940761244,
      "learning_rate": 2.6724398860701453e-05,
      "loss": 0.1428,
      "step": 3726
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3341419896545442,
      "learning_rate": 2.6717698480483794e-05,
      "loss": 0.0727,
      "step": 3727
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41081486456692684,
      "learning_rate": 2.6710997250299012e-05,
      "loss": 0.0774,
      "step": 3728
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34533978350119077,
      "learning_rate": 2.670429517099499e-05,
      "loss": 0.0693,
      "step": 3729
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4583082535413163,
      "learning_rate": 2.6697592243419723e-05,
      "loss": 0.0753,
      "step": 3730
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.30079207462797986,
      "learning_rate": 2.669088846842131e-05,
      "loss": 0.0623,
      "step": 3731
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3539198992793908,
      "learning_rate": 2.668418384684795e-05,
      "loss": 0.0869,
      "step": 3732
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.36161592723463565,
      "learning_rate": 2.667747837954796e-05,
      "loss": 0.0859,
      "step": 3733
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4716474312756721,
      "learning_rate": 2.6670772067369754e-05,
      "loss": 0.0894,
      "step": 3734
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37577267357443184,
      "learning_rate": 2.6664064911161865e-05,
      "loss": 0.0764,
      "step": 3735
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.35757159656568394,
      "learning_rate": 2.6657356911772922e-05,
      "loss": 0.0763,
      "step": 3736
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3981749286514307,
      "learning_rate": 2.665064807005166e-05,
      "loss": 0.0707,
      "step": 3737
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34343839740722704,
      "learning_rate": 2.6643938386846945e-05,
      "loss": 0.0684,
      "step": 3738
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3895658247714165,
      "learning_rate": 2.6637227863007708e-05,
      "loss": 0.0745,
      "step": 3739
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.395385195716134,
      "learning_rate": 2.663051649938303e-05,
      "loss": 0.1103,
      "step": 3740
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.556052502921623,
      "learning_rate": 2.6623804296822047e-05,
      "loss": 0.1137,
      "step": 3741
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4019299017571043,
      "learning_rate": 2.6617091256174058e-05,
      "loss": 0.0755,
      "step": 3742
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42738327875679344,
      "learning_rate": 2.6610377378288427e-05,
      "loss": 0.1077,
      "step": 3743
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41713783525742254,
      "learning_rate": 2.6603662664014644e-05,
      "loss": 0.0797,
      "step": 3744
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.35580642088456604,
      "learning_rate": 2.6596947114202292e-05,
      "loss": 0.0965,
      "step": 3745
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3278912665447424,
      "learning_rate": 2.659023072970107e-05,
      "loss": 0.0846,
      "step": 3746
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43495188901933723,
      "learning_rate": 2.658351351136078e-05,
      "loss": 0.1064,
      "step": 3747
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.396502903096658,
      "learning_rate": 2.657679546003133e-05,
      "loss": 0.0887,
      "step": 3748
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3620310775894723,
      "learning_rate": 2.6570076576562726e-05,
      "loss": 0.0608,
      "step": 3749
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.364516894780986,
      "learning_rate": 2.656335686180509e-05,
      "loss": 0.0711,
      "step": 3750
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.36470807757013285,
      "learning_rate": 2.6556636316608638e-05,
      "loss": 0.0618,
      "step": 3751
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37263699927680105,
      "learning_rate": 2.6549914941823713e-05,
      "loss": 0.1033,
      "step": 3752
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.36668609311198197,
      "learning_rate": 2.6543192738300722e-05,
      "loss": 0.1009,
      "step": 3753
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3415438996509172,
      "learning_rate": 2.6536469706890226e-05,
      "loss": 0.0915,
      "step": 3754
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4262886873290953,
      "learning_rate": 2.6529745848442848e-05,
      "loss": 0.1118,
      "step": 3755
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4442084160889862,
      "learning_rate": 2.652302116380935e-05,
      "loss": 0.0868,
      "step": 3756
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3483900940297324,
      "learning_rate": 2.651629565384057e-05,
      "loss": 0.0897,
      "step": 3757
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41044330542745466,
      "learning_rate": 2.6509569319387477e-05,
      "loss": 0.0887,
      "step": 3758
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40089516833237204,
      "learning_rate": 2.6502842161301122e-05,
      "loss": 0.0638,
      "step": 3759
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3713163009677597,
      "learning_rate": 2.6496114180432672e-05,
      "loss": 0.0808,
      "step": 3760
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4287045072103712,
      "learning_rate": 2.6489385377633386e-05,
      "loss": 0.0985,
      "step": 3761
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42671047969529485,
      "learning_rate": 2.6482655753754657e-05,
      "loss": 0.089,
      "step": 3762
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5120856738625476,
      "learning_rate": 2.647592530964793e-05,
      "loss": 0.1093,
      "step": 3763
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4575059658714247,
      "learning_rate": 2.6469194046164818e-05,
      "loss": 0.0993,
      "step": 3764
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.29188314845128427,
      "learning_rate": 2.6462461964156987e-05,
      "loss": 0.066,
      "step": 3765
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34060030068957126,
      "learning_rate": 2.6455729064476227e-05,
      "loss": 0.0856,
      "step": 3766
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4166301969621244,
      "learning_rate": 2.6448995347974423e-05,
      "loss": 0.0729,
      "step": 3767
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.39070079068205676,
      "learning_rate": 2.6442260815503575e-05,
      "loss": 0.092,
      "step": 3768
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31346189755751785,
      "learning_rate": 2.6435525467915775e-05,
      "loss": 0.0607,
      "step": 3769
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.46441492769011705,
      "learning_rate": 2.6428789306063233e-05,
      "loss": 0.0963,
      "step": 3770
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.33865248504711404,
      "learning_rate": 2.6422052330798246e-05,
      "loss": 0.0783,
      "step": 3771
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31178134299341204,
      "learning_rate": 2.6415314542973214e-05,
      "loss": 0.0598,
      "step": 3772
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3745344172516076,
      "learning_rate": 2.6408575943440663e-05,
      "loss": 0.0859,
      "step": 3773
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.33082243838175024,
      "learning_rate": 2.6401836533053186e-05,
      "loss": 0.0788,
      "step": 3774
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.49266325950777917,
      "learning_rate": 2.6395096312663503e-05,
      "loss": 0.1089,
      "step": 3775
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5309467852826272,
      "learning_rate": 2.6388355283124435e-05,
      "loss": 0.0993,
      "step": 3776
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38705393204583444,
      "learning_rate": 2.6381613445288904e-05,
      "loss": 0.0942,
      "step": 3777
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.39752438575390575,
      "learning_rate": 2.637487080000992e-05,
      "loss": 0.0879,
      "step": 3778
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.435103888246342,
      "learning_rate": 2.6368127348140617e-05,
      "loss": 0.0761,
      "step": 3779
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3783821335098992,
      "learning_rate": 2.636138309053421e-05,
      "loss": 0.0614,
      "step": 3780
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5078508215896558,
      "learning_rate": 2.6354638028044042e-05,
      "loss": 0.0948,
      "step": 3781
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.34976695680993125,
      "learning_rate": 2.634789216152353e-05,
      "loss": 0.0645,
      "step": 3782
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4342542720676777,
      "learning_rate": 2.634114549182621e-05,
      "loss": 0.1061,
      "step": 3783
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4177716583741957,
      "learning_rate": 2.63343980198057e-05,
      "loss": 0.0839,
      "step": 3784
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.33537022800299376,
      "learning_rate": 2.6327649746315765e-05,
      "loss": 0.0806,
      "step": 3785
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3686832642377394,
      "learning_rate": 2.6320900672210216e-05,
      "loss": 0.0631,
      "step": 3786
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.35588485341823284,
      "learning_rate": 2.6314150798343e-05,
      "loss": 0.0804,
      "step": 3787
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4261561676683239,
      "learning_rate": 2.6307400125568147e-05,
      "loss": 0.0799,
      "step": 3788
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.597507009385306,
      "learning_rate": 2.6300648654739807e-05,
      "loss": 0.1348,
      "step": 3789
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4175857678847839,
      "learning_rate": 2.629389638671221e-05,
      "loss": 0.0712,
      "step": 3790
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40433647588384997,
      "learning_rate": 2.6287143322339708e-05,
      "loss": 0.103,
      "step": 3791
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.472471500576694,
      "learning_rate": 2.6280389462476733e-05,
      "loss": 0.0878,
      "step": 3792
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3449964142523197,
      "learning_rate": 2.6273634807977835e-05,
      "loss": 0.057,
      "step": 3793
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4295187879012468,
      "learning_rate": 2.6266879359697647e-05,
      "loss": 0.0954,
      "step": 3794
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3290037015312466,
      "learning_rate": 2.6260123118490923e-05,
      "loss": 0.088,
      "step": 3795
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3803637483056497,
      "learning_rate": 2.6253366085212503e-05,
      "loss": 0.0841,
      "step": 3796
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.33345049299185653,
      "learning_rate": 2.624660826071733e-05,
      "loss": 0.079,
      "step": 3797
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6368027146097052,
      "learning_rate": 2.6239849645860447e-05,
      "loss": 0.1143,
      "step": 3798
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.44586020801627463,
      "learning_rate": 2.6233090241497002e-05,
      "loss": 0.1039,
      "step": 3799
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.34690206138625723,
      "learning_rate": 2.6226330048482233e-05,
      "loss": 0.0628,
      "step": 3800
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.502676327784346,
      "learning_rate": 2.6219569067671492e-05,
      "loss": 0.1143,
      "step": 3801
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.32050167870565444,
      "learning_rate": 2.6212807299920218e-05,
      "loss": 0.0525,
      "step": 3802
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4719324796922636,
      "learning_rate": 2.620604474608395e-05,
      "loss": 0.114,
      "step": 3803
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.31918097886414964,
      "learning_rate": 2.6199281407018338e-05,
      "loss": 0.049,
      "step": 3804
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3494760269653155,
      "learning_rate": 2.6192517283579123e-05,
      "loss": 0.0743,
      "step": 3805
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3240728164767029,
      "learning_rate": 2.618575237662214e-05,
      "loss": 0.0634,
      "step": 3806
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3565976588093727,
      "learning_rate": 2.617898668700333e-05,
      "loss": 0.0686,
      "step": 3807
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38766959847374577,
      "learning_rate": 2.6172220215578743e-05,
      "loss": 0.0752,
      "step": 3808
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4119011411634392,
      "learning_rate": 2.6165452963204506e-05,
      "loss": 0.1115,
      "step": 3809
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38317223598408934,
      "learning_rate": 2.615868493073686e-05,
      "loss": 0.0786,
      "step": 3810
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27981722982421026,
      "learning_rate": 2.615191611903214e-05,
      "loss": 0.0609,
      "step": 3811
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3477730608375392,
      "learning_rate": 2.614514652894678e-05,
      "loss": 0.0709,
      "step": 3812
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5658702550461188,
      "learning_rate": 2.613837616133731e-05,
      "loss": 0.1093,
      "step": 3813
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3478805911135369,
      "learning_rate": 2.613160501706037e-05,
      "loss": 0.0701,
      "step": 3814
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37509291213098606,
      "learning_rate": 2.6124833096972677e-05,
      "loss": 0.0725,
      "step": 3815
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2818346833791343,
      "learning_rate": 2.6118060401931073e-05,
      "loss": 0.0612,
      "step": 3816
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38897457508093614,
      "learning_rate": 2.611128693279247e-05,
      "loss": 0.0706,
      "step": 3817
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5441346857605304,
      "learning_rate": 2.6104512690413906e-05,
      "loss": 0.1419,
      "step": 3818
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4475984958854371,
      "learning_rate": 2.6097737675652487e-05,
      "loss": 0.0892,
      "step": 3819
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.42761378569786174,
      "learning_rate": 2.609096188936544e-05,
      "loss": 0.0782,
      "step": 3820
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36967465962854595,
      "learning_rate": 2.6084185332410083e-05,
      "loss": 0.0803,
      "step": 3821
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.689075139420687,
      "learning_rate": 2.607740800564383e-05,
      "loss": 0.1065,
      "step": 3822
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4114882864692848,
      "learning_rate": 2.6070629909924185e-05,
      "loss": 0.0761,
      "step": 3823
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.42522050522698307,
      "learning_rate": 2.6063851046108766e-05,
      "loss": 0.1191,
      "step": 3824
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37927305026909475,
      "learning_rate": 2.6057071415055274e-05,
      "loss": 0.0932,
      "step": 3825
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5835468605334241,
      "learning_rate": 2.605029101762152e-05,
      "loss": 0.1433,
      "step": 3826
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37207057282045725,
      "learning_rate": 2.604350985466539e-05,
      "loss": 0.0851,
      "step": 3827
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3467701860965927,
      "learning_rate": 2.6036727927044897e-05,
      "loss": 0.0813,
      "step": 3828
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4110846230723617,
      "learning_rate": 2.6029945235618124e-05,
      "loss": 0.0853,
      "step": 3829
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.32372267815837463,
      "learning_rate": 2.602316178124327e-05,
      "loss": 0.0623,
      "step": 3830
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3964777602973468,
      "learning_rate": 2.6016377564778612e-05,
      "loss": 0.0718,
      "step": 3831
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4110845868239569,
      "learning_rate": 2.6009592587082538e-05,
      "loss": 0.0867,
      "step": 3832
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.388139369532742,
      "learning_rate": 2.6002806849013535e-05,
      "loss": 0.0865,
      "step": 3833
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37089527243277676,
      "learning_rate": 2.5996020351430163e-05,
      "loss": 0.0884,
      "step": 3834
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3232354800170887,
      "learning_rate": 2.598923309519111e-05,
      "loss": 0.0937,
      "step": 3835
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4911285998765178,
      "learning_rate": 2.598244508115513e-05,
      "loss": 0.1024,
      "step": 3836
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3604990959539731,
      "learning_rate": 2.5975656310181103e-05,
      "loss": 0.0909,
      "step": 3837
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3541039897585125,
      "learning_rate": 2.596886678312797e-05,
      "loss": 0.0724,
      "step": 3838
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5309020473685049,
      "learning_rate": 2.5962076500854804e-05,
      "loss": 0.1191,
      "step": 3839
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4487732212094087,
      "learning_rate": 2.5955285464220738e-05,
      "loss": 0.1102,
      "step": 3840
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3782163021106454,
      "learning_rate": 2.5948493674085037e-05,
      "loss": 0.0765,
      "step": 3841
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.41965018065998755,
      "learning_rate": 2.594170113130703e-05,
      "loss": 0.0857,
      "step": 3842
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3497844875516106,
      "learning_rate": 2.5934907836746163e-05,
      "loss": 0.055,
      "step": 3843
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3404159989616977,
      "learning_rate": 2.5928113791261952e-05,
      "loss": 0.0573,
      "step": 3844
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4448014723555292,
      "learning_rate": 2.5921318995714044e-05,
      "loss": 0.0887,
      "step": 3845
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.35883558259336706,
      "learning_rate": 2.5914523450962147e-05,
      "loss": 0.0742,
      "step": 3846
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3590736784493898,
      "learning_rate": 2.590772715786609e-05,
      "loss": 0.0705,
      "step": 3847
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4567968305310795,
      "learning_rate": 2.590093011728577e-05,
      "loss": 0.1125,
      "step": 3848
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.29760697124594165,
      "learning_rate": 2.5894132330081205e-05,
      "loss": 0.0494,
      "step": 3849
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39466525153293547,
      "learning_rate": 2.588733379711248e-05,
      "loss": 0.0771,
      "step": 3850
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3584027199138524,
      "learning_rate": 2.588053451923981e-05,
      "loss": 0.0608,
      "step": 3851
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.43275859235390585,
      "learning_rate": 2.587373449732347e-05,
      "loss": 0.0755,
      "step": 3852
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4936958844672623,
      "learning_rate": 2.586693373222385e-05,
      "loss": 0.0934,
      "step": 3853
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.33962295199700926,
      "learning_rate": 2.5860132224801424e-05,
      "loss": 0.0641,
      "step": 3854
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.35557506354818935,
      "learning_rate": 2.5853329975916767e-05,
      "loss": 0.063,
      "step": 3855
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.44622992660082794,
      "learning_rate": 2.584652698643054e-05,
      "loss": 0.0846,
      "step": 3856
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3580092309901829,
      "learning_rate": 2.58397232572035e-05,
      "loss": 0.0455,
      "step": 3857
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.362799581516218,
      "learning_rate": 2.58329187890965e-05,
      "loss": 0.064,
      "step": 3858
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3767330930836038,
      "learning_rate": 2.5826113582970493e-05,
      "loss": 0.0745,
      "step": 3859
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4787991053842968,
      "learning_rate": 2.581930763968651e-05,
      "loss": 0.1001,
      "step": 3860
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4000528352830764,
      "learning_rate": 2.581250096010569e-05,
      "loss": 0.0761,
      "step": 3861
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.49798320288462,
      "learning_rate": 2.580569354508925e-05,
      "loss": 0.099,
      "step": 3862
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3794917076685017,
      "learning_rate": 2.5798885395498518e-05,
      "loss": 0.0764,
      "step": 3863
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.47551176836190107,
      "learning_rate": 2.5792076512194895e-05,
      "loss": 0.0974,
      "step": 3864
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4815467897757516,
      "learning_rate": 2.5785266896039897e-05,
      "loss": 0.1136,
      "step": 3865
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5750016046584585,
      "learning_rate": 2.5778456547895117e-05,
      "loss": 0.1074,
      "step": 3866
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3401514830494021,
      "learning_rate": 2.577164546862224e-05,
      "loss": 0.0758,
      "step": 3867
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40080134097896225,
      "learning_rate": 2.5764833659083053e-05,
      "loss": 0.0784,
      "step": 3868
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.507016746889598,
      "learning_rate": 2.5758021120139427e-05,
      "loss": 0.2175,
      "step": 3869
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39052492767220004,
      "learning_rate": 2.5751207852653334e-05,
      "loss": 0.073,
      "step": 3870
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38043666759166145,
      "learning_rate": 2.574439385748683e-05,
      "loss": 0.0897,
      "step": 3871
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.36342193586589333,
      "learning_rate": 2.5737579135502068e-05,
      "loss": 0.0729,
      "step": 3872
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4488029545962984,
      "learning_rate": 2.573076368756129e-05,
      "loss": 0.1009,
      "step": 3873
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.34412541263831325,
      "learning_rate": 2.572394751452683e-05,
      "loss": 0.0664,
      "step": 3874
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3933295275549851,
      "learning_rate": 2.571713061726111e-05,
      "loss": 0.0671,
      "step": 3875
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44359016293388487,
      "learning_rate": 2.5710312996626667e-05,
      "loss": 0.1214,
      "step": 3876
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3049379077814045,
      "learning_rate": 2.5703494653486084e-05,
      "loss": 0.0718,
      "step": 3877
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3725122145435892,
      "learning_rate": 2.569667558870209e-05,
      "loss": 0.0762,
      "step": 3878
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35834157632987024,
      "learning_rate": 2.5689855803137453e-05,
      "loss": 0.0893,
      "step": 3879
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.527352905193966,
      "learning_rate": 2.5683035297655076e-05,
      "loss": 0.1036,
      "step": 3880
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38136529820622933,
      "learning_rate": 2.567621407311792e-05,
      "loss": 0.0634,
      "step": 3881
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38202547408342197,
      "learning_rate": 2.566939213038906e-05,
      "loss": 0.0956,
      "step": 3882
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41750340092010485,
      "learning_rate": 2.5662569470331645e-05,
      "loss": 0.0833,
      "step": 3883
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40203247251002283,
      "learning_rate": 2.5655746093808934e-05,
      "loss": 0.0878,
      "step": 3884
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3290538583473927,
      "learning_rate": 2.564892200168425e-05,
      "loss": 0.0732,
      "step": 3885
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.36282668849068805,
      "learning_rate": 2.564209719482104e-05,
      "loss": 0.0605,
      "step": 3886
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5207934427879871,
      "learning_rate": 2.5635271674082805e-05,
      "loss": 0.1313,
      "step": 3887
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38136279751346147,
      "learning_rate": 2.5628445440333164e-05,
      "loss": 0.0857,
      "step": 3888
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.43607489905622726,
      "learning_rate": 2.562161849443582e-05,
      "loss": 0.0786,
      "step": 3889
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41371748603462327,
      "learning_rate": 2.5614790837254555e-05,
      "loss": 0.0724,
      "step": 3890
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37772882530451274,
      "learning_rate": 2.5607962469653253e-05,
      "loss": 0.0848,
      "step": 3891
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6028516505744399,
      "learning_rate": 2.5601133392495886e-05,
      "loss": 0.0796,
      "step": 3892
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3652959353539588,
      "learning_rate": 2.5594303606646503e-05,
      "loss": 0.0651,
      "step": 3893
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.46588016021589185,
      "learning_rate": 2.558747311296926e-05,
      "loss": 0.097,
      "step": 3894
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5002860800577672,
      "learning_rate": 2.5580641912328402e-05,
      "loss": 0.113,
      "step": 3895
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3927328269902187,
      "learning_rate": 2.5573810005588245e-05,
      "loss": 0.0885,
      "step": 3896
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4902156963057521,
      "learning_rate": 2.5566977393613224e-05,
      "loss": 0.1,
      "step": 3897
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4662990229558147,
      "learning_rate": 2.5560144077267826e-05,
      "loss": 0.097,
      "step": 3898
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.43230247675543354,
      "learning_rate": 2.555331005741666e-05,
      "loss": 0.0779,
      "step": 3899
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3246016252421191,
      "learning_rate": 2.5546475334924398e-05,
      "loss": 0.0541,
      "step": 3900
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37721433137971877,
      "learning_rate": 2.5539639910655827e-05,
      "loss": 0.0908,
      "step": 3901
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.32857107313689266,
      "learning_rate": 2.5532803785475802e-05,
      "loss": 0.0728,
      "step": 3902
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3931847814642713,
      "learning_rate": 2.552596696024928e-05,
      "loss": 0.0752,
      "step": 3903
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4531462269776736,
      "learning_rate": 2.5519129435841298e-05,
      "loss": 0.1185,
      "step": 3904
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44381032184495556,
      "learning_rate": 2.5512291213116984e-05,
      "loss": 0.0802,
      "step": 3905
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33268377474201083,
      "learning_rate": 2.550545229294155e-05,
      "loss": 0.0807,
      "step": 3906
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5052706558746696,
      "learning_rate": 2.549861267618031e-05,
      "loss": 0.1216,
      "step": 3907
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39228884158921085,
      "learning_rate": 2.549177236369865e-05,
      "loss": 0.07,
      "step": 3908
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3824918143325257,
      "learning_rate": 2.5484931356362057e-05,
      "loss": 0.0831,
      "step": 3909
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38590434082532665,
      "learning_rate": 2.5478089655036086e-05,
      "loss": 0.0599,
      "step": 3910
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.30644358238767555,
      "learning_rate": 2.5471247260586416e-05,
      "loss": 0.0674,
      "step": 3911
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3954896522402884,
      "learning_rate": 2.5464404173878775e-05,
      "loss": 0.0868,
      "step": 3912
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35155877005399805,
      "learning_rate": 2.5457560395779003e-05,
      "loss": 0.0624,
      "step": 3913
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35703782257107175,
      "learning_rate": 2.5450715927153012e-05,
      "loss": 0.0591,
      "step": 3914
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3296240442686724,
      "learning_rate": 2.5443870768866816e-05,
      "loss": 0.0662,
      "step": 3915
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39548439615890185,
      "learning_rate": 2.54370249217865e-05,
      "loss": 0.0521,
      "step": 3916
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5679655513122402,
      "learning_rate": 2.543017838677826e-05,
      "loss": 0.1113,
      "step": 3917
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3715630224431169,
      "learning_rate": 2.542333116470835e-05,
      "loss": 0.0679,
      "step": 3918
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4330067140309773,
      "learning_rate": 2.5416483256443135e-05,
      "loss": 0.0935,
      "step": 3919
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4378313444712211,
      "learning_rate": 2.5409634662849053e-05,
      "loss": 0.0766,
      "step": 3920
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40397802855095727,
      "learning_rate": 2.540278538479263e-05,
      "loss": 0.0997,
      "step": 3921
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3989035367892105,
      "learning_rate": 2.5395935423140487e-05,
      "loss": 0.059,
      "step": 3922
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38609519933523473,
      "learning_rate": 2.5389084778759324e-05,
      "loss": 0.1034,
      "step": 3923
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38970434413991617,
      "learning_rate": 2.5382233452515927e-05,
      "loss": 0.0812,
      "step": 3924
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39774739583000324,
      "learning_rate": 2.5375381445277176e-05,
      "loss": 0.0678,
      "step": 3925
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5641328107384846,
      "learning_rate": 2.5368528757910027e-05,
      "loss": 0.1254,
      "step": 3926
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.408759108100696,
      "learning_rate": 2.5361675391281523e-05,
      "loss": 0.0795,
      "step": 3927
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5993204570134232,
      "learning_rate": 2.5354821346258813e-05,
      "loss": 0.0689,
      "step": 3928
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4668339989875443,
      "learning_rate": 2.5347966623709094e-05,
      "loss": 0.1132,
      "step": 3929
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.377967004633986,
      "learning_rate": 2.534111122449969e-05,
      "loss": 0.1083,
      "step": 3930
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3502302464061644,
      "learning_rate": 2.533425514949797e-05,
      "loss": 0.0809,
      "step": 3931
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5067858368030378,
      "learning_rate": 2.532739839957143e-05,
      "loss": 0.1293,
      "step": 3932
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.31964821668288834,
      "learning_rate": 2.5320540975587617e-05,
      "loss": 0.0754,
      "step": 3933
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3985285748380224,
      "learning_rate": 2.5313682878414185e-05,
      "loss": 0.0714,
      "step": 3934
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3863101303183432,
      "learning_rate": 2.5306824108918857e-05,
      "loss": 0.1071,
      "step": 3935
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.36849703517249066,
      "learning_rate": 2.529996466796946e-05,
      "loss": 0.0837,
      "step": 3936
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3209938385793989,
      "learning_rate": 2.5293104556433888e-05,
      "loss": 0.0866,
      "step": 3937
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.42146012428243845,
      "learning_rate": 2.5286243775180128e-05,
      "loss": 0.0821,
      "step": 3938
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.351332801378856,
      "learning_rate": 2.527938232507625e-05,
      "loss": 0.0863,
      "step": 3939
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3176020173379286,
      "learning_rate": 2.5272520206990418e-05,
      "loss": 0.0598,
      "step": 3940
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3273192002734414,
      "learning_rate": 2.526565742179086e-05,
      "loss": 0.0747,
      "step": 3941
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5135218760279558,
      "learning_rate": 2.5258793970345908e-05,
      "loss": 0.0884,
      "step": 3942
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.34495288454595063,
      "learning_rate": 2.5251929853523968e-05,
      "loss": 0.0845,
      "step": 3943
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3591409003684262,
      "learning_rate": 2.5245065072193534e-05,
      "loss": 0.0627,
      "step": 3944
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3573670267014216,
      "learning_rate": 2.523819962722318e-05,
      "loss": 0.0583,
      "step": 3945
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.49663062593992613,
      "learning_rate": 2.5231333519481577e-05,
      "loss": 0.1108,
      "step": 3946
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3766905903111116,
      "learning_rate": 2.5224466749837454e-05,
      "loss": 0.0649,
      "step": 3947
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.34673657017796344,
      "learning_rate": 2.5217599319159654e-05,
      "loss": 0.0647,
      "step": 3948
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4860739292578158,
      "learning_rate": 2.521073122831708e-05,
      "loss": 0.1043,
      "step": 3949
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33606990156772215,
      "learning_rate": 2.5203862478178732e-05,
      "loss": 0.048,
      "step": 3950
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5270395602519095,
      "learning_rate": 2.5196993069613688e-05,
      "loss": 0.1245,
      "step": 3951
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3578753278786093,
      "learning_rate": 2.519012300349111e-05,
      "loss": 0.085,
      "step": 3952
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4156731283674217,
      "learning_rate": 2.5183252280680242e-05,
      "loss": 0.0793,
      "step": 3953
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3773470189365616,
      "learning_rate": 2.5176380902050418e-05,
      "loss": 0.0728,
      "step": 3954
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3470449323503195,
      "learning_rate": 2.5169508868471048e-05,
      "loss": 0.0757,
      "step": 3955
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4141831851916955,
      "learning_rate": 2.516263618081162e-05,
      "loss": 0.079,
      "step": 3956
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.298639599540234,
      "learning_rate": 2.5155762839941726e-05,
      "loss": 0.051,
      "step": 3957
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4796281610125741,
      "learning_rate": 2.5148888846731007e-05,
      "loss": 0.1312,
      "step": 3958
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35793940280217373,
      "learning_rate": 2.5142014202049222e-05,
      "loss": 0.0809,
      "step": 3959
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4509660655997747,
      "learning_rate": 2.5135138906766185e-05,
      "loss": 0.0933,
      "step": 3960
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41579605182275076,
      "learning_rate": 2.512826296175181e-05,
      "loss": 0.0902,
      "step": 3961
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.48505695082384526,
      "learning_rate": 2.512138636787608e-05,
      "loss": 0.0906,
      "step": 3962
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4352575461595124,
      "learning_rate": 2.5114509126009076e-05,
      "loss": 0.1035,
      "step": 3963
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4130864064459698,
      "learning_rate": 2.510763123702094e-05,
      "loss": 0.0773,
      "step": 3964
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.33966078162726066,
      "learning_rate": 2.5100752701781922e-05,
      "loss": 0.0669,
      "step": 3965
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4258009879760604,
      "learning_rate": 2.5093873521162323e-05,
      "loss": 0.0557,
      "step": 3966
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3693169268673469,
      "learning_rate": 2.5086993696032556e-05,
      "loss": 0.0948,
      "step": 3967
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5109936184643279,
      "learning_rate": 2.5080113227263093e-05,
      "loss": 0.118,
      "step": 3968
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3214970874706362,
      "learning_rate": 2.50732321157245e-05,
      "loss": 0.0739,
      "step": 3969
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28465755175314744,
      "learning_rate": 2.5066350362287407e-05,
      "loss": 0.0499,
      "step": 3970
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.27761344154614254,
      "learning_rate": 2.5059467967822562e-05,
      "loss": 0.0609,
      "step": 3971
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.31501857787934107,
      "learning_rate": 2.5052584933200756e-05,
      "loss": 0.072,
      "step": 3972
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5530520493413115,
      "learning_rate": 2.5045701259292878e-05,
      "loss": 0.1232,
      "step": 3973
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3356339613643383,
      "learning_rate": 2.5038816946969894e-05,
      "loss": 0.0767,
      "step": 3974
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3295410572254891,
      "learning_rate": 2.5031931997102857e-05,
      "loss": 0.0836,
      "step": 3975
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42266841736941374,
      "learning_rate": 2.5025046410562888e-05,
      "loss": 0.0914,
      "step": 3976
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.46362502352276597,
      "learning_rate": 2.5018160188221208e-05,
      "loss": 0.0771,
      "step": 3977
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3267458816055714,
      "learning_rate": 2.501127333094909e-05,
      "loss": 0.0656,
      "step": 3978
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3580363260704932,
      "learning_rate": 2.5004385839617915e-05,
      "loss": 0.0699,
      "step": 3979
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42407254673098055,
      "learning_rate": 2.4997497715099134e-05,
      "loss": 0.082,
      "step": 3980
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.282839548282268,
      "learning_rate": 2.499060895826428e-05,
      "loss": 0.0617,
      "step": 3981
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4409810102476525,
      "learning_rate": 2.4983719569984955e-05,
      "loss": 0.0892,
      "step": 3982
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.32086296001518605,
      "learning_rate": 2.4976829551132857e-05,
      "loss": 0.0832,
      "step": 3983
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3701342173661875,
      "learning_rate": 2.496993890257975e-05,
      "loss": 0.0731,
      "step": 3984
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39735504914365444,
      "learning_rate": 2.496304762519749e-05,
      "loss": 0.074,
      "step": 3985
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.32314524968961716,
      "learning_rate": 2.4956155719858e-05,
      "loss": 0.0565,
      "step": 3986
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.35724220554479313,
      "learning_rate": 2.4949263187433298e-05,
      "loss": 0.0757,
      "step": 3987
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34499910295756825,
      "learning_rate": 2.4942370028795456e-05,
      "loss": 0.0723,
      "step": 3988
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3835298084799812,
      "learning_rate": 2.4935476244816662e-05,
      "loss": 0.0653,
      "step": 3989
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5876722975966179,
      "learning_rate": 2.4928581836369147e-05,
      "loss": 0.1437,
      "step": 3990
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4231232965961918,
      "learning_rate": 2.492168680432525e-05,
      "loss": 0.0795,
      "step": 3991
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5047027443865797,
      "learning_rate": 2.4914791149557358e-05,
      "loss": 0.0805,
      "step": 3992
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38703264155347544,
      "learning_rate": 2.490789487293797e-05,
      "loss": 0.0692,
      "step": 3993
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3729902294351956,
      "learning_rate": 2.490099797533964e-05,
      "loss": 0.0824,
      "step": 3994
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3241176447579964,
      "learning_rate": 2.489410045763501e-05,
      "loss": 0.0763,
      "step": 3995
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3486858295551772,
      "learning_rate": 2.48872023206968e-05,
      "loss": 0.0575,
      "step": 3996
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6431566781860486,
      "learning_rate": 2.4880303565397807e-05,
      "loss": 0.1838,
      "step": 3997
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3290287922171316,
      "learning_rate": 2.487340419261091e-05,
      "loss": 0.063,
      "step": 3998
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.37963091974224433,
      "learning_rate": 2.4866504203209053e-05,
      "loss": 0.0605,
      "step": 3999
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.36840963900559937,
      "learning_rate": 2.485960359806528e-05,
      "loss": 0.0723,
      "step": 4000
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.43069675866398194,
      "learning_rate": 2.4852702378052694e-05,
      "loss": 0.0991,
      "step": 4001
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3638299469413789,
      "learning_rate": 2.4845800544044483e-05,
      "loss": 0.0693,
      "step": 4002
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3283430464916656,
      "learning_rate": 2.483889809691391e-05,
      "loss": 0.0718,
      "step": 4003
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.303546419385742,
      "learning_rate": 2.4831995037534325e-05,
      "loss": 0.0723,
      "step": 4004
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3495961938003505,
      "learning_rate": 2.4825091366779136e-05,
      "loss": 0.0533,
      "step": 4005
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39181733310947897,
      "learning_rate": 2.481818708552185e-05,
      "loss": 0.0864,
      "step": 4006
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42711427235381716,
      "learning_rate": 2.4811282194636043e-05,
      "loss": 0.0966,
      "step": 4007
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.44701205066037386,
      "learning_rate": 2.480437669499537e-05,
      "loss": 0.1001,
      "step": 4008
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3949992359733436,
      "learning_rate": 2.479747058747354e-05,
      "loss": 0.1057,
      "step": 4009
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.33476006626779853,
      "learning_rate": 2.479056387294438e-05,
      "loss": 0.0857,
      "step": 4010
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4069646271932735,
      "learning_rate": 2.4783656552281766e-05,
      "loss": 0.0778,
      "step": 4011
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3299116358070327,
      "learning_rate": 2.4776748626359656e-05,
      "loss": 0.0695,
      "step": 4012
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4402597103596856,
      "learning_rate": 2.4769840096052083e-05,
      "loss": 0.1146,
      "step": 4013
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3558807291267144,
      "learning_rate": 2.4762930962233164e-05,
      "loss": 0.0752,
      "step": 4014
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.430625100654845,
      "learning_rate": 2.475602122577709e-05,
      "loss": 0.0754,
      "step": 4015
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39156348978025585,
      "learning_rate": 2.4749110887558114e-05,
      "loss": 0.0938,
      "step": 4016
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39420705638087716,
      "learning_rate": 2.4742199948450596e-05,
      "loss": 0.0697,
      "step": 4017
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.41255108271373403,
      "learning_rate": 2.4735288409328937e-05,
      "loss": 0.0907,
      "step": 4018
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.41751037847445255,
      "learning_rate": 2.4728376271067643e-05,
      "loss": 0.0855,
      "step": 4019
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.33957999506676667,
      "learning_rate": 2.472146353454127e-05,
      "loss": 0.0771,
      "step": 4020
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.33474912702932336,
      "learning_rate": 2.471455020062447e-05,
      "loss": 0.0824,
      "step": 4021
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3534427588802655,
      "learning_rate": 2.4707636270191956e-05,
      "loss": 0.0714,
      "step": 4022
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.41002302276704905,
      "learning_rate": 2.4700721744118535e-05,
      "loss": 0.0804,
      "step": 4023
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40380173000420444,
      "learning_rate": 2.4693806623279074e-05,
      "loss": 0.0719,
      "step": 4024
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4583213725118816,
      "learning_rate": 2.4686890908548517e-05,
      "loss": 0.1047,
      "step": 4025
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4138015788679207,
      "learning_rate": 2.4679974600801882e-05,
      "loss": 0.0746,
      "step": 4026
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3183128580383109,
      "learning_rate": 2.4673057700914277e-05,
      "loss": 0.0826,
      "step": 4027
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.581030067755732,
      "learning_rate": 2.4666140209760862e-05,
      "loss": 0.1351,
      "step": 4028
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34632531980501,
      "learning_rate": 2.4659222128216895e-05,
      "loss": 0.0547,
      "step": 4029
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.31736662603331023,
      "learning_rate": 2.4652303457157677e-05,
      "loss": 0.0688,
      "step": 4030
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.46296102254072374,
      "learning_rate": 2.464538419745862e-05,
      "loss": 0.0914,
      "step": 4031
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.32812791777630473,
      "learning_rate": 2.4638464349995186e-05,
      "loss": 0.0753,
      "step": 4032
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.323221327001955,
      "learning_rate": 2.463154391564293e-05,
      "loss": 0.0541,
      "step": 4033
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38021630013188373,
      "learning_rate": 2.4624622895277462e-05,
      "loss": 0.0684,
      "step": 4034
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.27850444123872053,
      "learning_rate": 2.4617701289774477e-05,
      "loss": 0.0638,
      "step": 4035
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4160848568901733,
      "learning_rate": 2.461077910000974e-05,
      "loss": 0.0695,
      "step": 4036
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40860013500935866,
      "learning_rate": 2.4603856326859094e-05,
      "loss": 0.0919,
      "step": 4037
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.36864440100999635,
      "learning_rate": 2.4596932971198446e-05,
      "loss": 0.0667,
      "step": 4038
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.36940348313656174,
      "learning_rate": 2.4590009033903796e-05,
      "loss": 0.0781,
      "step": 4039
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38139553966119516,
      "learning_rate": 2.4583084515851194e-05,
      "loss": 0.0787,
      "step": 4040
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4145859342841729,
      "learning_rate": 2.457615941791679e-05,
      "loss": 0.0802,
      "step": 4041
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.48217560200280074,
      "learning_rate": 2.456923374097678e-05,
      "loss": 0.0854,
      "step": 4042
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3827295018742603,
      "learning_rate": 2.4562307485907456e-05,
      "loss": 0.0827,
      "step": 4043
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29226932960738133,
      "learning_rate": 2.4555380653585158e-05,
      "loss": 0.0631,
      "step": 4044
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4157865368128142,
      "learning_rate": 2.454845324488633e-05,
      "loss": 0.0776,
      "step": 4045
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42253312240647284,
      "learning_rate": 2.4541525260687468e-05,
      "loss": 0.0776,
      "step": 4046
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5335129895957245,
      "learning_rate": 2.453459670186514e-05,
      "loss": 0.1128,
      "step": 4047
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3807513744106521,
      "learning_rate": 2.4527667569295996e-05,
      "loss": 0.074,
      "step": 4048
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3970192181634317,
      "learning_rate": 2.4520737863856758e-05,
      "loss": 0.0682,
      "step": 4049
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4232716224902008,
      "learning_rate": 2.4513807586424214e-05,
      "loss": 0.0648,
      "step": 4050
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.383405304282245,
      "learning_rate": 2.450687673787523e-05,
      "loss": 0.088,
      "step": 4051
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3586610253084197,
      "learning_rate": 2.449994531908675e-05,
      "loss": 0.0658,
      "step": 4052
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3253901588097721,
      "learning_rate": 2.4493013330935762e-05,
      "loss": 0.0674,
      "step": 4053
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4307136593586819,
      "learning_rate": 2.4486080774299364e-05,
      "loss": 0.0802,
      "step": 4054
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3483167332115282,
      "learning_rate": 2.4479147650054703e-05,
      "loss": 0.0704,
      "step": 4055
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3625551987419031,
      "learning_rate": 2.4472213959079002e-05,
      "loss": 0.0621,
      "step": 4056
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4295798947239526,
      "learning_rate": 2.4465279702249555e-05,
      "loss": 0.1044,
      "step": 4057
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40744321992138677,
      "learning_rate": 2.4458344880443735e-05,
      "loss": 0.0983,
      "step": 4058
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3277954535693289,
      "learning_rate": 2.4451409494538982e-05,
      "loss": 0.0661,
      "step": 4059
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.41520874145278486,
      "learning_rate": 2.4444473545412804e-05,
      "loss": 0.0648,
      "step": 4060
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3482586859422912,
      "learning_rate": 2.4437537033942778e-05,
      "loss": 0.0786,
      "step": 4061
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43237411524739194,
      "learning_rate": 2.4430599961006563e-05,
      "loss": 0.0846,
      "step": 4062
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36528353436764405,
      "learning_rate": 2.4423662327481874e-05,
      "loss": 0.0719,
      "step": 4063
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3149174407593809,
      "learning_rate": 2.441672413424652e-05,
      "loss": 0.0607,
      "step": 4064
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2749867013403368,
      "learning_rate": 2.440978538217835e-05,
      "loss": 0.0661,
      "step": 4065
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.35795634597474113,
      "learning_rate": 2.4402846072155313e-05,
      "loss": 0.0736,
      "step": 4066
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4154880025419162,
      "learning_rate": 2.4395906205055412e-05,
      "loss": 0.0686,
      "step": 4067
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4540987240282171,
      "learning_rate": 2.4388965781756727e-05,
      "loss": 0.0931,
      "step": 4068
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2978381419675977,
      "learning_rate": 2.4382024803137396e-05,
      "loss": 0.0669,
      "step": 4069
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.44983027754362054,
      "learning_rate": 2.437508327007565e-05,
      "loss": 0.1031,
      "step": 4070
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33034941004204904,
      "learning_rate": 2.4368141183449766e-05,
      "loss": 0.0753,
      "step": 4071
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.24719202112242059,
      "learning_rate": 2.4361198544138117e-05,
      "loss": 0.0507,
      "step": 4072
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2852300522287845,
      "learning_rate": 2.435425535301911e-05,
      "loss": 0.0604,
      "step": 4073
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4154557952066227,
      "learning_rate": 2.4347311610971255e-05,
      "loss": 0.07,
      "step": 4074
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.48120340146458357,
      "learning_rate": 2.434036731887312e-05,
      "loss": 0.1039,
      "step": 4075
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43451232523277183,
      "learning_rate": 2.4333422477603342e-05,
      "loss": 0.0911,
      "step": 4076
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38873196621368605,
      "learning_rate": 2.432647708804063e-05,
      "loss": 0.0673,
      "step": 4077
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3995967904026898,
      "learning_rate": 2.4319531151063753e-05,
      "loss": 0.0769,
      "step": 4078
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3683685625210861,
      "learning_rate": 2.4312584667551563e-05,
      "loss": 0.063,
      "step": 4079
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.31037331775542776,
      "learning_rate": 2.4305637638382967e-05,
      "loss": 0.042,
      "step": 4080
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.46054758940585494,
      "learning_rate": 2.429869006443695e-05,
      "loss": 0.0836,
      "step": 4081
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.30119911496164103,
      "learning_rate": 2.4291741946592575e-05,
      "loss": 0.0637,
      "step": 4082
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.30680140756902124,
      "learning_rate": 2.4284793285728956e-05,
      "loss": 0.0545,
      "step": 4083
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4628624404503353,
      "learning_rate": 2.427784408272528e-05,
      "loss": 0.0955,
      "step": 4084
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4396367007518765,
      "learning_rate": 2.4270894338460805e-05,
      "loss": 0.0834,
      "step": 4085
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3704367996397135,
      "learning_rate": 2.4263944053814866e-05,
      "loss": 0.0793,
      "step": 4086
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4086022137263114,
      "learning_rate": 2.4256993229666854e-05,
      "loss": 0.0796,
      "step": 4087
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.34001792083872495,
      "learning_rate": 2.4250041866896234e-05,
      "loss": 0.0665,
      "step": 4088
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.49202109354535645,
      "learning_rate": 2.4243089966382534e-05,
      "loss": 0.1127,
      "step": 4089
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3956157772441568,
      "learning_rate": 2.4236137529005355e-05,
      "loss": 0.0655,
      "step": 4090
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.44245615302808183,
      "learning_rate": 2.422918455564437e-05,
      "loss": 0.097,
      "step": 4091
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4112510782882793,
      "learning_rate": 2.4222231047179303e-05,
      "loss": 0.0728,
      "step": 4092
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3733790451650575,
      "learning_rate": 2.4215277004489975e-05,
      "loss": 0.0654,
      "step": 4093
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4219935568480347,
      "learning_rate": 2.420832242845624e-05,
      "loss": 0.0728,
      "step": 4094
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39155022719893856,
      "learning_rate": 2.4201367319958047e-05,
      "loss": 0.0704,
      "step": 4095
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4190375308819891,
      "learning_rate": 2.41944116798754e-05,
      "loss": 0.0963,
      "step": 4096
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40209634822410095,
      "learning_rate": 2.418745550908837e-05,
      "loss": 0.0704,
      "step": 4097
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43143862317185117,
      "learning_rate": 2.4180498808477096e-05,
      "loss": 0.0939,
      "step": 4098
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3643987960635374,
      "learning_rate": 2.4173541578921785e-05,
      "loss": 0.074,
      "step": 4099
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3656240821892068,
      "learning_rate": 2.4166583821302712e-05,
      "loss": 0.0869,
      "step": 4100
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33639262663049807,
      "learning_rate": 2.4159625536500223e-05,
      "loss": 0.0727,
      "step": 4101
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3486097632730796,
      "learning_rate": 2.4152666725394717e-05,
      "loss": 0.0702,
      "step": 4102
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39061588276713655,
      "learning_rate": 2.414570738886668e-05,
      "loss": 0.0995,
      "step": 4103
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.32947586914608973,
      "learning_rate": 2.413874752779664e-05,
      "loss": 0.058,
      "step": 4104
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38648395924635454,
      "learning_rate": 2.4131787143065216e-05,
      "loss": 0.0703,
      "step": 4105
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40905455722193673,
      "learning_rate": 2.412482623555307e-05,
      "loss": 0.1096,
      "step": 4106
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.46141518878384324,
      "learning_rate": 2.4117864806140948e-05,
      "loss": 0.0866,
      "step": 4107
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3523039522553531,
      "learning_rate": 2.411090285570965e-05,
      "loss": 0.0716,
      "step": 4108
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3467673177513218,
      "learning_rate": 2.410394038514005e-05,
      "loss": 0.0898,
      "step": 4109
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3338156491445141,
      "learning_rate": 2.4096977395313096e-05,
      "loss": 0.0701,
      "step": 4110
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.41313677896510176,
      "learning_rate": 2.4090013887109776e-05,
      "loss": 0.0836,
      "step": 4111
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4007530432994274,
      "learning_rate": 2.4083049861411173e-05,
      "loss": 0.0662,
      "step": 4112
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4856953008131061,
      "learning_rate": 2.4076085319098406e-05,
      "loss": 0.091,
      "step": 4113
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3949056303190438,
      "learning_rate": 2.4069120261052682e-05,
      "loss": 0.0844,
      "step": 4114
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38848210978586917,
      "learning_rate": 2.406215468815526e-05,
      "loss": 0.0709,
      "step": 4115
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38983537974574095,
      "learning_rate": 2.4055188601287483e-05,
      "loss": 0.0787,
      "step": 4116
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40612616852497985,
      "learning_rate": 2.4048222001330727e-05,
      "loss": 0.0534,
      "step": 4117
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2572359661216907,
      "learning_rate": 2.404125488916647e-05,
      "loss": 0.0703,
      "step": 4118
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4374908718791743,
      "learning_rate": 2.4034287265676223e-05,
      "loss": 0.0826,
      "step": 4119
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4128683612923811,
      "learning_rate": 2.402731913174159e-05,
      "loss": 0.0676,
      "step": 4120
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5551200040177134,
      "learning_rate": 2.402035048824421e-05,
      "loss": 0.1235,
      "step": 4121
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4619953495303507,
      "learning_rate": 2.4013381336065805e-05,
      "loss": 0.0751,
      "step": 4122
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.30160906437127283,
      "learning_rate": 2.400641167608816e-05,
      "loss": 0.0466,
      "step": 4123
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40222303190568415,
      "learning_rate": 2.399944150919313e-05,
      "loss": 0.1006,
      "step": 4124
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3784822274070643,
      "learning_rate": 2.3992470836262607e-05,
      "loss": 0.0719,
      "step": 4125
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.32452008595041476,
      "learning_rate": 2.398549965817858e-05,
      "loss": 0.0721,
      "step": 4126
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.334409287290163,
      "learning_rate": 2.3978527975823082e-05,
      "loss": 0.0846,
      "step": 4127
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3274186342297903,
      "learning_rate": 2.3971555790078228e-05,
      "loss": 0.0565,
      "step": 4128
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40135385904202164,
      "learning_rate": 2.3964583101826172e-05,
      "loss": 0.0948,
      "step": 4129
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39576479251467833,
      "learning_rate": 2.3957609911949146e-05,
      "loss": 0.0774,
      "step": 4130
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.32390304499437667,
      "learning_rate": 2.3950636221329444e-05,
      "loss": 0.0607,
      "step": 4131
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3444486151313712,
      "learning_rate": 2.3943662030849426e-05,
      "loss": 0.0746,
      "step": 4132
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4120789466942001,
      "learning_rate": 2.3936687341391505e-05,
      "loss": 0.0913,
      "step": 4133
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40921932565620195,
      "learning_rate": 2.3929712153838173e-05,
      "loss": 0.0689,
      "step": 4134
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5156134979814723,
      "learning_rate": 2.3922736469071978e-05,
      "loss": 0.1108,
      "step": 4135
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.49798963629230075,
      "learning_rate": 2.3915760287975515e-05,
      "loss": 0.0985,
      "step": 4136
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.46270510017860483,
      "learning_rate": 2.390878361143147e-05,
      "loss": 0.0904,
      "step": 4137
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38336795291201153,
      "learning_rate": 2.390180644032257e-05,
      "loss": 0.0791,
      "step": 4138
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3238420941292933,
      "learning_rate": 2.389482877553161e-05,
      "loss": 0.0741,
      "step": 4139
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4355136875800226,
      "learning_rate": 2.3887850617941464e-05,
      "loss": 0.0953,
      "step": 4140
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33931770389014304,
      "learning_rate": 2.388087196843504e-05,
      "loss": 0.0821,
      "step": 4141
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.29874379083732944,
      "learning_rate": 2.3873892827895332e-05,
      "loss": 0.0574,
      "step": 4142
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3452346127041587,
      "learning_rate": 2.3866913197205376e-05,
      "loss": 0.0884,
      "step": 4143
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26044678831736495,
      "learning_rate": 2.3859933077248285e-05,
      "loss": 0.0553,
      "step": 4144
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26998422402192723,
      "learning_rate": 2.3852952468907237e-05,
      "loss": 0.0525,
      "step": 4145
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41913255539775846,
      "learning_rate": 2.3845971373065452e-05,
      "loss": 0.0684,
      "step": 4146
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32751998299865487,
      "learning_rate": 2.3838989790606238e-05,
      "loss": 0.0652,
      "step": 4147
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39188418570769346,
      "learning_rate": 2.3832007722412934e-05,
      "loss": 0.0811,
      "step": 4148
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3928881886835845,
      "learning_rate": 2.3825025169368966e-05,
      "loss": 0.0761,
      "step": 4149
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39091810673852523,
      "learning_rate": 2.3818042132357812e-05,
      "loss": 0.0641,
      "step": 4150
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34110191538146306,
      "learning_rate": 2.381105861226301e-05,
      "loss": 0.0721,
      "step": 4151
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4511295645555022,
      "learning_rate": 2.3804074609968158e-05,
      "loss": 0.0822,
      "step": 4152
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3163040372797685,
      "learning_rate": 2.3797090126356928e-05,
      "loss": 0.0676,
      "step": 4153
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4131503043742734,
      "learning_rate": 2.3790105162313032e-05,
      "loss": 0.0712,
      "step": 4154
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39081183733655933,
      "learning_rate": 2.3783119718720257e-05,
      "loss": 0.0654,
      "step": 4155
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43488972101846396,
      "learning_rate": 2.3776133796462446e-05,
      "loss": 0.0754,
      "step": 4156
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3792544662386112,
      "learning_rate": 2.376914739642351e-05,
      "loss": 0.0697,
      "step": 4157
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.48525745239313234,
      "learning_rate": 2.3762160519487402e-05,
      "loss": 0.1131,
      "step": 4158
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39531456264515613,
      "learning_rate": 2.375517316653816e-05,
      "loss": 0.0773,
      "step": 4159
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3666997876524798,
      "learning_rate": 2.3748185338459847e-05,
      "loss": 0.078,
      "step": 4160
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3415948870417214,
      "learning_rate": 2.3741197036136642e-05,
      "loss": 0.089,
      "step": 4161
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33878134044079133,
      "learning_rate": 2.3734208260452727e-05,
      "loss": 0.062,
      "step": 4162
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4572641154384413,
      "learning_rate": 2.3727219012292377e-05,
      "loss": 0.0985,
      "step": 4163
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.549819295631519,
      "learning_rate": 2.372022929253991e-05,
      "loss": 0.1263,
      "step": 4164
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3773641766376178,
      "learning_rate": 2.371323910207972e-05,
      "loss": 0.068,
      "step": 4165
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4302853673111398,
      "learning_rate": 2.3706248441796246e-05,
      "loss": 0.1107,
      "step": 4166
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38086676957214566,
      "learning_rate": 2.369925731257399e-05,
      "loss": 0.0964,
      "step": 4167
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4197541900105286,
      "learning_rate": 2.369226571529752e-05,
      "loss": 0.106,
      "step": 4168
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34064601605878425,
      "learning_rate": 2.3685273650851458e-05,
      "loss": 0.076,
      "step": 4169
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3711951067187312,
      "learning_rate": 2.3678281120120485e-05,
      "loss": 0.0745,
      "step": 4170
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3664991338324102,
      "learning_rate": 2.3671288123989342e-05,
      "loss": 0.0483,
      "step": 4171
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28306199865450504,
      "learning_rate": 2.366429466334283e-05,
      "loss": 0.0641,
      "step": 4172
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.42092622811249925,
      "learning_rate": 2.36573007390658e-05,
      "loss": 0.0798,
      "step": 4173
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34973236137123703,
      "learning_rate": 2.3650306352043182e-05,
      "loss": 0.081,
      "step": 4174
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3989890339080592,
      "learning_rate": 2.3643311503159937e-05,
      "loss": 0.075,
      "step": 4175
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3480454527272493,
      "learning_rate": 2.3636316193301107e-05,
      "loss": 0.0541,
      "step": 4176
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5366767389327688,
      "learning_rate": 2.362932042335178e-05,
      "loss": 0.0987,
      "step": 4177
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3581306226065945,
      "learning_rate": 2.3622324194197118e-05,
      "loss": 0.0718,
      "step": 4178
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43685258919750963,
      "learning_rate": 2.3615327506722317e-05,
      "loss": 0.099,
      "step": 4179
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2597478272500462,
      "learning_rate": 2.3608330361812652e-05,
      "loss": 0.053,
      "step": 4180
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3593804317561444,
      "learning_rate": 2.3601332760353438e-05,
      "loss": 0.0572,
      "step": 4181
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3173307170661564,
      "learning_rate": 2.3594334703230065e-05,
      "loss": 0.0737,
      "step": 4182
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39167436793378535,
      "learning_rate": 2.358733619132797e-05,
      "loss": 0.0723,
      "step": 4183
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.42147820850913487,
      "learning_rate": 2.3580337225532663e-05,
      "loss": 0.0848,
      "step": 4184
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41178175934471073,
      "learning_rate": 2.357333780672967e-05,
      "loss": 0.0768,
      "step": 4185
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.42869854607194285,
      "learning_rate": 2.356633793580463e-05,
      "loss": 0.083,
      "step": 4186
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34809470656133645,
      "learning_rate": 2.3559337613643198e-05,
      "loss": 0.0829,
      "step": 4187
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3671463273685135,
      "learning_rate": 2.355233684113111e-05,
      "loss": 0.0823,
      "step": 4188
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.30917442592016336,
      "learning_rate": 2.354533561915414e-05,
      "loss": 0.0694,
      "step": 4189
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4699587178869484,
      "learning_rate": 2.3538333948598142e-05,
      "loss": 0.0782,
      "step": 4190
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41889348632081136,
      "learning_rate": 2.3531331830348994e-05,
      "loss": 0.096,
      "step": 4191
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5880135442542722,
      "learning_rate": 2.3524329265292668e-05,
      "loss": 0.1454,
      "step": 4192
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.45799279747530225,
      "learning_rate": 2.351732625431516e-05,
      "loss": 0.0873,
      "step": 4193
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37676002814243087,
      "learning_rate": 2.3510322798302553e-05,
      "loss": 0.0923,
      "step": 4194
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3287243364927219,
      "learning_rate": 2.3503318898140952e-05,
      "loss": 0.0751,
      "step": 4195
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3115377871278743,
      "learning_rate": 2.3496314554716543e-05,
      "loss": 0.0506,
      "step": 4196
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4253234151635005,
      "learning_rate": 2.3489309768915564e-05,
      "loss": 0.099,
      "step": 4197
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33258154650687444,
      "learning_rate": 2.348230454162431e-05,
      "loss": 0.066,
      "step": 4198
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3094640964757944,
      "learning_rate": 2.347529887372912e-05,
      "loss": 0.0755,
      "step": 4199
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.40265021642709675,
      "learning_rate": 2.34682927661164e-05,
      "loss": 0.067,
      "step": 4200
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3810043059203361,
      "learning_rate": 2.346128621967261e-05,
      "loss": 0.0506,
      "step": 4201
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.27806708879514996,
      "learning_rate": 2.3454279235284264e-05,
      "loss": 0.069,
      "step": 4202
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2994116527363354,
      "learning_rate": 2.3447271813837928e-05,
      "loss": 0.0531,
      "step": 4203
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32228744895327155,
      "learning_rate": 2.344026395622023e-05,
      "loss": 0.055,
      "step": 4204
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.44270619410577666,
      "learning_rate": 2.3433255663317845e-05,
      "loss": 0.0755,
      "step": 4205
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.460650500157179,
      "learning_rate": 2.3426246936017514e-05,
      "loss": 0.1092,
      "step": 4206
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3341713209236365,
      "learning_rate": 2.3419237775206026e-05,
      "loss": 0.0692,
      "step": 4207
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3803600069714685,
      "learning_rate": 2.3412228181770224e-05,
      "loss": 0.0747,
      "step": 4208
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32009740911505996,
      "learning_rate": 2.3405218156597012e-05,
      "loss": 0.073,
      "step": 4209
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.344758796805806,
      "learning_rate": 2.3398207700573336e-05,
      "loss": 0.0767,
      "step": 4210
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3060500294927936,
      "learning_rate": 2.3391196814586215e-05,
      "loss": 0.0746,
      "step": 4211
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34794130318990363,
      "learning_rate": 2.3384185499522696e-05,
      "loss": 0.0913,
      "step": 4212
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3214541072200943,
      "learning_rate": 2.3377173756269914e-05,
      "loss": 0.0603,
      "step": 4213
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3397570861283747,
      "learning_rate": 2.337016158571503e-05,
      "loss": 0.0605,
      "step": 4214
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3270237813279916,
      "learning_rate": 2.3363148988745277e-05,
      "loss": 0.047,
      "step": 4215
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.46267321667903893,
      "learning_rate": 2.335613596624793e-05,
      "loss": 0.1162,
      "step": 4216
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28229137851142017,
      "learning_rate": 2.334912251911032e-05,
      "loss": 0.0607,
      "step": 4217
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3400327880210369,
      "learning_rate": 2.334210864821984e-05,
      "loss": 0.0589,
      "step": 4218
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32967501069769883,
      "learning_rate": 2.3335094354463932e-05,
      "loss": 0.069,
      "step": 4219
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4260663818925664,
      "learning_rate": 2.3328079638730073e-05,
      "loss": 0.0998,
      "step": 4220
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43401975581915014,
      "learning_rate": 2.3321064501905837e-05,
      "loss": 0.0746,
      "step": 4221
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32703676737629755,
      "learning_rate": 2.3314048944878804e-05,
      "loss": 0.0728,
      "step": 4222
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5122762366017956,
      "learning_rate": 2.3307032968536642e-05,
      "loss": 0.1224,
      "step": 4223
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3322454322690402,
      "learning_rate": 2.330001657376705e-05,
      "loss": 0.0733,
      "step": 4224
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.29419388510503663,
      "learning_rate": 2.3292999761457794e-05,
      "loss": 0.0617,
      "step": 4225
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4531682421341326,
      "learning_rate": 2.3285982532496676e-05,
      "loss": 0.1073,
      "step": 4226
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.30350394109872575,
      "learning_rate": 2.3278964887771576e-05,
      "loss": 0.0549,
      "step": 4227
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3078286186103161,
      "learning_rate": 2.32719468281704e-05,
      "loss": 0.0812,
      "step": 4228
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.518183458571294,
      "learning_rate": 2.3264928354581126e-05,
      "loss": 0.1055,
      "step": 4229
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37898957427639357,
      "learning_rate": 2.325790946789178e-05,
      "loss": 0.0923,
      "step": 4230
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36397027742686516,
      "learning_rate": 2.3250890168990426e-05,
      "loss": 0.0595,
      "step": 4231
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3852768807758676,
      "learning_rate": 2.32438704587652e-05,
      "loss": 0.0695,
      "step": 4232
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3622624251812756,
      "learning_rate": 2.323685033810428e-05,
      "loss": 0.05,
      "step": 4233
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3582332344670945,
      "learning_rate": 2.3229829807895904e-05,
      "loss": 0.0845,
      "step": 4234
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3058011658527163,
      "learning_rate": 2.3222808869028342e-05,
      "loss": 0.054,
      "step": 4235
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4245133025440607,
      "learning_rate": 2.3215787522389935e-05,
      "loss": 0.0772,
      "step": 4236
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35221508705049276,
      "learning_rate": 2.3208765768869074e-05,
      "loss": 0.0653,
      "step": 4237
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42820459586567544,
      "learning_rate": 2.3201743609354187e-05,
      "loss": 0.0863,
      "step": 4238
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.318704675275583,
      "learning_rate": 2.3194721044733776e-05,
      "loss": 0.0633,
      "step": 4239
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3090389637523935,
      "learning_rate": 2.3187698075896378e-05,
      "loss": 0.0504,
      "step": 4240
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4908221850831954,
      "learning_rate": 2.318067470373058e-05,
      "loss": 0.099,
      "step": 4241
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4647447176033386,
      "learning_rate": 2.317365092912503e-05,
      "loss": 0.0786,
      "step": 4242
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38551470652283654,
      "learning_rate": 2.3166626752968418e-05,
      "loss": 0.0746,
      "step": 4243
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5598766560190404,
      "learning_rate": 2.3159602176149493e-05,
      "loss": 0.1262,
      "step": 4244
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4753533015649203,
      "learning_rate": 2.3152577199557045e-05,
      "loss": 0.1013,
      "step": 4245
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4116600805533102,
      "learning_rate": 2.314555182407992e-05,
      "loss": 0.0654,
      "step": 4246
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.34096132875242663,
      "learning_rate": 2.313852605060702e-05,
      "loss": 0.0706,
      "step": 4247
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4382657094230001,
      "learning_rate": 2.3131499880027294e-05,
      "loss": 0.0821,
      "step": 4248
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44984770157970866,
      "learning_rate": 2.3124473313229726e-05,
      "loss": 0.0781,
      "step": 4249
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4121756116238622,
      "learning_rate": 2.311744635110338e-05,
      "loss": 0.0961,
      "step": 4250
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37420217760060026,
      "learning_rate": 2.3110418994537332e-05,
      "loss": 0.0733,
      "step": 4251
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3456929954294772,
      "learning_rate": 2.3103391244420754e-05,
      "loss": 0.0766,
      "step": 4252
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38187667746417386,
      "learning_rate": 2.3096363101642822e-05,
      "loss": 0.0679,
      "step": 4253
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4710642272403325,
      "learning_rate": 2.30893345670928e-05,
      "loss": 0.1012,
      "step": 4254
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3570202932199923,
      "learning_rate": 2.3082305641659968e-05,
      "loss": 0.0992,
      "step": 4255
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33272146403693625,
      "learning_rate": 2.3075276326233676e-05,
      "loss": 0.0793,
      "step": 4256
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.32790787643048164,
      "learning_rate": 2.3068246621703325e-05,
      "loss": 0.0676,
      "step": 4257
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33956588698350965,
      "learning_rate": 2.306121652895836e-05,
      "loss": 0.0734,
      "step": 4258
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3716345610814525,
      "learning_rate": 2.3054186048888264e-05,
      "loss": 0.0866,
      "step": 4259
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3037221115362737,
      "learning_rate": 2.3047155182382584e-05,
      "loss": 0.0565,
      "step": 4260
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3006909594078138,
      "learning_rate": 2.304012393033092e-05,
      "loss": 0.0553,
      "step": 4261
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33139467363943415,
      "learning_rate": 2.3033092293622903e-05,
      "loss": 0.0739,
      "step": 4262
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37087028199689787,
      "learning_rate": 2.3026060273148217e-05,
      "loss": 0.0758,
      "step": 4263
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.31561170115638904,
      "learning_rate": 2.3019027869796607e-05,
      "loss": 0.0632,
      "step": 4264
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.45581582653255054,
      "learning_rate": 2.301199508445786e-05,
      "loss": 0.0874,
      "step": 4265
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28860069450362924,
      "learning_rate": 2.3004961918021804e-05,
      "loss": 0.0534,
      "step": 4266
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2994410393811674,
      "learning_rate": 2.299792837137833e-05,
      "loss": 0.0632,
      "step": 4267
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.398660597330533,
      "learning_rate": 2.2990894445417355e-05,
      "loss": 0.0965,
      "step": 4268
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3990051676133505,
      "learning_rate": 2.2983860141028874e-05,
      "loss": 0.0733,
      "step": 4269
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3596882491514092,
      "learning_rate": 2.2976825459102898e-05,
      "loss": 0.0704,
      "step": 4270
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4121888793621453,
      "learning_rate": 2.2969790400529508e-05,
      "loss": 0.0776,
      "step": 4271
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42568210882464785,
      "learning_rate": 2.2962754966198815e-05,
      "loss": 0.0996,
      "step": 4272
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3176280321459141,
      "learning_rate": 2.295571915700101e-05,
      "loss": 0.0737,
      "step": 4273
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3501518345919991,
      "learning_rate": 2.2948682973826292e-05,
      "loss": 0.0796,
      "step": 4274
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39159007067478147,
      "learning_rate": 2.2941646417564932e-05,
      "loss": 0.0534,
      "step": 4275
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35058927401599344,
      "learning_rate": 2.2934609489107236e-05,
      "loss": 0.0786,
      "step": 4276
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3067731267306875,
      "learning_rate": 2.292757218934357e-05,
      "loss": 0.0714,
      "step": 4277
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42958478566925395,
      "learning_rate": 2.292053451916433e-05,
      "loss": 0.0846,
      "step": 4278
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43159341338176777,
      "learning_rate": 2.2913496479459977e-05,
      "loss": 0.0684,
      "step": 4279
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35833149214574656,
      "learning_rate": 2.2906458071121e-05,
      "loss": 0.0607,
      "step": 4280
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3847828391624839,
      "learning_rate": 2.2899419295037952e-05,
      "loss": 0.061,
      "step": 4281
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36272176092506414,
      "learning_rate": 2.289238015210142e-05,
      "loss": 0.0535,
      "step": 4282
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.40597251438606413,
      "learning_rate": 2.2885340643202053e-05,
      "loss": 0.0589,
      "step": 4283
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3679154667271321,
      "learning_rate": 2.2878300769230522e-05,
      "loss": 0.0598,
      "step": 4284
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37538474928815474,
      "learning_rate": 2.2871260531077565e-05,
      "loss": 0.0733,
      "step": 4285
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.41514535776938283,
      "learning_rate": 2.2864219929633956e-05,
      "loss": 0.0816,
      "step": 4286
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4352916945053992,
      "learning_rate": 2.2857178965790522e-05,
      "loss": 0.0849,
      "step": 4287
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3696158040192482,
      "learning_rate": 2.2850137640438126e-05,
      "loss": 0.0925,
      "step": 4288
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33078205506772335,
      "learning_rate": 2.2843095954467684e-05,
      "loss": 0.0544,
      "step": 4289
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43250343773418265,
      "learning_rate": 2.2836053908770165e-05,
      "loss": 0.1071,
      "step": 4290
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4032859695258915,
      "learning_rate": 2.2829011504236563e-05,
      "loss": 0.0884,
      "step": 4291
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4340043229017442,
      "learning_rate": 2.2821968741757935e-05,
      "loss": 0.0741,
      "step": 4292
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3598764901085625,
      "learning_rate": 2.281492562222537e-05,
      "loss": 0.084,
      "step": 4293
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.34655353238720704,
      "learning_rate": 2.280788214653003e-05,
      "loss": 0.0785,
      "step": 4294
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3143926645287379,
      "learning_rate": 2.2800838315563074e-05,
      "loss": 0.0692,
      "step": 4295
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3617421021143069,
      "learning_rate": 2.2793794130215753e-05,
      "loss": 0.0624,
      "step": 4296
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3426628042934496,
      "learning_rate": 2.2786749591379333e-05,
      "loss": 0.0765,
      "step": 4297
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3368312125685827,
      "learning_rate": 2.2779704699945136e-05,
      "loss": 0.0756,
      "step": 4298
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.29035318748566485,
      "learning_rate": 2.2772659456804537e-05,
      "loss": 0.0661,
      "step": 4299
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3398279866037583,
      "learning_rate": 2.2765613862848936e-05,
      "loss": 0.0593,
      "step": 4300
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.339979813183332,
      "learning_rate": 2.2758567918969794e-05,
      "loss": 0.0648,
      "step": 4301
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36565660356400675,
      "learning_rate": 2.2751521626058607e-05,
      "loss": 0.0901,
      "step": 4302
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3442851149712656,
      "learning_rate": 2.2744474985006918e-05,
      "loss": 0.0693,
      "step": 4303
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3941907262952854,
      "learning_rate": 2.2737427996706316e-05,
      "loss": 0.0646,
      "step": 4304
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3059710463191362,
      "learning_rate": 2.2730380662048433e-05,
      "loss": 0.0482,
      "step": 4305
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3449319762395715,
      "learning_rate": 2.2723332981924937e-05,
      "loss": 0.064,
      "step": 4306
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37771199977685216,
      "learning_rate": 2.2716284957227547e-05,
      "loss": 0.0727,
      "step": 4307
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.338424789598626,
      "learning_rate": 2.2709236588848036e-05,
      "loss": 0.0472,
      "step": 4308
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3650090139399516,
      "learning_rate": 2.2702187877678198e-05,
      "loss": 0.0838,
      "step": 4309
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4490433060082027,
      "learning_rate": 2.269513882460989e-05,
      "loss": 0.1151,
      "step": 4310
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3573037040766145,
      "learning_rate": 2.2688089430535003e-05,
      "loss": 0.0535,
      "step": 4311
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3710508422138282,
      "learning_rate": 2.268103969634547e-05,
      "loss": 0.0719,
      "step": 4312
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4024471779631351,
      "learning_rate": 2.267398962293327e-05,
      "loss": 0.085,
      "step": 4313
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33924127207894916,
      "learning_rate": 2.266693921119042e-05,
      "loss": 0.0636,
      "step": 4314
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4911987121152769,
      "learning_rate": 2.2659888462008988e-05,
      "loss": 0.0999,
      "step": 4315
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5085896633939151,
      "learning_rate": 2.2652837376281087e-05,
      "loss": 0.1133,
      "step": 4316
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3126775952191896,
      "learning_rate": 2.2645785954898858e-05,
      "loss": 0.0646,
      "step": 4317
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.352744298987547,
      "learning_rate": 2.2638734198754496e-05,
      "loss": 0.0724,
      "step": 4318
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3695248819008147,
      "learning_rate": 2.2631682108740236e-05,
      "loss": 0.0616,
      "step": 4319
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.31011539449580666,
      "learning_rate": 2.2624629685748353e-05,
      "loss": 0.0562,
      "step": 4320
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2795273023681627,
      "learning_rate": 2.2617576930671165e-05,
      "loss": 0.0753,
      "step": 4321
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44186316953196797,
      "learning_rate": 2.261052384440104e-05,
      "loss": 0.0826,
      "step": 4322
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4266623907539318,
      "learning_rate": 2.2603470427830364e-05,
      "loss": 0.0756,
      "step": 4323
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3183496859895677,
      "learning_rate": 2.2596416681851595e-05,
      "loss": 0.0607,
      "step": 4324
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3967897338585108,
      "learning_rate": 2.258936260735722e-05,
      "loss": 0.0694,
      "step": 4325
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41219262100404985,
      "learning_rate": 2.2582308205239757e-05,
      "loss": 0.0886,
      "step": 4326
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3138468447954937,
      "learning_rate": 2.257525347639179e-05,
      "loss": 0.0514,
      "step": 4327
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3353097950736494,
      "learning_rate": 2.256819842170591e-05,
      "loss": 0.0598,
      "step": 4328
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34194531078617957,
      "learning_rate": 2.2561143042074784e-05,
      "loss": 0.0623,
      "step": 4329
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3495438050738101,
      "learning_rate": 2.2554087338391098e-05,
      "loss": 0.0815,
      "step": 4330
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.44620250972792613,
      "learning_rate": 2.254703131154759e-05,
      "loss": 0.0824,
      "step": 4331
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4488103917854197,
      "learning_rate": 2.2539974962437022e-05,
      "loss": 0.1149,
      "step": 4332
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4020510228999266,
      "learning_rate": 2.2532918291952228e-05,
      "loss": 0.0702,
      "step": 4333
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.416747009955414,
      "learning_rate": 2.252586130098605e-05,
      "loss": 0.0928,
      "step": 4334
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3172327126849917,
      "learning_rate": 2.2518803990431394e-05,
      "loss": 0.0633,
      "step": 4335
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36649567787697573,
      "learning_rate": 2.251174636118119e-05,
      "loss": 0.0653,
      "step": 4336
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3008968948390082,
      "learning_rate": 2.2504688414128423e-05,
      "loss": 0.0494,
      "step": 4337
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3460202567470304,
      "learning_rate": 2.2497630150166102e-05,
      "loss": 0.0867,
      "step": 4338
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35697108122758936,
      "learning_rate": 2.249057157018729e-05,
      "loss": 0.0878,
      "step": 4339
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.298349832970042,
      "learning_rate": 2.2483512675085085e-05,
      "loss": 0.0518,
      "step": 4340
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.28430174470920294,
      "learning_rate": 2.2476453465752615e-05,
      "loss": 0.0409,
      "step": 4341
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3674662728954839,
      "learning_rate": 2.2469393943083068e-05,
      "loss": 0.0766,
      "step": 4342
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3394765185162822,
      "learning_rate": 2.2462334107969667e-05,
      "loss": 0.0676,
      "step": 4343
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38048047511776334,
      "learning_rate": 2.245527396130565e-05,
      "loss": 0.0777,
      "step": 4344
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38107137403999786,
      "learning_rate": 2.2448213503984328e-05,
      "loss": 0.1032,
      "step": 4345
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3902662346314301,
      "learning_rate": 2.2441152736899026e-05,
      "loss": 0.0718,
      "step": 4346
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35041807188386576,
      "learning_rate": 2.243409166094313e-05,
      "loss": 0.0527,
      "step": 4347
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.46102904364394653,
      "learning_rate": 2.242703027701004e-05,
      "loss": 0.0978,
      "step": 4348
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32414640056018557,
      "learning_rate": 2.241996858599322e-05,
      "loss": 0.0705,
      "step": 4349
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2800981059588493,
      "learning_rate": 2.2412906588786147e-05,
      "loss": 0.0558,
      "step": 4350
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39614114797534544,
      "learning_rate": 2.240584428628237e-05,
      "loss": 0.0913,
      "step": 4351
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.500656531361286,
      "learning_rate": 2.2398781679375445e-05,
      "loss": 0.1052,
      "step": 4352
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35953443556480524,
      "learning_rate": 2.239171876895898e-05,
      "loss": 0.0779,
      "step": 4353
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3983575142253801,
      "learning_rate": 2.2384655555926625e-05,
      "loss": 0.0998,
      "step": 4354
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3651771296553642,
      "learning_rate": 2.237759204117206e-05,
      "loss": 0.0674,
      "step": 4355
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.344367383190738,
      "learning_rate": 2.237052822558901e-05,
      "loss": 0.0839,
      "step": 4356
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38632957064798795,
      "learning_rate": 2.236346411007124e-05,
      "loss": 0.0874,
      "step": 4357
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.29366639043096426,
      "learning_rate": 2.235639969551253e-05,
      "loss": 0.0522,
      "step": 4358
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3648458648228259,
      "learning_rate": 2.234933498280673e-05,
      "loss": 0.0673,
      "step": 4359
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3266488887286235,
      "learning_rate": 2.2342269972847718e-05,
      "loss": 0.0689,
      "step": 4360
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.371242352898185,
      "learning_rate": 2.2335204666529396e-05,
      "loss": 0.0664,
      "step": 4361
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3198542689991614,
      "learning_rate": 2.232813906474572e-05,
      "loss": 0.0521,
      "step": 4362
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3903140165301045,
      "learning_rate": 2.2321073168390663e-05,
      "loss": 0.0626,
      "step": 4363
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3364296346811924,
      "learning_rate": 2.2314006978358263e-05,
      "loss": 0.0545,
      "step": 4364
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4239453975319239,
      "learning_rate": 2.2306940495542576e-05,
      "loss": 0.0626,
      "step": 4365
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3996125639697796,
      "learning_rate": 2.2299873720837692e-05,
      "loss": 0.0561,
      "step": 4366
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37965688385778257,
      "learning_rate": 2.2292806655137752e-05,
      "loss": 0.0762,
      "step": 4367
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3320122769488699,
      "learning_rate": 2.2285739299336933e-05,
      "loss": 0.0457,
      "step": 4368
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.27999074150003067,
      "learning_rate": 2.227867165432943e-05,
      "loss": 0.0624,
      "step": 4369
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39799841828247606,
      "learning_rate": 2.22716037210095e-05,
      "loss": 0.074,
      "step": 4370
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36179208609406466,
      "learning_rate": 2.2264535500271417e-05,
      "loss": 0.0629,
      "step": 4371
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3608095726767989,
      "learning_rate": 2.2257466993009503e-05,
      "loss": 0.0832,
      "step": 4372
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39271884502774906,
      "learning_rate": 2.2250398200118104e-05,
      "loss": 0.0693,
      "step": 4373
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35879756313132233,
      "learning_rate": 2.2243329122491617e-05,
      "loss": 0.0724,
      "step": 4374
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.40921051346281273,
      "learning_rate": 2.2236259761024467e-05,
      "loss": 0.0969,
      "step": 4375
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3875987703822033,
      "learning_rate": 2.222919011661111e-05,
      "loss": 0.0683,
      "step": 4376
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5225450109414287,
      "learning_rate": 2.222212019014605e-05,
      "loss": 0.0891,
      "step": 4377
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36738611490569284,
      "learning_rate": 2.2215049982523827e-05,
      "loss": 0.0544,
      "step": 4378
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.44154949733328414,
      "learning_rate": 2.2207979494638988e-05,
      "loss": 0.0829,
      "step": 4379
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4373328025677027,
      "learning_rate": 2.2200908727386167e-05,
      "loss": 0.1092,
      "step": 4380
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.30308787718303515,
      "learning_rate": 2.2193837681659975e-05,
      "loss": 0.0621,
      "step": 4381
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.33813915043270465,
      "learning_rate": 2.2186766358355106e-05,
      "loss": 0.0875,
      "step": 4382
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37176416973463605,
      "learning_rate": 2.2179694758366254e-05,
      "loss": 0.0757,
      "step": 4383
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.33130705899813734,
      "learning_rate": 2.217262288258818e-05,
      "loss": 0.0692,
      "step": 4384
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.350758013561006,
      "learning_rate": 2.2165550731915664e-05,
      "loss": 0.0862,
      "step": 4385
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4075060646799392,
      "learning_rate": 2.2158478307243507e-05,
      "loss": 0.0699,
      "step": 4386
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35962987237625565,
      "learning_rate": 2.215140560946656e-05,
      "loss": 0.0752,
      "step": 4387
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.31553130978234917,
      "learning_rate": 2.2144332639479722e-05,
      "loss": 0.0583,
      "step": 4388
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38601402651423605,
      "learning_rate": 2.2137259398177903e-05,
      "loss": 0.0923,
      "step": 4389
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2764534428117698,
      "learning_rate": 2.213018588645605e-05,
      "loss": 0.0558,
      "step": 4390
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42265140650542365,
      "learning_rate": 2.2123112105209157e-05,
      "loss": 0.0728,
      "step": 4391
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4169266445992342,
      "learning_rate": 2.2116038055332238e-05,
      "loss": 0.0892,
      "step": 4392
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3182469151797622,
      "learning_rate": 2.2108963737720354e-05,
      "loss": 0.0607,
      "step": 4393
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4114832347465145,
      "learning_rate": 2.2101889153268595e-05,
      "loss": 0.0851,
      "step": 4394
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3590432377189472,
      "learning_rate": 2.209481430287208e-05,
      "loss": 0.0754,
      "step": 4395
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.28551474989336967,
      "learning_rate": 2.2087739187425967e-05,
      "loss": 0.0629,
      "step": 4396
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37412489667815196,
      "learning_rate": 2.2080663807825448e-05,
      "loss": 0.0726,
      "step": 4397
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3922163021864411,
      "learning_rate": 2.2073588164965737e-05,
      "loss": 0.0615,
      "step": 4398
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.33012003886442975,
      "learning_rate": 2.2066512259742107e-05,
      "loss": 0.0628,
      "step": 4399
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3737348234801594,
      "learning_rate": 2.205943609304983e-05,
      "loss": 0.068,
      "step": 4400
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2970666266920581,
      "learning_rate": 2.205235966578424e-05,
      "loss": 0.051,
      "step": 4401
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39051939490456916,
      "learning_rate": 2.2045282978840684e-05,
      "loss": 0.0646,
      "step": 4402
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26177338841156883,
      "learning_rate": 2.2038206033114565e-05,
      "loss": 0.045,
      "step": 4403
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3168128662212616,
      "learning_rate": 2.2031128829501293e-05,
      "loss": 0.053,
      "step": 4404
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36054623522138607,
      "learning_rate": 2.2024051368896324e-05,
      "loss": 0.0693,
      "step": 4405
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.40205587810540727,
      "learning_rate": 2.2016973652195145e-05,
      "loss": 0.0977,
      "step": 4406
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4264672379091293,
      "learning_rate": 2.2009895680293282e-05,
      "loss": 0.0817,
      "step": 4407
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3300946135606695,
      "learning_rate": 2.200281745408627e-05,
      "loss": 0.0702,
      "step": 4408
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37954877666084774,
      "learning_rate": 2.1995738974469712e-05,
      "loss": 0.0771,
      "step": 4409
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.27303177842303766,
      "learning_rate": 2.1988660242339205e-05,
      "loss": 0.0491,
      "step": 4410
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41433953068684676,
      "learning_rate": 2.198158125859041e-05,
      "loss": 0.0786,
      "step": 4411
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.27899680657678466,
      "learning_rate": 2.1974502024119002e-05,
      "loss": 0.0431,
      "step": 4412
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39167731639235287,
      "learning_rate": 2.196742253982069e-05,
      "loss": 0.0813,
      "step": 4413
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6102844445518398,
      "learning_rate": 2.196034280659122e-05,
      "loss": 0.1344,
      "step": 4414
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3274576576175134,
      "learning_rate": 2.1953262825326357e-05,
      "loss": 0.0593,
      "step": 4415
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32475552304367616,
      "learning_rate": 2.1946182596921917e-05,
      "loss": 0.0623,
      "step": 4416
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4327028417935204,
      "learning_rate": 2.193910212227374e-05,
      "loss": 0.0878,
      "step": 4417
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35699358017275484,
      "learning_rate": 2.1932021402277682e-05,
      "loss": 0.0825,
      "step": 4418
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.372314453125,
      "learning_rate": 2.1924940437829643e-05,
      "loss": 0.0811,
      "step": 4419
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33035941242852357,
      "learning_rate": 2.1917859229825565e-05,
      "loss": 0.0521,
      "step": 4420
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3699903044203544,
      "learning_rate": 2.19107777791614e-05,
      "loss": 0.0707,
      "step": 4421
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2854554161628383,
      "learning_rate": 2.1903696086733142e-05,
      "loss": 0.0717,
      "step": 4422
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3311293076220247,
      "learning_rate": 2.1896614153436804e-05,
      "loss": 0.0472,
      "step": 4423
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3148082362892187,
      "learning_rate": 2.188953198016846e-05,
      "loss": 0.066,
      "step": 4424
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3224210422173281,
      "learning_rate": 2.1882449567824165e-05,
      "loss": 0.049,
      "step": 4425
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3536928270191253,
      "learning_rate": 2.1875366917300057e-05,
      "loss": 0.0718,
      "step": 4426
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.56370309960082,
      "learning_rate": 2.1868284029492256e-05,
      "loss": 0.1338,
      "step": 4427
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38858855614836,
      "learning_rate": 2.1861200905296952e-05,
      "loss": 0.0716,
      "step": 4428
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3697511176143363,
      "learning_rate": 2.1854117545610345e-05,
      "loss": 0.0739,
      "step": 4429
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42592013132642514,
      "learning_rate": 2.1847033951328673e-05,
      "loss": 0.0827,
      "step": 4430
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.382245558713531,
      "learning_rate": 2.183995012334818e-05,
      "loss": 0.0574,
      "step": 4431
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.26570928863133125,
      "learning_rate": 2.1832866062565183e-05,
      "loss": 0.053,
      "step": 4432
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3227790672663503,
      "learning_rate": 2.1825781769875986e-05,
      "loss": 0.0625,
      "step": 4433
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.36195492377391036,
      "learning_rate": 2.1818697246176943e-05,
      "loss": 0.0705,
      "step": 4434
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3300240376416469,
      "learning_rate": 2.1811612492364436e-05,
      "loss": 0.0622,
      "step": 4435
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4537924750338406,
      "learning_rate": 2.1804527509334875e-05,
      "loss": 0.0832,
      "step": 4436
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3826781640528386,
      "learning_rate": 2.17974422979847e-05,
      "loss": 0.0693,
      "step": 4437
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3757246882960812,
      "learning_rate": 2.1790356859210378e-05,
      "loss": 0.0678,
      "step": 4438
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42329380088608826,
      "learning_rate": 2.17832711939084e-05,
      "loss": 0.0884,
      "step": 4439
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3207002247018268,
      "learning_rate": 2.17761853029753e-05,
      "loss": 0.0642,
      "step": 4440
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42221749496358135,
      "learning_rate": 2.1769099187307616e-05,
      "loss": 0.0606,
      "step": 4441
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.30211302457883304,
      "learning_rate": 2.176201284780195e-05,
      "loss": 0.0605,
      "step": 4442
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3349027110392718,
      "learning_rate": 2.1754926285354895e-05,
      "loss": 0.0812,
      "step": 4443
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4498126540677357,
      "learning_rate": 2.1747839500863096e-05,
      "loss": 0.0966,
      "step": 4444
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35671102992495873,
      "learning_rate": 2.1740752495223216e-05,
      "loss": 0.0594,
      "step": 4445
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38519754710689025,
      "learning_rate": 2.1733665269331953e-05,
      "loss": 0.0643,
      "step": 4446
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33666308356650826,
      "learning_rate": 2.1726577824086032e-05,
      "loss": 0.0617,
      "step": 4447
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3212680056086583,
      "learning_rate": 2.1719490160382196e-05,
      "loss": 0.0485,
      "step": 4448
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2859127643704896,
      "learning_rate": 2.1712402279117225e-05,
      "loss": 0.0421,
      "step": 4449
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.45362353031007613,
      "learning_rate": 2.1705314181187922e-05,
      "loss": 0.079,
      "step": 4450
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3033710552740286,
      "learning_rate": 2.169822586749113e-05,
      "loss": 0.0588,
      "step": 4451
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3022458965442612,
      "learning_rate": 2.169113733892369e-05,
      "loss": 0.078,
      "step": 4452
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.49415525030272484,
      "learning_rate": 2.1684048596382507e-05,
      "loss": 0.0998,
      "step": 4453
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40718191803204307,
      "learning_rate": 2.1676959640764484e-05,
      "loss": 0.056,
      "step": 4454
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3225641197588187,
      "learning_rate": 2.1669870472966567e-05,
      "loss": 0.0692,
      "step": 4455
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33938903642574786,
      "learning_rate": 2.166278109388572e-05,
      "loss": 0.0457,
      "step": 4456
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34952943836368505,
      "learning_rate": 2.165569150441895e-05,
      "loss": 0.074,
      "step": 4457
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3466505656301169,
      "learning_rate": 2.1648601705463263e-05,
      "loss": 0.0513,
      "step": 4458
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.28325509219377515,
      "learning_rate": 2.164151169791571e-05,
      "loss": 0.045,
      "step": 4459
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3090287052574003,
      "learning_rate": 2.1634421482673368e-05,
      "loss": 0.0444,
      "step": 4460
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3609038465259484,
      "learning_rate": 2.162733106063334e-05,
      "loss": 0.0925,
      "step": 4461
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25518660874576665,
      "learning_rate": 2.1620240432692737e-05,
      "loss": 0.0544,
      "step": 4462
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39989662399207204,
      "learning_rate": 2.1613149599748738e-05,
      "loss": 0.0845,
      "step": 4463
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4230374638997947,
      "learning_rate": 2.1606058562698496e-05,
      "loss": 0.1073,
      "step": 4464
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3639532047966518,
      "learning_rate": 2.1598967322439237e-05,
      "loss": 0.0736,
      "step": 4465
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40011851670918575,
      "learning_rate": 2.1591875879868177e-05,
      "loss": 0.0785,
      "step": 4466
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4000397982766122,
      "learning_rate": 2.1584784235882575e-05,
      "loss": 0.059,
      "step": 4467
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3629141562791462,
      "learning_rate": 2.157769239137971e-05,
      "loss": 0.0623,
      "step": 4468
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3069926992388575,
      "learning_rate": 2.1570600347256894e-05,
      "loss": 0.0533,
      "step": 4469
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.326456242937862,
      "learning_rate": 2.1563508104411457e-05,
      "loss": 0.072,
      "step": 4470
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.378637592203253,
      "learning_rate": 2.155641566374075e-05,
      "loss": 0.0645,
      "step": 4471
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2808094150001029,
      "learning_rate": 2.1549323026142168e-05,
      "loss": 0.0568,
      "step": 4472
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3555040027618403,
      "learning_rate": 2.15422301925131e-05,
      "loss": 0.0641,
      "step": 4473
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.41654415316289695,
      "learning_rate": 2.153513716375099e-05,
      "loss": 0.1115,
      "step": 4474
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34796017852490074,
      "learning_rate": 2.1528043940753293e-05,
      "loss": 0.0685,
      "step": 4475
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2984227691130474,
      "learning_rate": 2.1520950524417484e-05,
      "loss": 0.0503,
      "step": 4476
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42422315760337237,
      "learning_rate": 2.151385691564108e-05,
      "loss": 0.072,
      "step": 4477
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.31711807947648224,
      "learning_rate": 2.1506763115321602e-05,
      "loss": 0.0634,
      "step": 4478
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5635149593923158,
      "learning_rate": 2.1499669124356602e-05,
      "loss": 0.1025,
      "step": 4479
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.27240057929568373,
      "learning_rate": 2.1492574943643666e-05,
      "loss": 0.0536,
      "step": 4480
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3051025669012113,
      "learning_rate": 2.148548057408039e-05,
      "loss": 0.0573,
      "step": 4481
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4068695259604966,
      "learning_rate": 2.1478386016564406e-05,
      "loss": 0.0801,
      "step": 4482
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33064398707439074,
      "learning_rate": 2.1471291271993353e-05,
      "loss": 0.0698,
      "step": 4483
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2911332551949484,
      "learning_rate": 2.1464196341264915e-05,
      "loss": 0.0587,
      "step": 4484
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2713247831815962,
      "learning_rate": 2.1457101225276782e-05,
      "loss": 0.0491,
      "step": 4485
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4243575977133858,
      "learning_rate": 2.145000592492668e-05,
      "loss": 0.097,
      "step": 4486
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35213316060726657,
      "learning_rate": 2.1442910441112348e-05,
      "loss": 0.0641,
      "step": 4487
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.37737933952613795,
      "learning_rate": 2.1435814774731557e-05,
      "loss": 0.0579,
      "step": 4488
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.44708791475653537,
      "learning_rate": 2.1428718926682094e-05,
      "loss": 0.1056,
      "step": 4489
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.41213808366476695,
      "learning_rate": 2.1421622897861777e-05,
      "loss": 0.0719,
      "step": 4490
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2694959202707098,
      "learning_rate": 2.1414526689168436e-05,
      "loss": 0.0607,
      "step": 4491
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3137855551813448,
      "learning_rate": 2.1407430301499934e-05,
      "loss": 0.0573,
      "step": 4492
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40188769075755126,
      "learning_rate": 2.1400333735754146e-05,
      "loss": 0.0623,
      "step": 4493
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38618627167540487,
      "learning_rate": 2.139323699282899e-05,
      "loss": 0.0765,
      "step": 4494
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3887449032979908,
      "learning_rate": 2.1386140073622374e-05,
      "loss": 0.0731,
      "step": 4495
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39065961684382844,
      "learning_rate": 2.1379042979032256e-05,
      "loss": 0.0731,
      "step": 4496
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4004124705239988,
      "learning_rate": 2.1371945709956607e-05,
      "loss": 0.0641,
      "step": 4497
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.24999219166959438,
      "learning_rate": 2.1364848267293424e-05,
      "loss": 0.0549,
      "step": 4498
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.36493768728401355,
      "learning_rate": 2.135775065194071e-05,
      "loss": 0.0618,
      "step": 4499
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4038511943371582,
      "learning_rate": 2.1350652864796513e-05,
      "loss": 0.0869,
      "step": 4500
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.41569938639347065,
      "learning_rate": 2.1343554906758885e-05,
      "loss": 0.0717,
      "step": 4501
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3690034924460754,
      "learning_rate": 2.133645677872591e-05,
      "loss": 0.0695,
      "step": 4502
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3185312708066224,
      "learning_rate": 2.1329358481595684e-05,
      "loss": 0.0545,
      "step": 4503
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.32036721529825507,
      "learning_rate": 2.1322260016266337e-05,
      "loss": 0.0634,
      "step": 4504
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.304034928094034,
      "learning_rate": 2.1315161383636006e-05,
      "loss": 0.0482,
      "step": 4505
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2953906218320155,
      "learning_rate": 2.1308062584602865e-05,
      "loss": 0.0561,
      "step": 4506
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3825905214669043,
      "learning_rate": 2.1300963620065092e-05,
      "loss": 0.0689,
      "step": 4507
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4904463108516287,
      "learning_rate": 2.1293864490920897e-05,
      "loss": 0.0995,
      "step": 4508
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2820360141045328,
      "learning_rate": 2.1286765198068516e-05,
      "loss": 0.0456,
      "step": 4509
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37802929021619497,
      "learning_rate": 2.1279665742406187e-05,
      "loss": 0.0564,
      "step": 4510
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.39281526681415724,
      "learning_rate": 2.1272566124832183e-05,
      "loss": 0.0691,
      "step": 4511
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.45911288795727956,
      "learning_rate": 2.126546634624479e-05,
      "loss": 0.0859,
      "step": 4512
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.33890490363255027,
      "learning_rate": 2.1258366407542333e-05,
      "loss": 0.0793,
      "step": 4513
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3041221557778955,
      "learning_rate": 2.125126630962312e-05,
      "loss": 0.0592,
      "step": 4514
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3780662624987316,
      "learning_rate": 2.1244166053385528e-05,
      "loss": 0.0668,
      "step": 4515
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3318567602888956,
      "learning_rate": 2.1237065639727906e-05,
      "loss": 0.0622,
      "step": 4516
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3071385853611324,
      "learning_rate": 2.1229965069548658e-05,
      "loss": 0.0552,
      "step": 4517
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4638316251273483,
      "learning_rate": 2.1222864343746185e-05,
      "loss": 0.0741,
      "step": 4518
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.45571939367078695,
      "learning_rate": 2.1215763463218927e-05,
      "loss": 0.0758,
      "step": 4519
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3671358965060949,
      "learning_rate": 2.1208662428865326e-05,
      "loss": 0.0646,
      "step": 4520
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2814296704829106,
      "learning_rate": 2.1201561241583856e-05,
      "loss": 0.0637,
      "step": 4521
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34420202618949586,
      "learning_rate": 2.1194459902272997e-05,
      "loss": 0.0727,
      "step": 4522
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.39079298222199,
      "learning_rate": 2.1187358411831273e-05,
      "loss": 0.0867,
      "step": 4523
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23260615760667525,
      "learning_rate": 2.1180256771157194e-05,
      "loss": 0.0566,
      "step": 4524
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29157824822322,
      "learning_rate": 2.1173154981149322e-05,
      "loss": 0.0515,
      "step": 4525
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.367190543628821,
      "learning_rate": 2.1166053042706204e-05,
      "loss": 0.0623,
      "step": 4526
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.31907652604586173,
      "learning_rate": 2.1158950956726443e-05,
      "loss": 0.0728,
      "step": 4527
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25753795570889043,
      "learning_rate": 2.115184872410862e-05,
      "loss": 0.0548,
      "step": 4528
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3458239241932494,
      "learning_rate": 2.114474634575138e-05,
      "loss": 0.056,
      "step": 4529
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3719278097722103,
      "learning_rate": 2.113764382255334e-05,
      "loss": 0.0814,
      "step": 4530
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.44318147694856896,
      "learning_rate": 2.1130541155413167e-05,
      "loss": 0.0923,
      "step": 4531
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4849299051816181,
      "learning_rate": 2.1123438345229537e-05,
      "loss": 0.101,
      "step": 4532
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6268001143889941,
      "learning_rate": 2.1116335392901154e-05,
      "loss": 0.1052,
      "step": 4533
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3218850402747531,
      "learning_rate": 2.110923229932671e-05,
      "loss": 0.0636,
      "step": 4534
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3483550838112909,
      "learning_rate": 2.1102129065404952e-05,
      "loss": 0.0686,
      "step": 4535
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36008226209777056,
      "learning_rate": 2.1095025692034614e-05,
      "loss": 0.0648,
      "step": 4536
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3414932646387405,
      "learning_rate": 2.1087922180114474e-05,
      "loss": 0.0641,
      "step": 4537
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6524708789597221,
      "learning_rate": 2.1080818530543304e-05,
      "loss": 0.1284,
      "step": 4538
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.38230726416399574,
      "learning_rate": 2.107371474421991e-05,
      "loss": 0.0804,
      "step": 4539
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3675223102414846,
      "learning_rate": 2.106661082204311e-05,
      "loss": 0.0623,
      "step": 4540
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29304390261268176,
      "learning_rate": 2.105950676491173e-05,
      "loss": 0.0704,
      "step": 4541
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29718646224810447,
      "learning_rate": 2.105240257372464e-05,
      "loss": 0.0708,
      "step": 4542
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.24698758153739822,
      "learning_rate": 2.1045298249380687e-05,
      "loss": 0.0493,
      "step": 4543
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3347476691790936,
      "learning_rate": 2.1038193792778775e-05,
      "loss": 0.0781,
      "step": 4544
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37789108512934,
      "learning_rate": 2.103108920481779e-05,
      "loss": 0.0759,
      "step": 4545
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2991314389603449,
      "learning_rate": 2.102398448639667e-05,
      "loss": 0.0597,
      "step": 4546
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3272076225833289,
      "learning_rate": 2.1016879638414327e-05,
      "loss": 0.0509,
      "step": 4547
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3341306399587591,
      "learning_rate": 2.100977466176973e-05,
      "loss": 0.0658,
      "step": 4548
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4034330935425939,
      "learning_rate": 2.100266955736184e-05,
      "loss": 0.1087,
      "step": 4549
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34608031563270264,
      "learning_rate": 2.099556432608965e-05,
      "loss": 0.0657,
      "step": 4550
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2656803353958201,
      "learning_rate": 2.098845896885215e-05,
      "loss": 0.0393,
      "step": 4551
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28982713451213166,
      "learning_rate": 2.0981353486548363e-05,
      "loss": 0.0633,
      "step": 4552
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3471113284835435,
      "learning_rate": 2.0974247880077314e-05,
      "loss": 0.0764,
      "step": 4553
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.32785338558661425,
      "learning_rate": 2.096714215033806e-05,
      "loss": 0.0709,
      "step": 4554
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2781855206595912,
      "learning_rate": 2.0960036298229656e-05,
      "loss": 0.0779,
      "step": 4555
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4002477251306299,
      "learning_rate": 2.095293032465119e-05,
      "loss": 0.0717,
      "step": 4556
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4185926469116941,
      "learning_rate": 2.0945824230501735e-05,
      "loss": 0.0992,
      "step": 4557
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35372442328254866,
      "learning_rate": 2.0938718016680433e-05,
      "loss": 0.0695,
      "step": 4558
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2892305555773046,
      "learning_rate": 2.093161168408638e-05,
      "loss": 0.0611,
      "step": 4559
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3085020387758374,
      "learning_rate": 2.0924505233618734e-05,
      "loss": 0.056,
      "step": 4560
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3919079122152957,
      "learning_rate": 2.091739866617664e-05,
      "loss": 0.0735,
      "step": 4561
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35461793461008023,
      "learning_rate": 2.0910291982659277e-05,
      "loss": 0.0705,
      "step": 4562
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.30415871780712833,
      "learning_rate": 2.0903185183965814e-05,
      "loss": 0.0433,
      "step": 4563
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.41869706132389145,
      "learning_rate": 2.0896078270995463e-05,
      "loss": 0.0728,
      "step": 4564
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3011101992300247,
      "learning_rate": 2.0888971244647425e-05,
      "loss": 0.0652,
      "step": 4565
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3833903408507335,
      "learning_rate": 2.0881864105820936e-05,
      "loss": 0.0905,
      "step": 4566
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3821481662325426,
      "learning_rate": 2.0874756855415245e-05,
      "loss": 0.0682,
      "step": 4567
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2893226844427285,
      "learning_rate": 2.0867649494329587e-05,
      "loss": 0.0523,
      "step": 4568
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35811150314036255,
      "learning_rate": 2.0860542023463256e-05,
      "loss": 0.068,
      "step": 4569
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25479581963363895,
      "learning_rate": 2.085343444371551e-05,
      "loss": 0.05,
      "step": 4570
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35007637416790827,
      "learning_rate": 2.0846326755985672e-05,
      "loss": 0.055,
      "step": 4571
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.44856845392424505,
      "learning_rate": 2.083921896117303e-05,
      "loss": 0.0807,
      "step": 4572
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3368993009567651,
      "learning_rate": 2.0832111060176924e-05,
      "loss": 0.0743,
      "step": 4573
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3359518048101841,
      "learning_rate": 2.0825003053896686e-05,
      "loss": 0.0488,
      "step": 4574
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.38670225782581175,
      "learning_rate": 2.081789494323167e-05,
      "loss": 0.0943,
      "step": 4575
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35578533369626514,
      "learning_rate": 2.0810786729081237e-05,
      "loss": 0.0502,
      "step": 4576
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3729927263356071,
      "learning_rate": 2.0803678412344775e-05,
      "loss": 0.0587,
      "step": 4577
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3280344679235716,
      "learning_rate": 2.079656999392166e-05,
      "loss": 0.0474,
      "step": 4578
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5258295147581339,
      "learning_rate": 2.078946147471131e-05,
      "loss": 0.1168,
      "step": 4579
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.30115483355765493,
      "learning_rate": 2.0782352855613128e-05,
      "loss": 0.0428,
      "step": 4580
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5149466936985557,
      "learning_rate": 2.077524413752655e-05,
      "loss": 0.1027,
      "step": 4581
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.31386350951208836,
      "learning_rate": 2.0768135321351016e-05,
      "loss": 0.0701,
      "step": 4582
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3359317335476653,
      "learning_rate": 2.0761026407985977e-05,
      "loss": 0.063,
      "step": 4583
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.32639873632947247,
      "learning_rate": 2.0753917398330902e-05,
      "loss": 0.0703,
      "step": 4584
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28803314759807985,
      "learning_rate": 2.0746808293285278e-05,
      "loss": 0.0413,
      "step": 4585
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.48819399244752704,
      "learning_rate": 2.073969909374858e-05,
      "loss": 0.0667,
      "step": 4586
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4999769026666627,
      "learning_rate": 2.0732589800620328e-05,
      "loss": 0.0834,
      "step": 4587
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3498040195546038,
      "learning_rate": 2.0725480414800012e-05,
      "loss": 0.0707,
      "step": 4588
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.39269592643317175,
      "learning_rate": 2.0718370937187183e-05,
      "loss": 0.0627,
      "step": 4589
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37791925854549424,
      "learning_rate": 2.0711261368681356e-05,
      "loss": 0.0649,
      "step": 4590
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36467931256096925,
      "learning_rate": 2.0704151710182096e-05,
      "loss": 0.0441,
      "step": 4591
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3811198786388003,
      "learning_rate": 2.069704196258896e-05,
      "loss": 0.0611,
      "step": 4592
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36678401611668343,
      "learning_rate": 2.068993212680152e-05,
      "loss": 0.0585,
      "step": 4593
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.32629940258161994,
      "learning_rate": 2.068282220371936e-05,
      "loss": 0.0728,
      "step": 4594
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4451318006137944,
      "learning_rate": 2.067571219424207e-05,
      "loss": 0.0682,
      "step": 4595
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3159580470609586,
      "learning_rate": 2.066860209926925e-05,
      "loss": 0.0551,
      "step": 4596
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3331972629259536,
      "learning_rate": 2.066149191970053e-05,
      "loss": 0.0672,
      "step": 4597
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4961277206684615,
      "learning_rate": 2.0654381656435526e-05,
      "loss": 0.0947,
      "step": 4598
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30431991803303776,
      "learning_rate": 2.0647271310373877e-05,
      "loss": 0.0526,
      "step": 4599
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3006196392913737,
      "learning_rate": 2.064016088241523e-05,
      "loss": 0.0474,
      "step": 4600
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.41157576740172264,
      "learning_rate": 2.0633050373459246e-05,
      "loss": 0.0725,
      "step": 4601
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3708199545057789,
      "learning_rate": 2.0625939784405586e-05,
      "loss": 0.0556,
      "step": 4602
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4893523560811144,
      "learning_rate": 2.061882911615394e-05,
      "loss": 0.1029,
      "step": 4603
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3511242253922052,
      "learning_rate": 2.0611718369603982e-05,
      "loss": 0.0974,
      "step": 4604
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3298713442681895,
      "learning_rate": 2.060460754565542e-05,
      "loss": 0.0615,
      "step": 4605
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4140478167539539,
      "learning_rate": 2.0597496645207964e-05,
      "loss": 0.0996,
      "step": 4606
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37001847533545773,
      "learning_rate": 2.059038566916132e-05,
      "loss": 0.0694,
      "step": 4607
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4362299399696153,
      "learning_rate": 2.0583274618415227e-05,
      "loss": 0.0878,
      "step": 4608
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.28516118816123653,
      "learning_rate": 2.057616349386941e-05,
      "loss": 0.0565,
      "step": 4609
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31347394805483453,
      "learning_rate": 2.056905229642363e-05,
      "loss": 0.0619,
      "step": 4610
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.295383117951669,
      "learning_rate": 2.056194102697763e-05,
      "loss": 0.0464,
      "step": 4611
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3155562675345661,
      "learning_rate": 2.055482968643118e-05,
      "loss": 0.0624,
      "step": 4612
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.43115505679908894,
      "learning_rate": 2.0547718275684046e-05,
      "loss": 0.0851,
      "step": 4613
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3706003623479934,
      "learning_rate": 2.0540606795636022e-05,
      "loss": 0.0796,
      "step": 4614
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3756549004191829,
      "learning_rate": 2.053349524718689e-05,
      "loss": 0.0795,
      "step": 4615
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3475011522459234,
      "learning_rate": 2.0526383631236454e-05,
      "loss": 0.058,
      "step": 4616
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.253513986849672,
      "learning_rate": 2.051927194868451e-05,
      "loss": 0.0459,
      "step": 4617
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4014950662240892,
      "learning_rate": 2.0512160200430896e-05,
      "loss": 0.0846,
      "step": 4618
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3162283102709553,
      "learning_rate": 2.050504838737542e-05,
      "loss": 0.0737,
      "step": 4619
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.33028667114382493,
      "learning_rate": 2.0497936510417928e-05,
      "loss": 0.0632,
      "step": 4620
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.27218718089666394,
      "learning_rate": 2.0490824570458248e-05,
      "loss": 0.0414,
      "step": 4621
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4156168427778287,
      "learning_rate": 2.048371256839624e-05,
      "loss": 0.0722,
      "step": 4622
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3755848020158526,
      "learning_rate": 2.0476600505131747e-05,
      "loss": 0.0877,
      "step": 4623
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5013673979250173,
      "learning_rate": 2.046948838156465e-05,
      "loss": 0.085,
      "step": 4624
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3625648572056343,
      "learning_rate": 2.0462376198594813e-05,
      "loss": 0.0733,
      "step": 4625
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3110875036243839,
      "learning_rate": 2.0455263957122113e-05,
      "loss": 0.0619,
      "step": 4626
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.26201803261638423,
      "learning_rate": 2.044815165804645e-05,
      "loss": 0.0593,
      "step": 4627
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3392611036212399,
      "learning_rate": 2.04410393022677e-05,
      "loss": 0.064,
      "step": 4628
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3797220120227483,
      "learning_rate": 2.043392689068578e-05,
      "loss": 0.0564,
      "step": 4629
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4166361877102114,
      "learning_rate": 2.0426814424200592e-05,
      "loss": 0.0808,
      "step": 4630
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4395868564151161,
      "learning_rate": 2.041970190371206e-05,
      "loss": 0.0759,
      "step": 4631
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.49345372823805955,
      "learning_rate": 2.041258933012009e-05,
      "loss": 0.0978,
      "step": 4632
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3451715250637295,
      "learning_rate": 2.0405476704324624e-05,
      "loss": 0.0597,
      "step": 4633
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32081362438766614,
      "learning_rate": 2.0398364027225593e-05,
      "loss": 0.0644,
      "step": 4634
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40497959009005763,
      "learning_rate": 2.039125129972295e-05,
      "loss": 0.0887,
      "step": 4635
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3531975553461992,
      "learning_rate": 2.0384138522716626e-05,
      "loss": 0.0745,
      "step": 4636
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3677945090976669,
      "learning_rate": 2.0377025697106587e-05,
      "loss": 0.0661,
      "step": 4637
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34489797593609794,
      "learning_rate": 2.036991282379279e-05,
      "loss": 0.0634,
      "step": 4638
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.36388535671768535,
      "learning_rate": 2.0362799903675214e-05,
      "loss": 0.0777,
      "step": 4639
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.28581562623867446,
      "learning_rate": 2.0355686937653818e-05,
      "loss": 0.0448,
      "step": 4640
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3708941474975803,
      "learning_rate": 2.0348573926628586e-05,
      "loss": 0.0723,
      "step": 4641
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3063727381041822,
      "learning_rate": 2.03414608714995e-05,
      "loss": 0.0475,
      "step": 4642
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3477293430599361,
      "learning_rate": 2.033434777316655e-05,
      "loss": 0.0647,
      "step": 4643
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2592214463032209,
      "learning_rate": 2.0327234632529738e-05,
      "loss": 0.0422,
      "step": 4644
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.290287964238808,
      "learning_rate": 2.0320121450489062e-05,
      "loss": 0.0467,
      "step": 4645
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3955144057489601,
      "learning_rate": 2.0313008227944527e-05,
      "loss": 0.059,
      "step": 4646
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39261266452050225,
      "learning_rate": 2.0305894965796143e-05,
      "loss": 0.0927,
      "step": 4647
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3816971197257481,
      "learning_rate": 2.029878166494393e-05,
      "loss": 0.0732,
      "step": 4648
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31963227314258197,
      "learning_rate": 2.029166832628791e-05,
      "loss": 0.0548,
      "step": 4649
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4018415446319925,
      "learning_rate": 2.0284554950728106e-05,
      "loss": 0.101,
      "step": 4650
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38869851945958717,
      "learning_rate": 2.027744153916455e-05,
      "loss": 0.046,
      "step": 4651
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3655089300084946,
      "learning_rate": 2.0270328092497266e-05,
      "loss": 0.0567,
      "step": 4652
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2826903073957709,
      "learning_rate": 2.026321461162632e-05,
      "loss": 0.0683,
      "step": 4653
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4094382404663676,
      "learning_rate": 2.025610109745173e-05,
      "loss": 0.0747,
      "step": 4654
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3249037985372121,
      "learning_rate": 2.0248987550873566e-05,
      "loss": 0.0486,
      "step": 4655
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34473773624801163,
      "learning_rate": 2.024187397279186e-05,
      "loss": 0.0598,
      "step": 4656
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34700989377101266,
      "learning_rate": 2.0234760364106686e-05,
      "loss": 0.0716,
      "step": 4657
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.364102491548055,
      "learning_rate": 2.0227646725718085e-05,
      "loss": 0.0562,
      "step": 4658
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3672968721841312,
      "learning_rate": 2.0220533058526146e-05,
      "loss": 0.0976,
      "step": 4659
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.33257767089210566,
      "learning_rate": 2.021341936343091e-05,
      "loss": 0.051,
      "step": 4660
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.48849394171363797,
      "learning_rate": 2.0206305641332467e-05,
      "loss": 0.0886,
      "step": 4661
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32436748907547974,
      "learning_rate": 2.0199191893130893e-05,
      "loss": 0.0612,
      "step": 4662
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3713300253933642,
      "learning_rate": 2.0192078119726246e-05,
      "loss": 0.0716,
      "step": 4663
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4341402990927922,
      "learning_rate": 2.018496432201863e-05,
      "loss": 0.088,
      "step": 4664
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34245291828327334,
      "learning_rate": 2.017785050090811e-05,
      "loss": 0.0642,
      "step": 4665
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4056287563720189,
      "learning_rate": 2.017073665729479e-05,
      "loss": 0.0818,
      "step": 4666
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39500812000064944,
      "learning_rate": 2.016362279207875e-05,
      "loss": 0.0762,
      "step": 4667
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4728721251777227,
      "learning_rate": 2.0156508906160083e-05,
      "loss": 0.0875,
      "step": 4668
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3154172865375658,
      "learning_rate": 2.014939500043888e-05,
      "loss": 0.0579,
      "step": 4669
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3254570122780192,
      "learning_rate": 2.0142281075815253e-05,
      "loss": 0.064,
      "step": 4670
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34107747254476783,
      "learning_rate": 2.013516713318929e-05,
      "loss": 0.0484,
      "step": 4671
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.44921266302877566,
      "learning_rate": 2.0128053173461105e-05,
      "loss": 0.0927,
      "step": 4672
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.28672796333332295,
      "learning_rate": 2.012093919753079e-05,
      "loss": 0.0602,
      "step": 4673
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.26762949318211976,
      "learning_rate": 2.0113825206298458e-05,
      "loss": 0.0456,
      "step": 4674
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3094490126500517,
      "learning_rate": 2.0106711200664218e-05,
      "loss": 0.0589,
      "step": 4675
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4557519271253227,
      "learning_rate": 2.009959718152818e-05,
      "loss": 0.0876,
      "step": 4676
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.25029671465645154,
      "learning_rate": 2.0092483149790453e-05,
      "loss": 0.034,
      "step": 4677
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3868847500338196,
      "learning_rate": 2.008536910635115e-05,
      "loss": 0.0908,
      "step": 4678
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31288024894664096,
      "learning_rate": 2.0078255052110393e-05,
      "loss": 0.0475,
      "step": 4679
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3199749074724017,
      "learning_rate": 2.00711409879683e-05,
      "loss": 0.0564,
      "step": 4680
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31253077832287374,
      "learning_rate": 2.0064026914824976e-05,
      "loss": 0.067,
      "step": 4681
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37506683071988134,
      "learning_rate": 2.0056912833580557e-05,
      "loss": 0.0684,
      "step": 4682
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40580266385574204,
      "learning_rate": 2.0049798745135148e-05,
      "loss": 0.1004,
      "step": 4683
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3062097231011995,
      "learning_rate": 2.0042684650388882e-05,
      "loss": 0.0497,
      "step": 4684
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.29951505307784704,
      "learning_rate": 2.003557055024187e-05,
      "loss": 0.0664,
      "step": 4685
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38139977874329367,
      "learning_rate": 2.0028456445594234e-05,
      "loss": 0.0605,
      "step": 4686
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39084394041334164,
      "learning_rate": 2.0021342337346117e-05,
      "loss": 0.0633,
      "step": 4687
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4009913334492643,
      "learning_rate": 2.0014228226397618e-05,
      "loss": 0.0887,
      "step": 4688
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30262971114754145,
      "learning_rate": 2.0007114113648876e-05,
      "loss": 0.0603,
      "step": 4689
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3419804325403742,
      "learning_rate": 2e-05,
      "loss": 0.0627,
      "step": 4690
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3852560529177657,
      "learning_rate": 1.999288588635113e-05,
      "loss": 0.065,
      "step": 4691
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3104228005773945,
      "learning_rate": 1.998577177360239e-05,
      "loss": 0.0423,
      "step": 4692
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3383010400352988,
      "learning_rate": 1.997865766265389e-05,
      "loss": 0.0674,
      "step": 4693
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34263660275430635,
      "learning_rate": 1.997154355440577e-05,
      "loss": 0.0818,
      "step": 4694
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44871075973978786,
      "learning_rate": 1.9964429449758138e-05,
      "loss": 0.1121,
      "step": 4695
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.28127286076354807,
      "learning_rate": 1.995731534961113e-05,
      "loss": 0.0463,
      "step": 4696
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.23295536842010514,
      "learning_rate": 1.995020125486486e-05,
      "loss": 0.0492,
      "step": 4697
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3905686719336237,
      "learning_rate": 1.9943087166419453e-05,
      "loss": 0.066,
      "step": 4698
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40578542334841866,
      "learning_rate": 1.9935973085175024e-05,
      "loss": 0.0641,
      "step": 4699
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2559957237444201,
      "learning_rate": 1.9928859012031703e-05,
      "loss": 0.0979,
      "step": 4700
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3666195230448838,
      "learning_rate": 1.9921744947889607e-05,
      "loss": 0.0617,
      "step": 4701
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35480420381470423,
      "learning_rate": 1.991463089364885e-05,
      "loss": 0.0535,
      "step": 4702
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.30174684967570925,
      "learning_rate": 1.9907516850209554e-05,
      "loss": 0.0467,
      "step": 4703
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2803757909303582,
      "learning_rate": 1.9900402818471825e-05,
      "loss": 0.0603,
      "step": 4704
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3540602435303556,
      "learning_rate": 1.989328879933579e-05,
      "loss": 0.06,
      "step": 4705
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2808325636671105,
      "learning_rate": 1.9886174793701546e-05,
      "loss": 0.0359,
      "step": 4706
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4032442330564709,
      "learning_rate": 1.9879060802469217e-05,
      "loss": 0.0787,
      "step": 4707
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40026851745958963,
      "learning_rate": 1.98719468265389e-05,
      "loss": 0.0736,
      "step": 4708
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3548039098267231,
      "learning_rate": 1.9864832866810713e-05,
      "loss": 0.0617,
      "step": 4709
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3392706017068352,
      "learning_rate": 1.985771892418475e-05,
      "loss": 0.069,
      "step": 4710
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3207754186071412,
      "learning_rate": 1.9850604999561128e-05,
      "loss": 0.0678,
      "step": 4711
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31869558122259933,
      "learning_rate": 1.9843491093839927e-05,
      "loss": 0.0685,
      "step": 4712
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4083369460075834,
      "learning_rate": 1.9836377207921265e-05,
      "loss": 0.0751,
      "step": 4713
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.38764863009422956,
      "learning_rate": 1.982926334270522e-05,
      "loss": 0.054,
      "step": 4714
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.351851387010838,
      "learning_rate": 1.98221494990919e-05,
      "loss": 0.06,
      "step": 4715
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.27996641881555334,
      "learning_rate": 1.9815035677981378e-05,
      "loss": 0.0576,
      "step": 4716
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3828016980262593,
      "learning_rate": 1.9807921880273754e-05,
      "loss": 0.0801,
      "step": 4717
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3588459848136513,
      "learning_rate": 1.9800808106869117e-05,
      "loss": 0.0672,
      "step": 4718
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3661122099992246,
      "learning_rate": 1.9793694358667533e-05,
      "loss": 0.0418,
      "step": 4719
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32928118911616616,
      "learning_rate": 1.9786580636569092e-05,
      "loss": 0.0554,
      "step": 4720
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35582455444822636,
      "learning_rate": 1.977946694147386e-05,
      "loss": 0.0497,
      "step": 4721
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.29228866478177634,
      "learning_rate": 1.9772353274281918e-05,
      "loss": 0.0381,
      "step": 4722
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.26456319132803985,
      "learning_rate": 1.976523963589332e-05,
      "loss": 0.0443,
      "step": 4723
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.26419718653814606,
      "learning_rate": 1.9758126027208146e-05,
      "loss": 0.0372,
      "step": 4724
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3609036813721021,
      "learning_rate": 1.975101244912644e-05,
      "loss": 0.077,
      "step": 4725
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2774167032063228,
      "learning_rate": 1.9743898902548273e-05,
      "loss": 0.0608,
      "step": 4726
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3676231518854293,
      "learning_rate": 1.9736785388373686e-05,
      "loss": 0.0573,
      "step": 4727
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4060585597937581,
      "learning_rate": 1.972967190750274e-05,
      "loss": 0.0744,
      "step": 4728
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31194598441259513,
      "learning_rate": 1.9722558460835463e-05,
      "loss": 0.0762,
      "step": 4729
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3249208707030346,
      "learning_rate": 1.9715445049271907e-05,
      "loss": 0.0428,
      "step": 4730
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40405446769596115,
      "learning_rate": 1.97083316737121e-05,
      "loss": 0.0699,
      "step": 4731
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.41410809391838166,
      "learning_rate": 1.9701218335056076e-05,
      "loss": 0.089,
      "step": 4732
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4039627020988026,
      "learning_rate": 1.9694105034203867e-05,
      "loss": 0.0747,
      "step": 4733
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2972790578446354,
      "learning_rate": 1.9686991772055476e-05,
      "loss": 0.0588,
      "step": 4734
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3602559859726878,
      "learning_rate": 1.967987854951094e-05,
      "loss": 0.0815,
      "step": 4735
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.25467894428603094,
      "learning_rate": 1.9672765367470265e-05,
      "loss": 0.0445,
      "step": 4736
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3154520552413603,
      "learning_rate": 1.966565222683345e-05,
      "loss": 0.0608,
      "step": 4737
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.29233013486326725,
      "learning_rate": 1.9658539128500507e-05,
      "loss": 0.0596,
      "step": 4738
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3780755050044366,
      "learning_rate": 1.965142607337142e-05,
      "loss": 0.0659,
      "step": 4739
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.41711352546143093,
      "learning_rate": 1.964431306234619e-05,
      "loss": 0.0924,
      "step": 4740
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.30895279868877534,
      "learning_rate": 1.9637200096324792e-05,
      "loss": 0.0753,
      "step": 4741
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34942746890999127,
      "learning_rate": 1.9630087176207212e-05,
      "loss": 0.0722,
      "step": 4742
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.33469045188948415,
      "learning_rate": 1.9622974302893416e-05,
      "loss": 0.0618,
      "step": 4743
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4191969002721475,
      "learning_rate": 1.9615861477283384e-05,
      "loss": 0.0561,
      "step": 4744
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3587680542619685,
      "learning_rate": 1.9608748700277058e-05,
      "loss": 0.0643,
      "step": 4745
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3128015017409996,
      "learning_rate": 1.9601635972774414e-05,
      "loss": 0.0642,
      "step": 4746
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3329583408155374,
      "learning_rate": 1.9594523295675383e-05,
      "loss": 0.0877,
      "step": 4747
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.22941014060953932,
      "learning_rate": 1.958741066987992e-05,
      "loss": 0.0475,
      "step": 4748
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39091050204965927,
      "learning_rate": 1.958029809628795e-05,
      "loss": 0.0774,
      "step": 4749
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.30798737791365455,
      "learning_rate": 1.9573185575799414e-05,
      "loss": 0.0626,
      "step": 4750
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.33282125398826806,
      "learning_rate": 1.956607310931422e-05,
      "loss": 0.0495,
      "step": 4751
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3291413144517075,
      "learning_rate": 1.95589606977323e-05,
      "loss": 0.0695,
      "step": 4752
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3011719284354433,
      "learning_rate": 1.9551848341953556e-05,
      "loss": 0.0599,
      "step": 4753
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35773416852809714,
      "learning_rate": 1.9544736042877886e-05,
      "loss": 0.0637,
      "step": 4754
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2714101840949113,
      "learning_rate": 1.9537623801405194e-05,
      "loss": 0.0469,
      "step": 4755
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3227778785101817,
      "learning_rate": 1.9530511618435352e-05,
      "loss": 0.0629,
      "step": 4756
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.332783934213497,
      "learning_rate": 1.9523399494868256e-05,
      "loss": 0.0594,
      "step": 4757
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3196381704804296,
      "learning_rate": 1.9516287431603767e-05,
      "loss": 0.043,
      "step": 4758
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4336684910046564,
      "learning_rate": 1.950917542954176e-05,
      "loss": 0.0809,
      "step": 4759
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3099770145777829,
      "learning_rate": 1.950206348958208e-05,
      "loss": 0.0525,
      "step": 4760
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34155675900687366,
      "learning_rate": 1.9494951612624583e-05,
      "loss": 0.0619,
      "step": 4761
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3155040830387275,
      "learning_rate": 1.948783979956911e-05,
      "loss": 0.0462,
      "step": 4762
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5243635668320893,
      "learning_rate": 1.9480728051315493e-05,
      "loss": 0.0973,
      "step": 4763
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4387000349183044,
      "learning_rate": 1.9473616368763556e-05,
      "loss": 0.0875,
      "step": 4764
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.24595755638297326,
      "learning_rate": 1.9466504752813124e-05,
      "loss": 0.0405,
      "step": 4765
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3528262419005254,
      "learning_rate": 1.9459393204363988e-05,
      "loss": 0.0549,
      "step": 4766
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4792290798948304,
      "learning_rate": 1.9452281724315964e-05,
      "loss": 0.0901,
      "step": 4767
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3106596879760235,
      "learning_rate": 1.944517031356882e-05,
      "loss": 0.0555,
      "step": 4768
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.27229752647867894,
      "learning_rate": 1.9438058973022374e-05,
      "loss": 0.0626,
      "step": 4769
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3372626671077059,
      "learning_rate": 1.9430947703576373e-05,
      "loss": 0.0693,
      "step": 4770
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.28020790211149216,
      "learning_rate": 1.9423836506130593e-05,
      "loss": 0.0477,
      "step": 4771
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2888125163476407,
      "learning_rate": 1.9416725381584777e-05,
      "loss": 0.0471,
      "step": 4772
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3124671322704106,
      "learning_rate": 1.9409614330838685e-05,
      "loss": 0.0715,
      "step": 4773
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.255097412151316,
      "learning_rate": 1.9402503354792043e-05,
      "loss": 0.0522,
      "step": 4774
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.28330472241894744,
      "learning_rate": 1.9395392454344582e-05,
      "loss": 0.0413,
      "step": 4775
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3338411258207654,
      "learning_rate": 1.938828163039602e-05,
      "loss": 0.0625,
      "step": 4776
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.30842513294413454,
      "learning_rate": 1.938117088384607e-05,
      "loss": 0.0533,
      "step": 4777
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2830844498957535,
      "learning_rate": 1.9374060215594417e-05,
      "loss": 0.0444,
      "step": 4778
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2835374634426756,
      "learning_rate": 1.9366949626540764e-05,
      "loss": 0.0501,
      "step": 4779
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.43673514879586744,
      "learning_rate": 1.9359839117584775e-05,
      "loss": 0.0671,
      "step": 4780
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37202761748045315,
      "learning_rate": 1.9352728689626133e-05,
      "loss": 0.0625,
      "step": 4781
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37775016680650036,
      "learning_rate": 1.934561834356448e-05,
      "loss": 0.0656,
      "step": 4782
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39819017849884863,
      "learning_rate": 1.933850808029948e-05,
      "loss": 0.0683,
      "step": 4783
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3313787893728318,
      "learning_rate": 1.9331397900730754e-05,
      "loss": 0.0559,
      "step": 4784
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38201853101031413,
      "learning_rate": 1.9324287805757935e-05,
      "loss": 0.067,
      "step": 4785
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.32680930024413746,
      "learning_rate": 1.9317177796280643e-05,
      "loss": 0.0576,
      "step": 4786
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3409276864189273,
      "learning_rate": 1.9310067873198476e-05,
      "loss": 0.0573,
      "step": 4787
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.28473758009968864,
      "learning_rate": 1.930295803741104e-05,
      "loss": 0.0542,
      "step": 4788
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3075399029241916,
      "learning_rate": 1.9295848289817904e-05,
      "loss": 0.0416,
      "step": 4789
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3559080071827515,
      "learning_rate": 1.9288738631318648e-05,
      "loss": 0.0625,
      "step": 4790
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.32405360738458894,
      "learning_rate": 1.9281629062812827e-05,
      "loss": 0.0595,
      "step": 4791
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4237242716458942,
      "learning_rate": 1.9274519585199995e-05,
      "loss": 0.0909,
      "step": 4792
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41985948675807966,
      "learning_rate": 1.9267410199379682e-05,
      "loss": 0.0826,
      "step": 4793
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3238525390625,
      "learning_rate": 1.9260300906251422e-05,
      "loss": 0.0597,
      "step": 4794
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31939485572187754,
      "learning_rate": 1.9253191706714725e-05,
      "loss": 0.0546,
      "step": 4795
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3659308417778955,
      "learning_rate": 1.92460826016691e-05,
      "loss": 0.0723,
      "step": 4796
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.42463157897560067,
      "learning_rate": 1.9238973592014027e-05,
      "loss": 0.0949,
      "step": 4797
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2526853699614417,
      "learning_rate": 1.9231864678648994e-05,
      "loss": 0.052,
      "step": 4798
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3315363393129195,
      "learning_rate": 1.922475586247346e-05,
      "loss": 0.0821,
      "step": 4799
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.293977273109654,
      "learning_rate": 1.9217647144386885e-05,
      "loss": 0.0635,
      "step": 4800
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31837765225794823,
      "learning_rate": 1.92105385252887e-05,
      "loss": 0.0691,
      "step": 4801
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4162010173787922,
      "learning_rate": 1.9203430006078348e-05,
      "loss": 0.0788,
      "step": 4802
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3249224643651483,
      "learning_rate": 1.9196321587655228e-05,
      "loss": 0.075,
      "step": 4803
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41507273817883844,
      "learning_rate": 1.918921327091876e-05,
      "loss": 0.0713,
      "step": 4804
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3061640492320292,
      "learning_rate": 1.9182105056768335e-05,
      "loss": 0.05,
      "step": 4805
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31932931129043707,
      "learning_rate": 1.9174996946103318e-05,
      "loss": 0.0841,
      "step": 4806
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.309396604785433,
      "learning_rate": 1.9167888939823083e-05,
      "loss": 0.0619,
      "step": 4807
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.25655192637986973,
      "learning_rate": 1.9160781038826973e-05,
      "loss": 0.0526,
      "step": 4808
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.310471339606134,
      "learning_rate": 1.9153673244014334e-05,
      "loss": 0.0579,
      "step": 4809
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.30016985594827106,
      "learning_rate": 1.9146565556284492e-05,
      "loss": 0.0526,
      "step": 4810
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3116027587546304,
      "learning_rate": 1.9139457976536754e-05,
      "loss": 0.0605,
      "step": 4811
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3421233225869482,
      "learning_rate": 1.9132350505670416e-05,
      "loss": 0.065,
      "step": 4812
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.347588886206723,
      "learning_rate": 1.9125243144584762e-05,
      "loss": 0.0791,
      "step": 4813
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.32984032049524,
      "learning_rate": 1.9118135894179067e-05,
      "loss": 0.0533,
      "step": 4814
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.29665145488557776,
      "learning_rate": 1.9111028755352578e-05,
      "loss": 0.0697,
      "step": 4815
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.33921622294253856,
      "learning_rate": 1.910392172900455e-05,
      "loss": 0.0621,
      "step": 4816
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4300904725231287,
      "learning_rate": 1.9096814816034192e-05,
      "loss": 0.0605,
      "step": 4817
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37491111894151646,
      "learning_rate": 1.9089708017340733e-05,
      "loss": 0.0595,
      "step": 4818
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.29845438754649123,
      "learning_rate": 1.9082601333823364e-05,
      "loss": 0.0506,
      "step": 4819
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3584630424624439,
      "learning_rate": 1.9075494766381263e-05,
      "loss": 0.0416,
      "step": 4820
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.32079072469105685,
      "learning_rate": 1.906838831591362e-05,
      "loss": 0.058,
      "step": 4821
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3503191684765387,
      "learning_rate": 1.906128198331957e-05,
      "loss": 0.053,
      "step": 4822
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3260929331931471,
      "learning_rate": 1.9054175769498265e-05,
      "loss": 0.0643,
      "step": 4823
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3772245230594431,
      "learning_rate": 1.9047069675348816e-05,
      "loss": 0.0577,
      "step": 4824
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38575698244718737,
      "learning_rate": 1.903996370177035e-05,
      "loss": 0.0567,
      "step": 4825
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.34618073138954514,
      "learning_rate": 1.9032857849661942e-05,
      "loss": 0.0477,
      "step": 4826
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4194893849829217,
      "learning_rate": 1.902575211992269e-05,
      "loss": 0.0678,
      "step": 4827
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3421051814729771,
      "learning_rate": 1.901864651345164e-05,
      "loss": 0.062,
      "step": 4828
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35553313290571104,
      "learning_rate": 1.9011541031147856e-05,
      "loss": 0.0637,
      "step": 4829
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3159303854340297,
      "learning_rate": 1.9004435673910356e-05,
      "loss": 0.0542,
      "step": 4830
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3224906482560456,
      "learning_rate": 1.8997330442638165e-05,
      "loss": 0.0466,
      "step": 4831
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.306691621075204,
      "learning_rate": 1.8990225338230276e-05,
      "loss": 0.0592,
      "step": 4832
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.32547930897668065,
      "learning_rate": 1.8983120361585683e-05,
      "loss": 0.0556,
      "step": 4833
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.30747488454271754,
      "learning_rate": 1.8976015513603344e-05,
      "loss": 0.0545,
      "step": 4834
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3143407964585679,
      "learning_rate": 1.896891079518222e-05,
      "loss": 0.0511,
      "step": 4835
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3654993085737251,
      "learning_rate": 1.8961806207221235e-05,
      "loss": 0.0689,
      "step": 4836
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3067897142624209,
      "learning_rate": 1.8954701750619313e-05,
      "loss": 0.0631,
      "step": 4837
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3394790534157499,
      "learning_rate": 1.8947597426275368e-05,
      "loss": 0.0446,
      "step": 4838
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.313444688549121,
      "learning_rate": 1.8940493235088272e-05,
      "loss": 0.037,
      "step": 4839
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3423125382598025,
      "learning_rate": 1.8933389177956896e-05,
      "loss": 0.0674,
      "step": 4840
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3556325346512234,
      "learning_rate": 1.8926285255780095e-05,
      "loss": 0.0683,
      "step": 4841
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3509764024397666,
      "learning_rate": 1.8919181469456703e-05,
      "loss": 0.0707,
      "step": 4842
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2654621803319488,
      "learning_rate": 1.8912077819885533e-05,
      "loss": 0.0318,
      "step": 4843
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3555633712425293,
      "learning_rate": 1.8904974307965393e-05,
      "loss": 0.0769,
      "step": 4844
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3514757261210543,
      "learning_rate": 1.8897870934595055e-05,
      "loss": 0.058,
      "step": 4845
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.26660133892790533,
      "learning_rate": 1.8890767700673296e-05,
      "loss": 0.0302,
      "step": 4846
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.30750849180979273,
      "learning_rate": 1.8883664607098856e-05,
      "loss": 0.0595,
      "step": 4847
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.306579535163633,
      "learning_rate": 1.8876561654770466e-05,
      "loss": 0.0486,
      "step": 4848
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.26328504938842306,
      "learning_rate": 1.886945884458684e-05,
      "loss": 0.0374,
      "step": 4849
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.28228724794108406,
      "learning_rate": 1.8862356177446667e-05,
      "loss": 0.0581,
      "step": 4850
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3164625824125481,
      "learning_rate": 1.885525365424863e-05,
      "loss": 0.0504,
      "step": 4851
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3693746803656678,
      "learning_rate": 1.8848151275891383e-05,
      "loss": 0.0612,
      "step": 4852
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.29125390762360004,
      "learning_rate": 1.884104904327357e-05,
      "loss": 0.0514,
      "step": 4853
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3088892354992329,
      "learning_rate": 1.8833946957293796e-05,
      "loss": 0.0418,
      "step": 4854
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2739199061452179,
      "learning_rate": 1.882684501885068e-05,
      "loss": 0.0416,
      "step": 4855
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.27999998835580664,
      "learning_rate": 1.8819743228842806e-05,
      "loss": 0.068,
      "step": 4856
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3631608209653038,
      "learning_rate": 1.881264158816873e-05,
      "loss": 0.0631,
      "step": 4857
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37143338371921714,
      "learning_rate": 1.8805540097727003e-05,
      "loss": 0.067,
      "step": 4858
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.29118905244255,
      "learning_rate": 1.8798438758416147e-05,
      "loss": 0.048,
      "step": 4859
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3400818001995256,
      "learning_rate": 1.8791337571134677e-05,
      "loss": 0.0543,
      "step": 4860
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.504890338539092,
      "learning_rate": 1.8784236536781076e-05,
      "loss": 0.1036,
      "step": 4861
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4121225725564691,
      "learning_rate": 1.877713565625382e-05,
      "loss": 0.0649,
      "step": 4862
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3878110237308656,
      "learning_rate": 1.8770034930451345e-05,
      "loss": 0.0696,
      "step": 4863
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35091748937214295,
      "learning_rate": 1.8762934360272097e-05,
      "loss": 0.0459,
      "step": 4864
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2781925510495585,
      "learning_rate": 1.8755833946614475e-05,
      "loss": 0.0453,
      "step": 4865
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.36971593358902766,
      "learning_rate": 1.8748733690376883e-05,
      "loss": 0.0572,
      "step": 4866
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.30281658902079533,
      "learning_rate": 1.8741633592457677e-05,
      "loss": 0.0611,
      "step": 4867
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.29873236823623794,
      "learning_rate": 1.8734533653755216e-05,
      "loss": 0.0697,
      "step": 4868
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2923893221306974,
      "learning_rate": 1.8727433875167828e-05,
      "loss": 0.0525,
      "step": 4869
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38429033385912215,
      "learning_rate": 1.8720334257593826e-05,
      "loss": 0.067,
      "step": 4870
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31698419040221953,
      "learning_rate": 1.8713234801931494e-05,
      "loss": 0.0511,
      "step": 4871
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30124388451126555,
      "learning_rate": 1.8706135509079103e-05,
      "loss": 0.0606,
      "step": 4872
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.24919668416841298,
      "learning_rate": 1.869903637993491e-05,
      "loss": 0.0409,
      "step": 4873
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29466716554615313,
      "learning_rate": 1.869193741539714e-05,
      "loss": 0.0504,
      "step": 4874
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2730488740250263,
      "learning_rate": 1.8684838616363997e-05,
      "loss": 0.0442,
      "step": 4875
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.40023512262933153,
      "learning_rate": 1.8677739983733666e-05,
      "loss": 0.0526,
      "step": 4876
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.406152438406579,
      "learning_rate": 1.8670641518404322e-05,
      "loss": 0.0864,
      "step": 4877
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29053493457783014,
      "learning_rate": 1.8663543221274096e-05,
      "loss": 0.0486,
      "step": 4878
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3461570561292766,
      "learning_rate": 1.8656445093241122e-05,
      "loss": 0.0627,
      "step": 4879
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.28846751659263103,
      "learning_rate": 1.8649347135203494e-05,
      "loss": 0.0441,
      "step": 4880
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39885192710398476,
      "learning_rate": 1.8642249348059296e-05,
      "loss": 0.0649,
      "step": 4881
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3039924322867829,
      "learning_rate": 1.8635151732706586e-05,
      "loss": 0.0677,
      "step": 4882
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36083028365131947,
      "learning_rate": 1.86280542900434e-05,
      "loss": 0.0727,
      "step": 4883
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5207675479032882,
      "learning_rate": 1.862095702096775e-05,
      "loss": 0.1073,
      "step": 4884
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36658365229255785,
      "learning_rate": 1.8613859926377636e-05,
      "loss": 0.0721,
      "step": 4885
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.257427737642483,
      "learning_rate": 1.860676300717102e-05,
      "loss": 0.0391,
      "step": 4886
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30208050655270113,
      "learning_rate": 1.859966626424586e-05,
      "loss": 0.058,
      "step": 4887
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.524228821613645,
      "learning_rate": 1.8592569698500076e-05,
      "loss": 0.0803,
      "step": 4888
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37559932262356527,
      "learning_rate": 1.8585473310831567e-05,
      "loss": 0.0644,
      "step": 4889
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3664566233155714,
      "learning_rate": 1.8578377102138223e-05,
      "loss": 0.0722,
      "step": 4890
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2763762860097861,
      "learning_rate": 1.857128107331791e-05,
      "loss": 0.0672,
      "step": 4891
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30401333782933376,
      "learning_rate": 1.8564185225268446e-05,
      "loss": 0.0479,
      "step": 4892
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2695667132944537,
      "learning_rate": 1.8557089558887655e-05,
      "loss": 0.0452,
      "step": 4893
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21376175219313176,
      "learning_rate": 1.8549994075073327e-05,
      "loss": 0.0634,
      "step": 4894
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.25682047518176504,
      "learning_rate": 1.8542898774723225e-05,
      "loss": 0.0489,
      "step": 4895
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.28180622359236795,
      "learning_rate": 1.853580365873509e-05,
      "loss": 0.0453,
      "step": 4896
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33053754401600327,
      "learning_rate": 1.8528708728006654e-05,
      "loss": 0.0486,
      "step": 4897
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.28269531498796063,
      "learning_rate": 1.8521613983435604e-05,
      "loss": 0.0505,
      "step": 4898
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3289290748704466,
      "learning_rate": 1.8514519425919617e-05,
      "loss": 0.0694,
      "step": 4899
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29990069812160436,
      "learning_rate": 1.8507425056356338e-05,
      "loss": 0.0543,
      "step": 4900
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3723073489652655,
      "learning_rate": 1.8500330875643405e-05,
      "loss": 0.0699,
      "step": 4901
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3678210859179774,
      "learning_rate": 1.8493236884678405e-05,
      "loss": 0.0831,
      "step": 4902
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2765332184931764,
      "learning_rate": 1.848614308435893e-05,
      "loss": 0.0473,
      "step": 4903
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30530793037235593,
      "learning_rate": 1.847904947558252e-05,
      "loss": 0.049,
      "step": 4904
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3258362831086268,
      "learning_rate": 1.8471956059246717e-05,
      "loss": 0.0662,
      "step": 4905
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2574830551266415,
      "learning_rate": 1.8464862836249014e-05,
      "loss": 0.0487,
      "step": 4906
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.27446116521727665,
      "learning_rate": 1.8457769807486903e-05,
      "loss": 0.0441,
      "step": 4907
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3036688502548473,
      "learning_rate": 1.8450676973857842e-05,
      "loss": 0.0433,
      "step": 4908
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33868591341096527,
      "learning_rate": 1.8443584336259253e-05,
      "loss": 0.0585,
      "step": 4909
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3886296043221401,
      "learning_rate": 1.843649189558855e-05,
      "loss": 0.0664,
      "step": 4910
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3898702960254951,
      "learning_rate": 1.842939965274311e-05,
      "loss": 0.075,
      "step": 4911
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.27507723568981396,
      "learning_rate": 1.8422307608620292e-05,
      "loss": 0.0363,
      "step": 4912
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30849703951331714,
      "learning_rate": 1.8415215764117428e-05,
      "loss": 0.0581,
      "step": 4913
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.31653900658060236,
      "learning_rate": 1.840812412013183e-05,
      "loss": 0.0581,
      "step": 4914
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3325580120927327,
      "learning_rate": 1.8401032677560767e-05,
      "loss": 0.0533,
      "step": 4915
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36970293520695274,
      "learning_rate": 1.8393941437301507e-05,
      "loss": 0.0675,
      "step": 4916
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3585416834977342,
      "learning_rate": 1.838685040025127e-05,
      "loss": 0.0815,
      "step": 4917
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.23879248073882645,
      "learning_rate": 1.8379759567307266e-05,
      "loss": 0.0428,
      "step": 4918
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.43233937456007654,
      "learning_rate": 1.837266893936667e-05,
      "loss": 0.0737,
      "step": 4919
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36011360809702286,
      "learning_rate": 1.8365578517326642e-05,
      "loss": 0.0645,
      "step": 4920
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32384882355768296,
      "learning_rate": 1.83584883020843e-05,
      "loss": 0.0466,
      "step": 4921
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3139658404344843,
      "learning_rate": 1.8351398294536747e-05,
      "loss": 0.0498,
      "step": 4922
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4437869694582102,
      "learning_rate": 1.8344308495581056e-05,
      "loss": 0.0743,
      "step": 4923
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.326073157624937,
      "learning_rate": 1.833721890611428e-05,
      "loss": 0.0687,
      "step": 4924
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3790813120288228,
      "learning_rate": 1.8330129527033433e-05,
      "loss": 0.0823,
      "step": 4925
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4056979610144585,
      "learning_rate": 1.832304035923552e-05,
      "loss": 0.0776,
      "step": 4926
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.46710853866244517,
      "learning_rate": 1.8315951403617496e-05,
      "loss": 0.0872,
      "step": 4927
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3112583287974614,
      "learning_rate": 1.8308862661076313e-05,
      "loss": 0.0673,
      "step": 4928
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36728808871708474,
      "learning_rate": 1.8301774132508876e-05,
      "loss": 0.0767,
      "step": 4929
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36244579353849044,
      "learning_rate": 1.829468581881208e-05,
      "loss": 0.058,
      "step": 4930
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32660534317571355,
      "learning_rate": 1.828759772088278e-05,
      "loss": 0.0625,
      "step": 4931
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.28011761630334936,
      "learning_rate": 1.8280509839617814e-05,
      "loss": 0.0575,
      "step": 4932
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2744294837609691,
      "learning_rate": 1.8273422175913975e-05,
      "loss": 0.044,
      "step": 4933
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2619071823001568,
      "learning_rate": 1.8266334730668054e-05,
      "loss": 0.052,
      "step": 4934
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3509697685794147,
      "learning_rate": 1.8259247504776788e-05,
      "loss": 0.0683,
      "step": 4935
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3643907810543065,
      "learning_rate": 1.8252160499136914e-05,
      "loss": 0.072,
      "step": 4936
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2935391658656578,
      "learning_rate": 1.8245073714645115e-05,
      "loss": 0.0629,
      "step": 4937
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2858745985207618,
      "learning_rate": 1.8237987152198063e-05,
      "loss": 0.0617,
      "step": 4938
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3837281389874902,
      "learning_rate": 1.8230900812692387e-05,
      "loss": 0.0774,
      "step": 4939
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34214365125179147,
      "learning_rate": 1.822381469702471e-05,
      "loss": 0.0744,
      "step": 4940
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.28109274547539487,
      "learning_rate": 1.82167288060916e-05,
      "loss": 0.0489,
      "step": 4941
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.28707807720634265,
      "learning_rate": 1.8209643140789622e-05,
      "loss": 0.0594,
      "step": 4942
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3481909465478844,
      "learning_rate": 1.8202557702015303e-05,
      "loss": 0.0617,
      "step": 4943
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3282151212000431,
      "learning_rate": 1.8195472490665125e-05,
      "loss": 0.07,
      "step": 4944
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4418564247856527,
      "learning_rate": 1.8188387507635568e-05,
      "loss": 0.0819,
      "step": 4945
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3086632396829253,
      "learning_rate": 1.8181302753823064e-05,
      "loss": 0.0698,
      "step": 4946
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33104685546252693,
      "learning_rate": 1.8174218230124024e-05,
      "loss": 0.06,
      "step": 4947
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29924318995650334,
      "learning_rate": 1.8167133937434823e-05,
      "loss": 0.0624,
      "step": 4948
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5391362457694533,
      "learning_rate": 1.8160049876651822e-05,
      "loss": 0.0963,
      "step": 4949
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.40574559662851045,
      "learning_rate": 1.8152966048671334e-05,
      "loss": 0.0682,
      "step": 4950
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3352497401702295,
      "learning_rate": 1.814588245438966e-05,
      "loss": 0.0676,
      "step": 4951
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35281390944166424,
      "learning_rate": 1.813879909470305e-05,
      "loss": 0.0655,
      "step": 4952
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34141333723241074,
      "learning_rate": 1.813171597050775e-05,
      "loss": 0.0639,
      "step": 4953
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3164033300947929,
      "learning_rate": 1.8124633082699956e-05,
      "loss": 0.0551,
      "step": 4954
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4406131269329775,
      "learning_rate": 1.8117550432175845e-05,
      "loss": 0.1033,
      "step": 4955
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3741654009610003,
      "learning_rate": 1.8110468019831553e-05,
      "loss": 0.0705,
      "step": 4956
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36082336635934725,
      "learning_rate": 1.8103385846563202e-05,
      "loss": 0.0838,
      "step": 4957
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3024490734849216,
      "learning_rate": 1.8096303913266864e-05,
      "loss": 0.0837,
      "step": 4958
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36306395235853045,
      "learning_rate": 1.80892222208386e-05,
      "loss": 0.0698,
      "step": 4959
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2878906363871558,
      "learning_rate": 1.808214077017444e-05,
      "loss": 0.0591,
      "step": 4960
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3128928338019312,
      "learning_rate": 1.8075059562170357e-05,
      "loss": 0.0589,
      "step": 4961
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34881266594920257,
      "learning_rate": 1.8067978597722325e-05,
      "loss": 0.0753,
      "step": 4962
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2986322895691649,
      "learning_rate": 1.8060897877726267e-05,
      "loss": 0.0669,
      "step": 4963
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34748635802146344,
      "learning_rate": 1.8053817403078087e-05,
      "loss": 0.0779,
      "step": 4964
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.28748152092964574,
      "learning_rate": 1.8046737174673646e-05,
      "loss": 0.0512,
      "step": 4965
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2604562570359666,
      "learning_rate": 1.8039657193408788e-05,
      "loss": 0.0485,
      "step": 4966
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4140020362324788,
      "learning_rate": 1.803257746017932e-05,
      "loss": 0.0738,
      "step": 4967
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.28237213023797353,
      "learning_rate": 1.8025497975881004e-05,
      "loss": 0.0582,
      "step": 4968
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2752949162820628,
      "learning_rate": 1.8018418741409598e-05,
      "loss": 0.0463,
      "step": 4969
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32686183388709367,
      "learning_rate": 1.8011339757660798e-05,
      "loss": 0.0506,
      "step": 4970
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42394096876319953,
      "learning_rate": 1.8004261025530298e-05,
      "loss": 0.054,
      "step": 4971
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3079854426125247,
      "learning_rate": 1.7997182545913732e-05,
      "loss": 0.0512,
      "step": 4972
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.22254794816337783,
      "learning_rate": 1.7990104319706728e-05,
      "loss": 0.0342,
      "step": 4973
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.35234584427103316,
      "learning_rate": 1.798302634780486e-05,
      "loss": 0.0607,
      "step": 4974
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.29860853716507335,
      "learning_rate": 1.7975948631103676e-05,
      "loss": 0.0424,
      "step": 4975
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31554812165458185,
      "learning_rate": 1.796887117049871e-05,
      "loss": 0.053,
      "step": 4976
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.41391832612717944,
      "learning_rate": 1.796179396688544e-05,
      "loss": 0.0631,
      "step": 4977
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31691040093370726,
      "learning_rate": 1.7954717021159316e-05,
      "loss": 0.0734,
      "step": 4978
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33049958312936023,
      "learning_rate": 1.7947640334215762e-05,
      "loss": 0.0542,
      "step": 4979
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36660041951167466,
      "learning_rate": 1.7940563906950175e-05,
      "loss": 0.0576,
      "step": 4980
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38903620910670594,
      "learning_rate": 1.79334877402579e-05,
      "loss": 0.066,
      "step": 4981
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33668459389343924,
      "learning_rate": 1.7926411835034267e-05,
      "loss": 0.0649,
      "step": 4982
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3713500894302172,
      "learning_rate": 1.791933619217456e-05,
      "loss": 0.0736,
      "step": 4983
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3807993719322671,
      "learning_rate": 1.791226081257404e-05,
      "loss": 0.065,
      "step": 4984
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.35916288999325824,
      "learning_rate": 1.7905185697127924e-05,
      "loss": 0.0677,
      "step": 4985
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43874529311652233,
      "learning_rate": 1.7898110846731415e-05,
      "loss": 0.0778,
      "step": 4986
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4254177186462866,
      "learning_rate": 1.7891036262279653e-05,
      "loss": 0.0618,
      "step": 4987
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39054160181504416,
      "learning_rate": 1.7883961944667772e-05,
      "loss": 0.0774,
      "step": 4988
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2961933064234726,
      "learning_rate": 1.7876887894790856e-05,
      "loss": 0.0558,
      "step": 4989
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3789867630197014,
      "learning_rate": 1.786981411354396e-05,
      "loss": 0.0683,
      "step": 4990
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3350115701648776,
      "learning_rate": 1.7862740601822107e-05,
      "loss": 0.0519,
      "step": 4991
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3956848125230926,
      "learning_rate": 1.7855667360520277e-05,
      "loss": 0.0662,
      "step": 4992
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.23563540425590127,
      "learning_rate": 1.7848594390533442e-05,
      "loss": 0.0556,
      "step": 4993
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4090596025100565,
      "learning_rate": 1.7841521692756497e-05,
      "loss": 0.0663,
      "step": 4994
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3312390469593599,
      "learning_rate": 1.7834449268084343e-05,
      "loss": 0.0536,
      "step": 4995
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3368756922005246,
      "learning_rate": 1.782737711741182e-05,
      "loss": 0.0552,
      "step": 4996
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3748192351491326,
      "learning_rate": 1.782030524163375e-05,
      "loss": 0.0638,
      "step": 4997
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2708162690557804,
      "learning_rate": 1.7813233641644904e-05,
      "loss": 0.0421,
      "step": 4998
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.35805060114826714,
      "learning_rate": 1.7806162318340032e-05,
      "loss": 0.0535,
      "step": 4999
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.27394307943794116,
      "learning_rate": 1.7799091272613843e-05,
      "loss": 0.0406,
      "step": 5000
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2957267263006547,
      "learning_rate": 1.7792020505361015e-05,
      "loss": 0.0569,
      "step": 5001
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3501094034051673,
      "learning_rate": 1.778495001747618e-05,
      "loss": 0.0667,
      "step": 5002
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31723480294700446,
      "learning_rate": 1.7777879809853954e-05,
      "loss": 0.0503,
      "step": 5003
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4188699365490237,
      "learning_rate": 1.7770809883388896e-05,
      "loss": 0.0631,
      "step": 5004
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4369896568824533,
      "learning_rate": 1.7763740238975543e-05,
      "loss": 0.0734,
      "step": 5005
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3307078634995461,
      "learning_rate": 1.775667087750839e-05,
      "loss": 0.049,
      "step": 5006
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2965574322781026,
      "learning_rate": 1.774960179988191e-05,
      "loss": 0.0538,
      "step": 5007
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.29135011485126827,
      "learning_rate": 1.774253300699051e-05,
      "loss": 0.0491,
      "step": 5008
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3324820543587624,
      "learning_rate": 1.773546449972859e-05,
      "loss": 0.0602,
      "step": 5009
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2887968311686442,
      "learning_rate": 1.77283962789905e-05,
      "loss": 0.0426,
      "step": 5010
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32098663149013934,
      "learning_rate": 1.772132834567057e-05,
      "loss": 0.0577,
      "step": 5011
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34888167274651627,
      "learning_rate": 1.771426070066307e-05,
      "loss": 0.0775,
      "step": 5012
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3632791170447296,
      "learning_rate": 1.770719334486225e-05,
      "loss": 0.0723,
      "step": 5013
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2508698767332318,
      "learning_rate": 1.770012627916231e-05,
      "loss": 0.036,
      "step": 5014
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.25521742389344243,
      "learning_rate": 1.7693059504457434e-05,
      "loss": 0.0432,
      "step": 5015
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39834038167899866,
      "learning_rate": 1.768599302164174e-05,
      "loss": 0.062,
      "step": 5016
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2770387423040599,
      "learning_rate": 1.7678926831609344e-05,
      "loss": 0.068,
      "step": 5017
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39384864071118325,
      "learning_rate": 1.7671860935254285e-05,
      "loss": 0.0764,
      "step": 5018
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2637381717551651,
      "learning_rate": 1.7664795333470607e-05,
      "loss": 0.0423,
      "step": 5019
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.24678083658105257,
      "learning_rate": 1.7657730027152286e-05,
      "loss": 0.0388,
      "step": 5020
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.369029598594509,
      "learning_rate": 1.7650665017193272e-05,
      "loss": 0.0621,
      "step": 5021
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.29983160041020623,
      "learning_rate": 1.7643600304487475e-05,
      "loss": 0.0494,
      "step": 5022
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3832615145840354,
      "learning_rate": 1.7636535889928775e-05,
      "loss": 0.0633,
      "step": 5023
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2730398283609901,
      "learning_rate": 1.7629471774410997e-05,
      "loss": 0.0313,
      "step": 5024
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.41043388416108534,
      "learning_rate": 1.762240795882795e-05,
      "loss": 0.0636,
      "step": 5025
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3460301291241678,
      "learning_rate": 1.7615344444073385e-05,
      "loss": 0.0551,
      "step": 5026
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3388215619683655,
      "learning_rate": 1.7608281231041023e-05,
      "loss": 0.0661,
      "step": 5027
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3665732460690765,
      "learning_rate": 1.7601218320624562e-05,
      "loss": 0.0485,
      "step": 5028
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4283326202447252,
      "learning_rate": 1.7594155713717634e-05,
      "loss": 0.063,
      "step": 5029
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3892199232282073,
      "learning_rate": 1.7587093411213856e-05,
      "loss": 0.076,
      "step": 5030
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.24798691392797867,
      "learning_rate": 1.758003141400679e-05,
      "loss": 0.0571,
      "step": 5031
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3331656207928089,
      "learning_rate": 1.7572969722989967e-05,
      "loss": 0.0553,
      "step": 5032
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.45809062217281094,
      "learning_rate": 1.756590833905688e-05,
      "loss": 0.0901,
      "step": 5033
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34441775790379403,
      "learning_rate": 1.755884726310098e-05,
      "loss": 0.0656,
      "step": 5034
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4552007027109417,
      "learning_rate": 1.755178649601568e-05,
      "loss": 0.0816,
      "step": 5035
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2590029138802504,
      "learning_rate": 1.754472603869436e-05,
      "loss": 0.0542,
      "step": 5036
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34422459135497424,
      "learning_rate": 1.753766589203034e-05,
      "loss": 0.0445,
      "step": 5037
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31101417171289175,
      "learning_rate": 1.7530606056916935e-05,
      "loss": 0.0694,
      "step": 5038
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.27705897903012217,
      "learning_rate": 1.7523546534247388e-05,
      "loss": 0.0496,
      "step": 5039
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2658328897722623,
      "learning_rate": 1.751648732491493e-05,
      "loss": 0.0611,
      "step": 5040
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3552021032590606,
      "learning_rate": 1.750942842981272e-05,
      "loss": 0.0652,
      "step": 5041
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.28318241964483964,
      "learning_rate": 1.7502369849833908e-05,
      "loss": 0.0483,
      "step": 5042
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.355538770046406,
      "learning_rate": 1.7495311585871587e-05,
      "loss": 0.0601,
      "step": 5043
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2597771548817189,
      "learning_rate": 1.748825363881881e-05,
      "loss": 0.0416,
      "step": 5044
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3367107496616662,
      "learning_rate": 1.7481196009568606e-05,
      "loss": 0.0842,
      "step": 5045
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32081127875352194,
      "learning_rate": 1.7474138699013953e-05,
      "loss": 0.0598,
      "step": 5046
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2257297437320428,
      "learning_rate": 1.7467081708047775e-05,
      "loss": 0.0467,
      "step": 5047
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.30081102298173434,
      "learning_rate": 1.746002503756298e-05,
      "loss": 0.0499,
      "step": 5048
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2902564701382669,
      "learning_rate": 1.7452968688452418e-05,
      "loss": 0.0501,
      "step": 5049
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2717310655993475,
      "learning_rate": 1.7445912661608912e-05,
      "loss": 0.0437,
      "step": 5050
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34418557487269436,
      "learning_rate": 1.743885695792522e-05,
      "loss": 0.0519,
      "step": 5051
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3535292817739349,
      "learning_rate": 1.7431801578294097e-05,
      "loss": 0.0736,
      "step": 5052
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4060211087022254,
      "learning_rate": 1.7424746523608218e-05,
      "loss": 0.0762,
      "step": 5053
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3166670842648982,
      "learning_rate": 1.7417691794760247e-05,
      "loss": 0.0503,
      "step": 5054
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3585281345282278,
      "learning_rate": 1.7410637392642787e-05,
      "loss": 0.0554,
      "step": 5055
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3892928678553993,
      "learning_rate": 1.740358331814841e-05,
      "loss": 0.0568,
      "step": 5056
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3609059522308646,
      "learning_rate": 1.7396529572169642e-05,
      "loss": 0.0631,
      "step": 5057
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3522829092070097,
      "learning_rate": 1.7389476155598974e-05,
      "loss": 0.0559,
      "step": 5058
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30681902561408875,
      "learning_rate": 1.738242306932884e-05,
      "loss": 0.0569,
      "step": 5059
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3082430693991221,
      "learning_rate": 1.7375370314251657e-05,
      "loss": 0.0491,
      "step": 5060
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34872301788617277,
      "learning_rate": 1.7368317891259767e-05,
      "loss": 0.075,
      "step": 5061
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31598101403973267,
      "learning_rate": 1.7361265801245504e-05,
      "loss": 0.054,
      "step": 5062
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3369748046962167,
      "learning_rate": 1.7354214045101146e-05,
      "loss": 0.0548,
      "step": 5063
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2763976225253177,
      "learning_rate": 1.7347162623718913e-05,
      "loss": 0.0724,
      "step": 5064
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34208588506666104,
      "learning_rate": 1.7340111537991015e-05,
      "loss": 0.0628,
      "step": 5065
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3640915846769672,
      "learning_rate": 1.7333060788809582e-05,
      "loss": 0.044,
      "step": 5066
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3125153060979357,
      "learning_rate": 1.732601037706674e-05,
      "loss": 0.0394,
      "step": 5067
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3355524606650584,
      "learning_rate": 1.7318960303654534e-05,
      "loss": 0.0554,
      "step": 5068
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3964011006214421,
      "learning_rate": 1.7311910569465004e-05,
      "loss": 0.0787,
      "step": 5069
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4341759939375533,
      "learning_rate": 1.7304861175390112e-05,
      "loss": 0.0751,
      "step": 5070
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24737352304080354,
      "learning_rate": 1.7297812122321805e-05,
      "loss": 0.0433,
      "step": 5071
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34009618262966995,
      "learning_rate": 1.729076341115197e-05,
      "loss": 0.0922,
      "step": 5072
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2490521881859778,
      "learning_rate": 1.728371504277246e-05,
      "loss": 0.0561,
      "step": 5073
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3320103807061644,
      "learning_rate": 1.7276667018075073e-05,
      "loss": 0.0692,
      "step": 5074
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.26887293598725337,
      "learning_rate": 1.726961933795158e-05,
      "loss": 0.0426,
      "step": 5075
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3285078812093594,
      "learning_rate": 1.726257200329369e-05,
      "loss": 0.0636,
      "step": 5076
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2534061201262015,
      "learning_rate": 1.7255525014993092e-05,
      "loss": 0.047,
      "step": 5077
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30155677641739015,
      "learning_rate": 1.72484783739414e-05,
      "loss": 0.0635,
      "step": 5078
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35499942104534166,
      "learning_rate": 1.7241432081030206e-05,
      "loss": 0.0609,
      "step": 5079
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31672678123796466,
      "learning_rate": 1.7234386137151067e-05,
      "loss": 0.0481,
      "step": 5080
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24340283876742694,
      "learning_rate": 1.7227340543195466e-05,
      "loss": 0.0328,
      "step": 5081
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.29442590099299537,
      "learning_rate": 1.7220295300054867e-05,
      "loss": 0.0644,
      "step": 5082
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31362230709724537,
      "learning_rate": 1.7213250408620674e-05,
      "loss": 0.0538,
      "step": 5083
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34773400326608034,
      "learning_rate": 1.7206205869784254e-05,
      "loss": 0.0625,
      "step": 5084
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.25303537230156536,
      "learning_rate": 1.7199161684436933e-05,
      "loss": 0.0529,
      "step": 5085
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30992558547525695,
      "learning_rate": 1.719211785346998e-05,
      "loss": 0.0391,
      "step": 5086
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2615875940351599,
      "learning_rate": 1.7185074377774633e-05,
      "loss": 0.0447,
      "step": 5087
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4372570691712308,
      "learning_rate": 1.717803125824207e-05,
      "loss": 0.0812,
      "step": 5088
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2959408997135996,
      "learning_rate": 1.7170988495763443e-05,
      "loss": 0.0532,
      "step": 5089
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.281115897314965,
      "learning_rate": 1.716394609122984e-05,
      "loss": 0.0528,
      "step": 5090
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.32188896361867053,
      "learning_rate": 1.715690404553232e-05,
      "loss": 0.0504,
      "step": 5091
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.21733115063038813,
      "learning_rate": 1.714986235956188e-05,
      "loss": 0.0448,
      "step": 5092
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36235993984517734,
      "learning_rate": 1.7142821034209488e-05,
      "loss": 0.0635,
      "step": 5093
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3292087640682664,
      "learning_rate": 1.713578007036605e-05,
      "loss": 0.0678,
      "step": 5094
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.49258449226638445,
      "learning_rate": 1.7128739468922445e-05,
      "loss": 0.0901,
      "step": 5095
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.32377849700143196,
      "learning_rate": 1.712169923076948e-05,
      "loss": 0.0585,
      "step": 5096
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2785186863713594,
      "learning_rate": 1.711465935679795e-05,
      "loss": 0.0538,
      "step": 5097
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.22405879499120163,
      "learning_rate": 1.710761984789858e-05,
      "loss": 0.0422,
      "step": 5098
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.45103246012829096,
      "learning_rate": 1.7100580704962048e-05,
      "loss": 0.1284,
      "step": 5099
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37361622846841774,
      "learning_rate": 1.7093541928879004e-05,
      "loss": 0.0636,
      "step": 5100
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2924586241752771,
      "learning_rate": 1.7086503520540027e-05,
      "loss": 0.0509,
      "step": 5101
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3704349693558163,
      "learning_rate": 1.7079465480835677e-05,
      "loss": 0.0663,
      "step": 5102
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24524600986662592,
      "learning_rate": 1.7072427810656436e-05,
      "loss": 0.0426,
      "step": 5103
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.22810341262576617,
      "learning_rate": 1.7065390510892767e-05,
      "loss": 0.0539,
      "step": 5104
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34417395024447445,
      "learning_rate": 1.7058353582435075e-05,
      "loss": 0.0665,
      "step": 5105
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4585472622757748,
      "learning_rate": 1.7051317026173715e-05,
      "loss": 0.0674,
      "step": 5106
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3246709129495116,
      "learning_rate": 1.7044280842998998e-05,
      "loss": 0.0651,
      "step": 5107
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2828797302634436,
      "learning_rate": 1.703724503380119e-05,
      "loss": 0.0435,
      "step": 5108
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.29581321773284813,
      "learning_rate": 1.7030209599470505e-05,
      "loss": 0.0427,
      "step": 5109
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30279689266347104,
      "learning_rate": 1.7023174540897112e-05,
      "loss": 0.0486,
      "step": 5110
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35530447045801083,
      "learning_rate": 1.7016139858971136e-05,
      "loss": 0.062,
      "step": 5111
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3006020668664625,
      "learning_rate": 1.7009105554582652e-05,
      "loss": 0.058,
      "step": 5112
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3179507294968168,
      "learning_rate": 1.7002071628621675e-05,
      "loss": 0.0577,
      "step": 5113
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31721588439969883,
      "learning_rate": 1.6995038081978193e-05,
      "loss": 0.0669,
      "step": 5114
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2544246955897499,
      "learning_rate": 1.6988004915542143e-05,
      "loss": 0.0474,
      "step": 5115
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.33258999205429257,
      "learning_rate": 1.6980972130203396e-05,
      "loss": 0.0505,
      "step": 5116
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30920680030220316,
      "learning_rate": 1.697393972685179e-05,
      "loss": 0.0484,
      "step": 5117
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.29906068117710727,
      "learning_rate": 1.6966907706377103e-05,
      "loss": 0.0587,
      "step": 5118
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3245564620758247,
      "learning_rate": 1.6959876069669088e-05,
      "loss": 0.0674,
      "step": 5119
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31947071835832147,
      "learning_rate": 1.695284481761742e-05,
      "loss": 0.0501,
      "step": 5120
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3421745829125585,
      "learning_rate": 1.6945813951111746e-05,
      "loss": 0.0643,
      "step": 5121
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36055704269868233,
      "learning_rate": 1.6938783471041647e-05,
      "loss": 0.0628,
      "step": 5122
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.27731016788529445,
      "learning_rate": 1.6931753378296682e-05,
      "loss": 0.0571,
      "step": 5123
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3051614010283815,
      "learning_rate": 1.692472367376633e-05,
      "loss": 0.0428,
      "step": 5124
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.405505525247958,
      "learning_rate": 1.691769435834004e-05,
      "loss": 0.1074,
      "step": 5125
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4090393116952397,
      "learning_rate": 1.691066543290721e-05,
      "loss": 0.0775,
      "step": 5126
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.26558222145727184,
      "learning_rate": 1.6903636898357185e-05,
      "loss": 0.042,
      "step": 5127
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.25346701851094533,
      "learning_rate": 1.6896608755579256e-05,
      "loss": 0.0633,
      "step": 5128
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3878416654761144,
      "learning_rate": 1.688958100546267e-05,
      "loss": 0.0642,
      "step": 5129
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3484256460114992,
      "learning_rate": 1.6882553648896625e-05,
      "loss": 0.0562,
      "step": 5130
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2652885606387304,
      "learning_rate": 1.6875526686770277e-05,
      "loss": 0.0408,
      "step": 5131
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.28107621863987486,
      "learning_rate": 1.686850011997271e-05,
      "loss": 0.0459,
      "step": 5132
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3064934201039877,
      "learning_rate": 1.6861473949392983e-05,
      "loss": 0.051,
      "step": 5133
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30573218320867157,
      "learning_rate": 1.685444817592008e-05,
      "loss": 0.0843,
      "step": 5134
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3913446473091023,
      "learning_rate": 1.684742280044296e-05,
      "loss": 0.0874,
      "step": 5135
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3137450093750119,
      "learning_rate": 1.6840397823850513e-05,
      "loss": 0.0474,
      "step": 5136
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30506975697312927,
      "learning_rate": 1.683337324703159e-05,
      "loss": 0.0526,
      "step": 5137
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30696955733225617,
      "learning_rate": 1.6826349070874973e-05,
      "loss": 0.0483,
      "step": 5138
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.42842293974038276,
      "learning_rate": 1.6819325296269426e-05,
      "loss": 0.0723,
      "step": 5139
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2776156825074066,
      "learning_rate": 1.6812301924103626e-05,
      "loss": 0.0525,
      "step": 5140
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.44516781916100806,
      "learning_rate": 1.6805278955266227e-05,
      "loss": 0.0612,
      "step": 5141
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24392622116309437,
      "learning_rate": 1.6798256390645816e-05,
      "loss": 0.0463,
      "step": 5142
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35361049496138447,
      "learning_rate": 1.6791234231130937e-05,
      "loss": 0.0574,
      "step": 5143
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2711724334772591,
      "learning_rate": 1.6784212477610075e-05,
      "loss": 0.0446,
      "step": 5144
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29333892324297994,
      "learning_rate": 1.677719113097167e-05,
      "loss": 0.0356,
      "step": 5145
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.24203772682108785,
      "learning_rate": 1.6770170192104107e-05,
      "loss": 0.0452,
      "step": 5146
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29045960734621556,
      "learning_rate": 1.676314966189573e-05,
      "loss": 0.0535,
      "step": 5147
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.23727784181556422,
      "learning_rate": 1.67561295412348e-05,
      "loss": 0.0537,
      "step": 5148
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2979947855816416,
      "learning_rate": 1.6749109831009574e-05,
      "loss": 0.0528,
      "step": 5149
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3568537631956328,
      "learning_rate": 1.6742090532108228e-05,
      "loss": 0.0446,
      "step": 5150
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30903699887313796,
      "learning_rate": 1.6735071645418874e-05,
      "loss": 0.0504,
      "step": 5151
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4041219876687711,
      "learning_rate": 1.6728053171829603e-05,
      "loss": 0.0789,
      "step": 5152
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4368678191690414,
      "learning_rate": 1.6721035112228427e-05,
      "loss": 0.0816,
      "step": 5153
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6513108396392764,
      "learning_rate": 1.6714017467503328e-05,
      "loss": 0.1534,
      "step": 5154
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.28480162844272705,
      "learning_rate": 1.6707000238542213e-05,
      "loss": 0.0395,
      "step": 5155
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4079935731498508,
      "learning_rate": 1.6699983426232955e-05,
      "loss": 0.0733,
      "step": 5156
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2851944205459965,
      "learning_rate": 1.669296703146336e-05,
      "loss": 0.0497,
      "step": 5157
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4196943154424278,
      "learning_rate": 1.6685951055121203e-05,
      "loss": 0.0627,
      "step": 5158
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37183717246761716,
      "learning_rate": 1.667893549809417e-05,
      "loss": 0.0782,
      "step": 5159
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.26008156514893405,
      "learning_rate": 1.667192036126993e-05,
      "loss": 0.0414,
      "step": 5160
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.25079922775050034,
      "learning_rate": 1.666490564553608e-05,
      "loss": 0.0392,
      "step": 5161
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.22007195347266878,
      "learning_rate": 1.665789135178017e-05,
      "loss": 0.0436,
      "step": 5162
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39780031004251604,
      "learning_rate": 1.6650877480889685e-05,
      "loss": 0.0712,
      "step": 5163
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2112713485432251,
      "learning_rate": 1.664386403375208e-05,
      "loss": 0.0361,
      "step": 5164
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2686055915445793,
      "learning_rate": 1.6636851011254722e-05,
      "loss": 0.0449,
      "step": 5165
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29202897978706954,
      "learning_rate": 1.6629838414284972e-05,
      "loss": 0.0637,
      "step": 5166
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3112071234113945,
      "learning_rate": 1.662282624373009e-05,
      "loss": 0.0527,
      "step": 5167
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2880934762776518,
      "learning_rate": 1.6615814500477307e-05,
      "loss": 0.0452,
      "step": 5168
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2925294252413495,
      "learning_rate": 1.6608803185413792e-05,
      "loss": 0.0417,
      "step": 5169
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.320840679224972,
      "learning_rate": 1.6601792299426668e-05,
      "loss": 0.0546,
      "step": 5170
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30981612449297374,
      "learning_rate": 1.659478184340299e-05,
      "loss": 0.051,
      "step": 5171
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30845845552273965,
      "learning_rate": 1.658777181822978e-05,
      "loss": 0.0573,
      "step": 5172
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.315106666439429,
      "learning_rate": 1.6580762224793977e-05,
      "loss": 0.0407,
      "step": 5173
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29445459597759877,
      "learning_rate": 1.6573753063982492e-05,
      "loss": 0.054,
      "step": 5174
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.42009000739045305,
      "learning_rate": 1.6566744336682158e-05,
      "loss": 0.0932,
      "step": 5175
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.32205595651882796,
      "learning_rate": 1.655973604377978e-05,
      "loss": 0.0587,
      "step": 5176
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.27161409860179414,
      "learning_rate": 1.655272818616208e-05,
      "loss": 0.0381,
      "step": 5177
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37719153736295646,
      "learning_rate": 1.6545720764715746e-05,
      "loss": 0.058,
      "step": 5178
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36650687912005386,
      "learning_rate": 1.6538713780327395e-05,
      "loss": 0.0565,
      "step": 5179
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2800599990150764,
      "learning_rate": 1.6531707233883607e-05,
      "loss": 0.0527,
      "step": 5180
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34092566493233095,
      "learning_rate": 1.652470112627089e-05,
      "loss": 0.063,
      "step": 5181
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38688405674947124,
      "learning_rate": 1.651769545837569e-05,
      "loss": 0.0901,
      "step": 5182
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2778816648105283,
      "learning_rate": 1.6510690231084436e-05,
      "loss": 0.0454,
      "step": 5183
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.31072037126934776,
      "learning_rate": 1.650368544528346e-05,
      "loss": 0.0358,
      "step": 5184
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2944523187000291,
      "learning_rate": 1.6496681101859055e-05,
      "loss": 0.0516,
      "step": 5185
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.22673928006823216,
      "learning_rate": 1.6489677201697453e-05,
      "loss": 0.0329,
      "step": 5186
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37288048913248606,
      "learning_rate": 1.6482673745684842e-05,
      "loss": 0.0786,
      "step": 5187
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2682297700424568,
      "learning_rate": 1.6475670734707336e-05,
      "loss": 0.0485,
      "step": 5188
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30229954408695797,
      "learning_rate": 1.6468668169651012e-05,
      "loss": 0.0493,
      "step": 5189
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33750033025372206,
      "learning_rate": 1.6461666051401865e-05,
      "loss": 0.0475,
      "step": 5190
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3172191726205247,
      "learning_rate": 1.6454664380845862e-05,
      "loss": 0.0368,
      "step": 5191
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3362170320607557,
      "learning_rate": 1.6447663158868897e-05,
      "loss": 0.061,
      "step": 5192
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3913102053063349,
      "learning_rate": 1.644066238635681e-05,
      "loss": 0.0719,
      "step": 5193
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2654376210535113,
      "learning_rate": 1.6433662064195378e-05,
      "loss": 0.0334,
      "step": 5194
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34168859674079904,
      "learning_rate": 1.6426662193270336e-05,
      "loss": 0.0631,
      "step": 5195
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5171339036074991,
      "learning_rate": 1.641966277446735e-05,
      "loss": 0.1306,
      "step": 5196
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2924594393957121,
      "learning_rate": 1.6412663808672036e-05,
      "loss": 0.037,
      "step": 5197
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2529918638417605,
      "learning_rate": 1.6405665296769942e-05,
      "loss": 0.0448,
      "step": 5198
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.46661355796395015,
      "learning_rate": 1.6398667239646565e-05,
      "loss": 0.0877,
      "step": 5199
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30756346218281877,
      "learning_rate": 1.6391669638187355e-05,
      "loss": 0.0575,
      "step": 5200
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30400892646288813,
      "learning_rate": 1.6384672493277686e-05,
      "loss": 0.0449,
      "step": 5201
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.23404778049720912,
      "learning_rate": 1.6377675805802882e-05,
      "loss": 0.0515,
      "step": 5202
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.35008754742232134,
      "learning_rate": 1.637067957664822e-05,
      "loss": 0.0566,
      "step": 5203
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.24880728222613388,
      "learning_rate": 1.6363683806698896e-05,
      "loss": 0.0511,
      "step": 5204
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3532039680631212,
      "learning_rate": 1.635668849684007e-05,
      "loss": 0.0772,
      "step": 5205
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.27099999777241385,
      "learning_rate": 1.6349693647956824e-05,
      "loss": 0.0651,
      "step": 5206
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3232851028765911,
      "learning_rate": 1.6342699260934204e-05,
      "loss": 0.0569,
      "step": 5207
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3277576524654408,
      "learning_rate": 1.6335705336657176e-05,
      "loss": 0.0578,
      "step": 5208
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2576006249923336,
      "learning_rate": 1.6328711876010664e-05,
      "loss": 0.0472,
      "step": 5209
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3178874771888565,
      "learning_rate": 1.632171887987952e-05,
      "loss": 0.057,
      "step": 5210
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2348889200091216,
      "learning_rate": 1.631472634914855e-05,
      "loss": 0.0359,
      "step": 5211
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3040923272153393,
      "learning_rate": 1.6307734284702484e-05,
      "loss": 0.0711,
      "step": 5212
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.31607798042503377,
      "learning_rate": 1.6300742687426017e-05,
      "loss": 0.0595,
      "step": 5213
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3404649887997634,
      "learning_rate": 1.6293751558203764e-05,
      "loss": 0.0446,
      "step": 5214
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.27981268995260267,
      "learning_rate": 1.628676089792029e-05,
      "loss": 0.0532,
      "step": 5215
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33296990947977984,
      "learning_rate": 1.6279770707460096e-05,
      "loss": 0.0532,
      "step": 5216
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.22673408817906404,
      "learning_rate": 1.6272780987707623e-05,
      "loss": 0.0375,
      "step": 5217
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30006792273107746,
      "learning_rate": 1.6265791739547276e-05,
      "loss": 0.0613,
      "step": 5218
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.31690902559257167,
      "learning_rate": 1.625880296386336e-05,
      "loss": 0.0374,
      "step": 5219
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29312967331257844,
      "learning_rate": 1.625181466154015e-05,
      "loss": 0.0538,
      "step": 5220
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3474199156108664,
      "learning_rate": 1.6244826833461848e-05,
      "loss": 0.0535,
      "step": 5221
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3906577859471026,
      "learning_rate": 1.62378394805126e-05,
      "loss": 0.061,
      "step": 5222
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3951905850373451,
      "learning_rate": 1.6230852603576494e-05,
      "loss": 0.0654,
      "step": 5223
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2920592878423403,
      "learning_rate": 1.6223866203537558e-05,
      "loss": 0.0481,
      "step": 5224
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3050223735513972,
      "learning_rate": 1.6216880281279746e-05,
      "loss": 0.0493,
      "step": 5225
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3949634714981901,
      "learning_rate": 1.6209894837686974e-05,
      "loss": 0.0628,
      "step": 5226
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.27627084623194414,
      "learning_rate": 1.6202909873643075e-05,
      "loss": 0.0425,
      "step": 5227
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3592032353791254,
      "learning_rate": 1.6195925390031845e-05,
      "loss": 0.0579,
      "step": 5228
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.35142157167147936,
      "learning_rate": 1.6188941387736994e-05,
      "loss": 0.0521,
      "step": 5229
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2822674388389052,
      "learning_rate": 1.61819578676422e-05,
      "loss": 0.0343,
      "step": 5230
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33143301532436864,
      "learning_rate": 1.6174974830631044e-05,
      "loss": 0.0644,
      "step": 5231
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.25228726608151364,
      "learning_rate": 1.616799227758708e-05,
      "loss": 0.0472,
      "step": 5232
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2882526884946151,
      "learning_rate": 1.6161010209393775e-05,
      "loss": 0.0529,
      "step": 5233
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.26813759058399117,
      "learning_rate": 1.6154028626934548e-05,
      "loss": 0.0329,
      "step": 5234
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.261741637182857,
      "learning_rate": 1.6147047531092767e-05,
      "loss": 0.0421,
      "step": 5235
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.38538763435234974,
      "learning_rate": 1.6140066922751715e-05,
      "loss": 0.0776,
      "step": 5236
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2920774507587756,
      "learning_rate": 1.6133086802794627e-05,
      "loss": 0.0462,
      "step": 5237
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3395953531224787,
      "learning_rate": 1.612610717210467e-05,
      "loss": 0.0755,
      "step": 5238
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2589390302175967,
      "learning_rate": 1.6119128031564963e-05,
      "loss": 0.0582,
      "step": 5239
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3004705012841251,
      "learning_rate": 1.611214938205854e-05,
      "loss": 0.0454,
      "step": 5240
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.23203440834233513,
      "learning_rate": 1.6105171224468393e-05,
      "loss": 0.0463,
      "step": 5241
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33325620096594455,
      "learning_rate": 1.609819355967744e-05,
      "loss": 0.0455,
      "step": 5242
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.40789007201830385,
      "learning_rate": 1.6091216388568534e-05,
      "loss": 0.0724,
      "step": 5243
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.25309588888922263,
      "learning_rate": 1.6084239712024492e-05,
      "loss": 0.042,
      "step": 5244
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37112516972893506,
      "learning_rate": 1.6077263530928032e-05,
      "loss": 0.0652,
      "step": 5245
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29071635020640363,
      "learning_rate": 1.6070287846161834e-05,
      "loss": 0.0439,
      "step": 5246
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29490018600643625,
      "learning_rate": 1.6063312658608498e-05,
      "loss": 0.0464,
      "step": 5247
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34270070056860374,
      "learning_rate": 1.6056337969150584e-05,
      "loss": 0.0442,
      "step": 5248
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.27152861090681607,
      "learning_rate": 1.6049363778670563e-05,
      "loss": 0.0548,
      "step": 5249
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.31559746593822635,
      "learning_rate": 1.6042390088050864e-05,
      "loss": 0.0564,
      "step": 5250
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5388626059724194,
      "learning_rate": 1.603541689817383e-05,
      "loss": 0.0977,
      "step": 5251
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2396804968446206,
      "learning_rate": 1.6028444209921775e-05,
      "loss": 0.0301,
      "step": 5252
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.23117198981475298,
      "learning_rate": 1.6021472024176915e-05,
      "loss": 0.034,
      "step": 5253
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29093740039172244,
      "learning_rate": 1.601450034182142e-05,
      "loss": 0.0503,
      "step": 5254
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.30495534030512106,
      "learning_rate": 1.60075291637374e-05,
      "loss": 0.0608,
      "step": 5255
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2548163754924262,
      "learning_rate": 1.6000558490806877e-05,
      "loss": 0.0618,
      "step": 5256
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2636368763619328,
      "learning_rate": 1.5993588323911843e-05,
      "loss": 0.0368,
      "step": 5257
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.22409755608120996,
      "learning_rate": 1.59866186639342e-05,
      "loss": 0.0407,
      "step": 5258
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.25858331814840774,
      "learning_rate": 1.5979649511755798e-05,
      "loss": 0.0434,
      "step": 5259
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3853648597677454,
      "learning_rate": 1.597268086825842e-05,
      "loss": 0.0699,
      "step": 5260
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3070833812262413,
      "learning_rate": 1.596571273432378e-05,
      "loss": 0.0516,
      "step": 5261
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3222657174774962,
      "learning_rate": 1.5958745110833536e-05,
      "loss": 0.0626,
      "step": 5262
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2776293425866597,
      "learning_rate": 1.595177799866928e-05,
      "loss": 0.0495,
      "step": 5263
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37201421920585737,
      "learning_rate": 1.5944811398712527e-05,
      "loss": 0.0739,
      "step": 5264
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.40617995391870426,
      "learning_rate": 1.593784531184475e-05,
      "loss": 0.0821,
      "step": 5265
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.25384002702258873,
      "learning_rate": 1.5930879738947328e-05,
      "loss": 0.0513,
      "step": 5266
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2683584154936449,
      "learning_rate": 1.5923914680901607e-05,
      "loss": 0.0699,
      "step": 5267
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3641223399873448,
      "learning_rate": 1.5916950138588834e-05,
      "loss": 0.0544,
      "step": 5268
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29739697896510703,
      "learning_rate": 1.590998611289022e-05,
      "loss": 0.0585,
      "step": 5269
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2491987097763603,
      "learning_rate": 1.5903022604686908e-05,
      "loss": 0.0509,
      "step": 5270
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.28130431445505416,
      "learning_rate": 1.5896059614859946e-05,
      "loss": 0.0435,
      "step": 5271
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.28450910738696344,
      "learning_rate": 1.5889097144290357e-05,
      "loss": 0.049,
      "step": 5272
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2610857125762479,
      "learning_rate": 1.588213519385906e-05,
      "loss": 0.0574,
      "step": 5273
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29081575839727775,
      "learning_rate": 1.587517376444694e-05,
      "loss": 0.0536,
      "step": 5274
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.368844049389136,
      "learning_rate": 1.5868212856934794e-05,
      "loss": 0.0665,
      "step": 5275
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.24836798781070157,
      "learning_rate": 1.5861252472203367e-05,
      "loss": 0.0401,
      "step": 5276
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34309871704942396,
      "learning_rate": 1.5854292611133326e-05,
      "loss": 0.0694,
      "step": 5277
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2994736574191442,
      "learning_rate": 1.5847333274605286e-05,
      "loss": 0.045,
      "step": 5278
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37943419810030143,
      "learning_rate": 1.5840374463499784e-05,
      "loss": 0.0826,
      "step": 5279
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4250557869873433,
      "learning_rate": 1.5833416178697298e-05,
      "loss": 0.0943,
      "step": 5280
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.26947392669575115,
      "learning_rate": 1.5826458421078225e-05,
      "loss": 0.034,
      "step": 5281
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33450566706682255,
      "learning_rate": 1.5819501191522917e-05,
      "loss": 0.0567,
      "step": 5282
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.26646218481752504,
      "learning_rate": 1.581254449091164e-05,
      "loss": 0.042,
      "step": 5283
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35952295491887815,
      "learning_rate": 1.5805588320124607e-05,
      "loss": 0.0549,
      "step": 5284
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.30926456455127344,
      "learning_rate": 1.579863268004196e-05,
      "loss": 0.066,
      "step": 5285
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3250968944555532,
      "learning_rate": 1.5791677571543762e-05,
      "loss": 0.058,
      "step": 5286
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39262961054769485,
      "learning_rate": 1.5784722995510025e-05,
      "loss": 0.0876,
      "step": 5287
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3643003546286613,
      "learning_rate": 1.5777768952820697e-05,
      "loss": 0.07,
      "step": 5288
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3646054692587407,
      "learning_rate": 1.5770815444355635e-05,
      "loss": 0.0853,
      "step": 5289
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3955555451399972,
      "learning_rate": 1.576386247099465e-05,
      "loss": 0.0453,
      "step": 5290
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37631812495937944,
      "learning_rate": 1.575691003361747e-05,
      "loss": 0.0757,
      "step": 5291
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3075492178287895,
      "learning_rate": 1.5749958133103772e-05,
      "loss": 0.0761,
      "step": 5292
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.28585927341372847,
      "learning_rate": 1.574300677033315e-05,
      "loss": 0.0362,
      "step": 5293
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.26823545035004864,
      "learning_rate": 1.5736055946185137e-05,
      "loss": 0.0534,
      "step": 5294
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3189400835420362,
      "learning_rate": 1.5729105661539198e-05,
      "loss": 0.0563,
      "step": 5295
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.30321555505567427,
      "learning_rate": 1.572215591727473e-05,
      "loss": 0.0445,
      "step": 5296
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.32368159320742906,
      "learning_rate": 1.571520671427105e-05,
      "loss": 0.0744,
      "step": 5297
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.24333355063979578,
      "learning_rate": 1.570825805340743e-05,
      "loss": 0.0329,
      "step": 5298
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2896535218320554,
      "learning_rate": 1.5701309935563055e-05,
      "loss": 0.051,
      "step": 5299
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33976743652464303,
      "learning_rate": 1.5694362361617043e-05,
      "loss": 0.0603,
      "step": 5300
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35174937050118477,
      "learning_rate": 1.5687415332448447e-05,
      "loss": 0.0591,
      "step": 5301
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34926770739697627,
      "learning_rate": 1.568046884893626e-05,
      "loss": 0.08,
      "step": 5302
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3600447674572764,
      "learning_rate": 1.5673522911959376e-05,
      "loss": 0.0609,
      "step": 5303
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3149343445451943,
      "learning_rate": 1.5666577522396658e-05,
      "loss": 0.0616,
      "step": 5304
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34828043207249665,
      "learning_rate": 1.5659632681126883e-05,
      "loss": 0.046,
      "step": 5305
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3265256276829038,
      "learning_rate": 1.565268838902875e-05,
      "loss": 0.0576,
      "step": 5306
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3645142785024197,
      "learning_rate": 1.5645744646980898e-05,
      "loss": 0.0704,
      "step": 5307
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3109067115684409,
      "learning_rate": 1.5638801455861893e-05,
      "loss": 0.047,
      "step": 5308
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29999388400439836,
      "learning_rate": 1.5631858816550238e-05,
      "loss": 0.0514,
      "step": 5309
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2832519752201255,
      "learning_rate": 1.5624916729924354e-05,
      "loss": 0.0596,
      "step": 5310
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2857402804057479,
      "learning_rate": 1.5617975196862607e-05,
      "loss": 0.0326,
      "step": 5311
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.275001616364844,
      "learning_rate": 1.561103421824328e-05,
      "loss": 0.0439,
      "step": 5312
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4556139137217038,
      "learning_rate": 1.5604093794944595e-05,
      "loss": 0.0762,
      "step": 5313
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2608782238361229,
      "learning_rate": 1.5597153927844693e-05,
      "loss": 0.0417,
      "step": 5314
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37030109275277534,
      "learning_rate": 1.5590214617821658e-05,
      "loss": 0.0545,
      "step": 5315
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4052198259673694,
      "learning_rate": 1.5583275865753492e-05,
      "loss": 0.0676,
      "step": 5316
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.32192826385890533,
      "learning_rate": 1.5576337672518136e-05,
      "loss": 0.0579,
      "step": 5317
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4002354949393418,
      "learning_rate": 1.556940003899345e-05,
      "loss": 0.0519,
      "step": 5318
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39395280489513107,
      "learning_rate": 1.5562462966057236e-05,
      "loss": 0.0839,
      "step": 5319
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5002689234420536,
      "learning_rate": 1.55555264545872e-05,
      "loss": 0.0838,
      "step": 5320
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34034321792584765,
      "learning_rate": 1.554859050546102e-05,
      "loss": 0.0413,
      "step": 5321
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3024851603848303,
      "learning_rate": 1.5541655119556262e-05,
      "loss": 0.0525,
      "step": 5322
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34086222787100784,
      "learning_rate": 1.5534720297750448e-05,
      "loss": 0.0461,
      "step": 5323
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3072771039303064,
      "learning_rate": 1.5527786040921e-05,
      "loss": 0.0542,
      "step": 5324
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20857435629195625,
      "learning_rate": 1.5520852349945304e-05,
      "loss": 0.0261,
      "step": 5325
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3035151595771196,
      "learning_rate": 1.551391922570064e-05,
      "loss": 0.0477,
      "step": 5326
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.26215624894685435,
      "learning_rate": 1.5506986669064245e-05,
      "loss": 0.0563,
      "step": 5327
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4793901509721871,
      "learning_rate": 1.550005468091326e-05,
      "loss": 0.105,
      "step": 5328
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.29288639182559645,
      "learning_rate": 1.5493123262124773e-05,
      "loss": 0.0307,
      "step": 5329
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2710120668944695,
      "learning_rate": 1.548619241357579e-05,
      "loss": 0.0478,
      "step": 5330
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3193665469370426,
      "learning_rate": 1.547926213614325e-05,
      "loss": 0.0564,
      "step": 5331
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3998424778657557,
      "learning_rate": 1.5472332430704007e-05,
      "loss": 0.0749,
      "step": 5332
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2829291629957567,
      "learning_rate": 1.546540329813487e-05,
      "loss": 0.0536,
      "step": 5333
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24323200517559682,
      "learning_rate": 1.545847473931254e-05,
      "loss": 0.0424,
      "step": 5334
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33317123385688235,
      "learning_rate": 1.5451546755113676e-05,
      "loss": 0.064,
      "step": 5335
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9349120082370015,
      "learning_rate": 1.5444619346414845e-05,
      "loss": 0.1725,
      "step": 5336
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3779642843335106,
      "learning_rate": 1.5437692514092547e-05,
      "loss": 0.0601,
      "step": 5337
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2697827920718282,
      "learning_rate": 1.543076625902322e-05,
      "loss": 0.0514,
      "step": 5338
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2684082464280908,
      "learning_rate": 1.542384058208321e-05,
      "loss": 0.0506,
      "step": 5339
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.31119893551371064,
      "learning_rate": 1.5416915484148805e-05,
      "loss": 0.0566,
      "step": 5340
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33570136043376897,
      "learning_rate": 1.5409990966096207e-05,
      "loss": 0.0523,
      "step": 5341
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3699570160783724,
      "learning_rate": 1.5403067028801558e-05,
      "loss": 0.0723,
      "step": 5342
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28041338998505977,
      "learning_rate": 1.5396143673140913e-05,
      "loss": 0.0418,
      "step": 5343
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3013494129093857,
      "learning_rate": 1.5389220899990267e-05,
      "loss": 0.0788,
      "step": 5344
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28129126987722175,
      "learning_rate": 1.5382298710225527e-05,
      "loss": 0.0571,
      "step": 5345
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30394044404248366,
      "learning_rate": 1.5375377104722545e-05,
      "loss": 0.0554,
      "step": 5346
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3846943203455503,
      "learning_rate": 1.536845608435707e-05,
      "loss": 0.0709,
      "step": 5347
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30555999983219,
      "learning_rate": 1.5361535650004818e-05,
      "loss": 0.0592,
      "step": 5348
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.26164002444362655,
      "learning_rate": 1.5354615802541388e-05,
      "loss": 0.0478,
      "step": 5349
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3175599528200956,
      "learning_rate": 1.5347696542842333e-05,
      "loss": 0.0479,
      "step": 5350
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2580397500218949,
      "learning_rate": 1.534077787178312e-05,
      "loss": 0.0513,
      "step": 5351
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2770502525434804,
      "learning_rate": 1.5333859790239148e-05,
      "loss": 0.0484,
      "step": 5352
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28066084666789004,
      "learning_rate": 1.532694229908573e-05,
      "loss": 0.0503,
      "step": 5353
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24928148400863714,
      "learning_rate": 1.5320025399198125e-05,
      "loss": 0.0554,
      "step": 5354
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4968713688267873,
      "learning_rate": 1.5313109091451486e-05,
      "loss": 0.0901,
      "step": 5355
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36992235516010447,
      "learning_rate": 1.530619337672093e-05,
      "loss": 0.0643,
      "step": 5356
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.29671074690227967,
      "learning_rate": 1.5299278255881468e-05,
      "loss": 0.0486,
      "step": 5357
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2882430084833048,
      "learning_rate": 1.5292363729808048e-05,
      "loss": 0.0474,
      "step": 5358
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3683207858797972,
      "learning_rate": 1.528544979937554e-05,
      "loss": 0.0557,
      "step": 5359
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28178968572157664,
      "learning_rate": 1.5278536465458738e-05,
      "loss": 0.076,
      "step": 5360
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.34130176106218507,
      "learning_rate": 1.5271623728932364e-05,
      "loss": 0.0614,
      "step": 5361
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35080257512374047,
      "learning_rate": 1.5264711590671067e-05,
      "loss": 0.0737,
      "step": 5362
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3636087561552876,
      "learning_rate": 1.525780005154941e-05,
      "loss": 0.0635,
      "step": 5363
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3003510457765101,
      "learning_rate": 1.5250889112441889e-05,
      "loss": 0.0624,
      "step": 5364
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.25083123537835894,
      "learning_rate": 1.5243978774222918e-05,
      "loss": 0.0392,
      "step": 5365
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2769702760795757,
      "learning_rate": 1.5237069037766843e-05,
      "loss": 0.0512,
      "step": 5366
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3563446738948619,
      "learning_rate": 1.5230159903947923e-05,
      "loss": 0.0614,
      "step": 5367
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.25011045280944494,
      "learning_rate": 1.5223251373640354e-05,
      "loss": 0.0385,
      "step": 5368
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3033275330734785,
      "learning_rate": 1.5216343447718242e-05,
      "loss": 0.0536,
      "step": 5369
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2697809417291579,
      "learning_rate": 1.5209436127055627e-05,
      "loss": 0.0354,
      "step": 5370
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33350597942216426,
      "learning_rate": 1.5202529412526465e-05,
      "loss": 0.0634,
      "step": 5371
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3210924584764184,
      "learning_rate": 1.5195623305004637e-05,
      "loss": 0.047,
      "step": 5372
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2243579120292071,
      "learning_rate": 1.5188717805363957e-05,
      "loss": 0.0368,
      "step": 5373
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3201961422293201,
      "learning_rate": 1.5181812914478146e-05,
      "loss": 0.0552,
      "step": 5374
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.31539077046305714,
      "learning_rate": 1.5174908633220867e-05,
      "loss": 0.0691,
      "step": 5375
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3282751693416627,
      "learning_rate": 1.5168004962465681e-05,
      "loss": 0.0624,
      "step": 5376
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.25793776214545616,
      "learning_rate": 1.5161101903086095e-05,
      "loss": 0.0358,
      "step": 5377
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2321202621166454,
      "learning_rate": 1.5154199455955523e-05,
      "loss": 0.0378,
      "step": 5378
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2582491442692954,
      "learning_rate": 1.5147297621947313e-05,
      "loss": 0.0643,
      "step": 5379
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.26309087815208604,
      "learning_rate": 1.5140396401934725e-05,
      "loss": 0.045,
      "step": 5380
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.25247988970148555,
      "learning_rate": 1.513349579679095e-05,
      "loss": 0.0341,
      "step": 5381
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2934478656778279,
      "learning_rate": 1.5126595807389098e-05,
      "loss": 0.0623,
      "step": 5382
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.329260801735525,
      "learning_rate": 1.51196964346022e-05,
      "loss": 0.0512,
      "step": 5383
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2694806314012504,
      "learning_rate": 1.5112797679303206e-05,
      "loss": 0.0465,
      "step": 5384
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.27326908374573444,
      "learning_rate": 1.5105899542364999e-05,
      "loss": 0.0479,
      "step": 5385
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3268499806458816,
      "learning_rate": 1.5099002024660368e-05,
      "loss": 0.067,
      "step": 5386
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32742266193168806,
      "learning_rate": 1.5092105127062043e-05,
      "loss": 0.0627,
      "step": 5387
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2710233382542948,
      "learning_rate": 1.5085208850442649e-05,
      "loss": 0.0455,
      "step": 5388
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28720160060548783,
      "learning_rate": 1.5078313195674756e-05,
      "loss": 0.0523,
      "step": 5389
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4430665044551116,
      "learning_rate": 1.5071418163630855e-05,
      "loss": 0.0614,
      "step": 5390
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3116163396278419,
      "learning_rate": 1.506452375518334e-05,
      "loss": 0.0565,
      "step": 5391
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3179106914742854,
      "learning_rate": 1.5057629971204546e-05,
      "loss": 0.0653,
      "step": 5392
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3019330771873828,
      "learning_rate": 1.5050736812566709e-05,
      "loss": 0.0466,
      "step": 5393
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.29184498473307646,
      "learning_rate": 1.5043844280142005e-05,
      "loss": 0.0426,
      "step": 5394
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24366121142195443,
      "learning_rate": 1.5036952374802517e-05,
      "loss": 0.035,
      "step": 5395
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2628743862447861,
      "learning_rate": 1.5030061097420255e-05,
      "loss": 0.0633,
      "step": 5396
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24139787017450415,
      "learning_rate": 1.5023170448867148e-05,
      "loss": 0.0423,
      "step": 5397
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2582496924255401,
      "learning_rate": 1.5016280430015052e-05,
      "loss": 0.0395,
      "step": 5398
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30458403322206407,
      "learning_rate": 1.5009391041735725e-05,
      "loss": 0.0476,
      "step": 5399
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.21755135335534165,
      "learning_rate": 1.5002502284900871e-05,
      "loss": 0.0338,
      "step": 5400
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3300649651607083,
      "learning_rate": 1.4995614160382091e-05,
      "loss": 0.0754,
      "step": 5401
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2399360851773767,
      "learning_rate": 1.4988726669050917e-05,
      "loss": 0.0428,
      "step": 5402
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.38898043628789375,
      "learning_rate": 1.4981839811778805e-05,
      "loss": 0.0861,
      "step": 5403
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3379136116279355,
      "learning_rate": 1.4974953589437117e-05,
      "loss": 0.0478,
      "step": 5404
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3192701009375846,
      "learning_rate": 1.4968068002897153e-05,
      "loss": 0.0579,
      "step": 5405
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24777937905881006,
      "learning_rate": 1.4961183053030106e-05,
      "loss": 0.0436,
      "step": 5406
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.31597952854642736,
      "learning_rate": 1.4954298740707122e-05,
      "loss": 0.0707,
      "step": 5407
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2843534456195885,
      "learning_rate": 1.4947415066799247e-05,
      "loss": 0.036,
      "step": 5408
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.22456249079450785,
      "learning_rate": 1.4940532032177438e-05,
      "loss": 0.0362,
      "step": 5409
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2618308040952834,
      "learning_rate": 1.4933649637712593e-05,
      "loss": 0.0391,
      "step": 5410
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36086214277977924,
      "learning_rate": 1.4926767884275508e-05,
      "loss": 0.0487,
      "step": 5411
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32307999336199456,
      "learning_rate": 1.4919886772736915e-05,
      "loss": 0.0426,
      "step": 5412
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2584935930719824,
      "learning_rate": 1.491300630396745e-05,
      "loss": 0.0385,
      "step": 5413
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3644731922901051,
      "learning_rate": 1.4906126478837683e-05,
      "loss": 0.0493,
      "step": 5414
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.37091045872399747,
      "learning_rate": 1.4899247298218086e-05,
      "loss": 0.0435,
      "step": 5415
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.338807972068346,
      "learning_rate": 1.4892368762979067e-05,
      "loss": 0.0502,
      "step": 5416
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32978212752234964,
      "learning_rate": 1.4885490873990934e-05,
      "loss": 0.0486,
      "step": 5417
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2310971325078047,
      "learning_rate": 1.4878613632123928e-05,
      "loss": 0.033,
      "step": 5418
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37039938764459734,
      "learning_rate": 1.48717370382482e-05,
      "loss": 0.0658,
      "step": 5419
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.30192142976569014,
      "learning_rate": 1.4864861093233827e-05,
      "loss": 0.0539,
      "step": 5420
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2542736476480424,
      "learning_rate": 1.485798579795079e-05,
      "loss": 0.0475,
      "step": 5421
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.269254542410742,
      "learning_rate": 1.4851111153269005e-05,
      "loss": 0.0462,
      "step": 5422
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2654876915077558,
      "learning_rate": 1.4844237160058285e-05,
      "loss": 0.0446,
      "step": 5423
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3734391431892028,
      "learning_rate": 1.4837363819188379e-05,
      "loss": 0.0519,
      "step": 5424
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32455339740966643,
      "learning_rate": 1.4830491131528958e-05,
      "loss": 0.0712,
      "step": 5425
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.26414863434376623,
      "learning_rate": 1.4823619097949584e-05,
      "loss": 0.0412,
      "step": 5426
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3019541499822475,
      "learning_rate": 1.4816747719319762e-05,
      "loss": 0.0627,
      "step": 5427
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2368912725973804,
      "learning_rate": 1.4809876996508897e-05,
      "loss": 0.0487,
      "step": 5428
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.35748180032967825,
      "learning_rate": 1.480300693038632e-05,
      "loss": 0.0666,
      "step": 5429
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2602568406159695,
      "learning_rate": 1.4796137521821274e-05,
      "loss": 0.0264,
      "step": 5430
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2388690202943695,
      "learning_rate": 1.4789268771682927e-05,
      "loss": 0.0421,
      "step": 5431
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.25582170573152385,
      "learning_rate": 1.4782400680840352e-05,
      "loss": 0.0294,
      "step": 5432
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.30737770083103,
      "learning_rate": 1.4775533250162553e-05,
      "loss": 0.0474,
      "step": 5433
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.40678172892665165,
      "learning_rate": 1.4768666480518432e-05,
      "loss": 0.0602,
      "step": 5434
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2768167943029228,
      "learning_rate": 1.4761800372776827e-05,
      "loss": 0.0475,
      "step": 5435
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3198632136443426,
      "learning_rate": 1.4754934927806473e-05,
      "loss": 0.0637,
      "step": 5436
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.30841718523709777,
      "learning_rate": 1.4748070146476039e-05,
      "loss": 0.0376,
      "step": 5437
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.28145415790537537,
      "learning_rate": 1.4741206029654098e-05,
      "loss": 0.0485,
      "step": 5438
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2989088086678268,
      "learning_rate": 1.473434257820915e-05,
      "loss": 0.0599,
      "step": 5439
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3758936208420145,
      "learning_rate": 1.472747979300959e-05,
      "loss": 0.0475,
      "step": 5440
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3048547872386122,
      "learning_rate": 1.472061767492375e-05,
      "loss": 0.0659,
      "step": 5441
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3214409651713219,
      "learning_rate": 1.4713756224819872e-05,
      "loss": 0.0472,
      "step": 5442
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2049516378611804,
      "learning_rate": 1.4706895443566116e-05,
      "loss": 0.0471,
      "step": 5443
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32971052432906045,
      "learning_rate": 1.4700035332030545e-05,
      "loss": 0.0434,
      "step": 5444
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.34882422142792185,
      "learning_rate": 1.4693175891081146e-05,
      "loss": 0.0386,
      "step": 5445
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.27072139254498606,
      "learning_rate": 1.468631712158582e-05,
      "loss": 0.0661,
      "step": 5446
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36007833072377554,
      "learning_rate": 1.4679459024412391e-05,
      "loss": 0.0688,
      "step": 5447
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4667074844129673,
      "learning_rate": 1.4672601600428578e-05,
      "loss": 0.0762,
      "step": 5448
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.26789027474626065,
      "learning_rate": 1.4665744850502035e-05,
      "loss": 0.0432,
      "step": 5449
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.46465890879816074,
      "learning_rate": 1.465888877550032e-05,
      "loss": 0.0573,
      "step": 5450
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.30024749907834497,
      "learning_rate": 1.4652033376290912e-05,
      "loss": 0.0605,
      "step": 5451
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.253391903986646,
      "learning_rate": 1.4645178653741194e-05,
      "loss": 0.04,
      "step": 5452
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31819945689409157,
      "learning_rate": 1.463832460871848e-05,
      "loss": 0.0412,
      "step": 5453
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3135739231731931,
      "learning_rate": 1.4631471242089978e-05,
      "loss": 0.0506,
      "step": 5454
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2246633713543759,
      "learning_rate": 1.462461855472283e-05,
      "loss": 0.0639,
      "step": 5455
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36136274429672477,
      "learning_rate": 1.4617766547484075e-05,
      "loss": 0.0529,
      "step": 5456
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32574729925486706,
      "learning_rate": 1.4610915221240685e-05,
      "loss": 0.0444,
      "step": 5457
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23544005950111677,
      "learning_rate": 1.4604064576859513e-05,
      "loss": 0.0404,
      "step": 5458
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3685125426923099,
      "learning_rate": 1.459721461520737e-05,
      "loss": 0.0617,
      "step": 5459
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3273051783158187,
      "learning_rate": 1.459036533715095e-05,
      "loss": 0.0789,
      "step": 5460
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3440088251362481,
      "learning_rate": 1.4583516743556869e-05,
      "loss": 0.0481,
      "step": 5461
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2894821342932941,
      "learning_rate": 1.4576668835291654e-05,
      "loss": 0.0471,
      "step": 5462
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.29190433418095274,
      "learning_rate": 1.4569821613221743e-05,
      "loss": 0.0555,
      "step": 5463
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.27589054589487266,
      "learning_rate": 1.4562975078213504e-05,
      "loss": 0.0483,
      "step": 5464
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38874992468732,
      "learning_rate": 1.4556129231133191e-05,
      "loss": 0.0625,
      "step": 5465
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4230095654041868,
      "learning_rate": 1.4549284072846996e-05,
      "loss": 0.0615,
      "step": 5466
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.26943575508049955,
      "learning_rate": 1.4542439604221005e-05,
      "loss": 0.0383,
      "step": 5467
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37535682232288126,
      "learning_rate": 1.4535595826121233e-05,
      "loss": 0.0656,
      "step": 5468
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.26949227092679146,
      "learning_rate": 1.452875273941359e-05,
      "loss": 0.0546,
      "step": 5469
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.33942434586189885,
      "learning_rate": 1.4521910344963918e-05,
      "loss": 0.0753,
      "step": 5470
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21007208432797356,
      "learning_rate": 1.4515068643637953e-05,
      "loss": 0.034,
      "step": 5471
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.27001918634044286,
      "learning_rate": 1.450822763630136e-05,
      "loss": 0.038,
      "step": 5472
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3848606127917712,
      "learning_rate": 1.4501387323819697e-05,
      "loss": 0.0728,
      "step": 5473
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.28779162874870956,
      "learning_rate": 1.4494547707058459e-05,
      "loss": 0.0479,
      "step": 5474
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.27645190662444785,
      "learning_rate": 1.4487708786883024e-05,
      "loss": 0.0428,
      "step": 5475
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.28981545045382007,
      "learning_rate": 1.4480870564158704e-05,
      "loss": 0.0331,
      "step": 5476
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.27892617654007784,
      "learning_rate": 1.4474033039750718e-05,
      "loss": 0.0589,
      "step": 5477
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31212467065411087,
      "learning_rate": 1.44671962145242e-05,
      "loss": 0.0439,
      "step": 5478
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32269838348985047,
      "learning_rate": 1.4460360089344177e-05,
      "loss": 0.0449,
      "step": 5479
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.30413646264979755,
      "learning_rate": 1.4453524665075607e-05,
      "loss": 0.052,
      "step": 5480
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.26150877201363676,
      "learning_rate": 1.4446689942583348e-05,
      "loss": 0.0414,
      "step": 5481
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2854209611852699,
      "learning_rate": 1.4439855922732182e-05,
      "loss": 0.0395,
      "step": 5482
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2286295158686908,
      "learning_rate": 1.4433022606386781e-05,
      "loss": 0.056,
      "step": 5483
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31209123818954804,
      "learning_rate": 1.4426189994411756e-05,
      "loss": 0.0512,
      "step": 5484
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3384832576347095,
      "learning_rate": 1.4419358087671603e-05,
      "loss": 0.0444,
      "step": 5485
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31930474178516005,
      "learning_rate": 1.4412526887030745e-05,
      "loss": 0.0643,
      "step": 5486
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2956159395115791,
      "learning_rate": 1.4405696393353504e-05,
      "loss": 0.0716,
      "step": 5487
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3250006473974969,
      "learning_rate": 1.4398866607504128e-05,
      "loss": 0.0535,
      "step": 5488
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2354419503025549,
      "learning_rate": 1.4392037530346754e-05,
      "loss": 0.0355,
      "step": 5489
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3262860104173666,
      "learning_rate": 1.4385209162745453e-05,
      "loss": 0.0455,
      "step": 5490
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3416518312330896,
      "learning_rate": 1.4378381505564187e-05,
      "loss": 0.0693,
      "step": 5491
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2731465017379261,
      "learning_rate": 1.4371554559666843e-05,
      "loss": 0.0524,
      "step": 5492
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.42357427492474237,
      "learning_rate": 1.4364728325917199e-05,
      "loss": 0.0695,
      "step": 5493
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.33148932268746417,
      "learning_rate": 1.4357902805178965e-05,
      "loss": 0.0493,
      "step": 5494
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3387095609682479,
      "learning_rate": 1.4351077998315752e-05,
      "loss": 0.0435,
      "step": 5495
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2948594185082858,
      "learning_rate": 1.434425390619107e-05,
      "loss": 0.0891,
      "step": 5496
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2746070915984592,
      "learning_rate": 1.4337430529668357e-05,
      "loss": 0.0393,
      "step": 5497
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32797564785969896,
      "learning_rate": 1.4330607869610945e-05,
      "loss": 0.0529,
      "step": 5498
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.367200546844497,
      "learning_rate": 1.4323785926882088e-05,
      "loss": 0.0507,
      "step": 5499
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.40407725833100544,
      "learning_rate": 1.431696470234493e-05,
      "loss": 0.0756,
      "step": 5500
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31546558849003214,
      "learning_rate": 1.4310144196862552e-05,
      "loss": 0.0773,
      "step": 5501
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1970754216328647,
      "learning_rate": 1.4303324411297918e-05,
      "loss": 0.0333,
      "step": 5502
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3304312357791281,
      "learning_rate": 1.4296505346513918e-05,
      "loss": 0.0546,
      "step": 5503
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3110057751059579,
      "learning_rate": 1.4289687003373342e-05,
      "loss": 0.0563,
      "step": 5504
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23295923031930088,
      "learning_rate": 1.4282869382738893e-05,
      "loss": 0.0394,
      "step": 5505
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2885551898188161,
      "learning_rate": 1.4276052485473177e-05,
      "loss": 0.0306,
      "step": 5506
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2889744005753278,
      "learning_rate": 1.4269236312438718e-05,
      "loss": 0.0586,
      "step": 5507
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.272575108387623,
      "learning_rate": 1.4262420864497939e-05,
      "loss": 0.0402,
      "step": 5508
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.20621237664534375,
      "learning_rate": 1.4255606142513179e-05,
      "loss": 0.0342,
      "step": 5509
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2726226791438715,
      "learning_rate": 1.4248792147346668e-05,
      "loss": 0.0381,
      "step": 5510
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3190446627653352,
      "learning_rate": 1.4241978879860575e-05,
      "loss": 0.0526,
      "step": 5511
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2903752802544899,
      "learning_rate": 1.4235166340916955e-05,
      "loss": 0.0576,
      "step": 5512
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.25975112587755445,
      "learning_rate": 1.4228354531377764e-05,
      "loss": 0.0383,
      "step": 5513
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32837716131802325,
      "learning_rate": 1.4221543452104891e-05,
      "loss": 0.0457,
      "step": 5514
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30506092809449165,
      "learning_rate": 1.4214733103960106e-05,
      "loss": 0.0539,
      "step": 5515
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4490399875718616,
      "learning_rate": 1.4207923487805108e-05,
      "loss": 0.0925,
      "step": 5516
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38683417553188865,
      "learning_rate": 1.4201114604501489e-05,
      "loss": 0.0544,
      "step": 5517
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3301998342910457,
      "learning_rate": 1.4194306454910757e-05,
      "loss": 0.0482,
      "step": 5518
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3232851028765911,
      "learning_rate": 1.4187499039894318e-05,
      "loss": 0.0737,
      "step": 5519
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35437926288173494,
      "learning_rate": 1.4180692360313494e-05,
      "loss": 0.0501,
      "step": 5520
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.23675211397484017,
      "learning_rate": 1.4173886417029512e-05,
      "loss": 0.036,
      "step": 5521
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29212148859087067,
      "learning_rate": 1.4167081210903501e-05,
      "loss": 0.0575,
      "step": 5522
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3708292369759153,
      "learning_rate": 1.4160276742796509e-05,
      "loss": 0.0582,
      "step": 5523
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39275229087443136,
      "learning_rate": 1.4153473013569468e-05,
      "loss": 0.0512,
      "step": 5524
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34021670605474824,
      "learning_rate": 1.4146670024083241e-05,
      "loss": 0.0558,
      "step": 5525
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.28483389571109996,
      "learning_rate": 1.413986777519858e-05,
      "loss": 0.0435,
      "step": 5526
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2538840064062279,
      "learning_rate": 1.4133066267776148e-05,
      "loss": 0.0434,
      "step": 5527
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5287867829535119,
      "learning_rate": 1.412626550267653e-05,
      "loss": 0.1275,
      "step": 5528
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.25951080443834945,
      "learning_rate": 1.4119465480760192e-05,
      "loss": 0.0475,
      "step": 5529
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.553872167402259,
      "learning_rate": 1.4112666202887522e-05,
      "loss": 0.1174,
      "step": 5530
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3106050017076968,
      "learning_rate": 1.4105867669918803e-05,
      "loss": 0.0522,
      "step": 5531
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.21425685482454435,
      "learning_rate": 1.4099069882714236e-05,
      "loss": 0.0427,
      "step": 5532
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2665975381736101,
      "learning_rate": 1.4092272842133916e-05,
      "loss": 0.0499,
      "step": 5533
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2540361218993623,
      "learning_rate": 1.4085476549037856e-05,
      "loss": 0.0514,
      "step": 5534
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.270432775122475,
      "learning_rate": 1.4078681004285959e-05,
      "loss": 0.0512,
      "step": 5535
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.27399114689046483,
      "learning_rate": 1.4071886208738053e-05,
      "loss": 0.0424,
      "step": 5536
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.27262848655328736,
      "learning_rate": 1.4065092163253845e-05,
      "loss": 0.0553,
      "step": 5537
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2644290084773797,
      "learning_rate": 1.4058298868692979e-05,
      "loss": 0.0505,
      "step": 5538
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29287387583277114,
      "learning_rate": 1.405150632591497e-05,
      "loss": 0.0469,
      "step": 5539
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2658416481621058,
      "learning_rate": 1.4044714535779269e-05,
      "loss": 0.0471,
      "step": 5540
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.21789998709725442,
      "learning_rate": 1.4037923499145207e-05,
      "loss": 0.0469,
      "step": 5541
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3261424753793335,
      "learning_rate": 1.403113321687204e-05,
      "loss": 0.0702,
      "step": 5542
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2641477035435995,
      "learning_rate": 1.4024343689818909e-05,
      "loss": 0.0418,
      "step": 5543
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.278911818654436,
      "learning_rate": 1.4017554918844872e-05,
      "loss": 0.0551,
      "step": 5544
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.22926134714822902,
      "learning_rate": 1.4010766904808897e-05,
      "loss": 0.0401,
      "step": 5545
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32269592456877333,
      "learning_rate": 1.4003979648569839e-05,
      "loss": 0.0705,
      "step": 5546
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.24983462436893616,
      "learning_rate": 1.3997193150986472e-05,
      "loss": 0.0374,
      "step": 5547
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.26690762456663464,
      "learning_rate": 1.3990407412917462e-05,
      "loss": 0.0368,
      "step": 5548
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34141004198006764,
      "learning_rate": 1.3983622435221391e-05,
      "loss": 0.055,
      "step": 5549
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3245052198091605,
      "learning_rate": 1.3976838218756733e-05,
      "loss": 0.0594,
      "step": 5550
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32239717052882894,
      "learning_rate": 1.3970054764381879e-05,
      "loss": 0.0471,
      "step": 5551
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3174391828508528,
      "learning_rate": 1.3963272072955106e-05,
      "loss": 0.0554,
      "step": 5552
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2602591165160619,
      "learning_rate": 1.3956490145334613e-05,
      "loss": 0.0478,
      "step": 5553
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.24961398542799765,
      "learning_rate": 1.3949708982378487e-05,
      "loss": 0.0497,
      "step": 5554
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3065719041538778,
      "learning_rate": 1.3942928584944733e-05,
      "loss": 0.0393,
      "step": 5555
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2654547426189175,
      "learning_rate": 1.3936148953891242e-05,
      "loss": 0.0468,
      "step": 5556
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43763271090075617,
      "learning_rate": 1.3929370090075827e-05,
      "loss": 0.0719,
      "step": 5557
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.31957515890533356,
      "learning_rate": 1.392259199435618e-05,
      "loss": 0.0612,
      "step": 5558
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.24393513996547408,
      "learning_rate": 1.3915814667589929e-05,
      "loss": 0.0567,
      "step": 5559
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.46093817888630156,
      "learning_rate": 1.3909038110634567e-05,
      "loss": 0.0753,
      "step": 5560
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29493192946113983,
      "learning_rate": 1.3902262324347522e-05,
      "loss": 0.0541,
      "step": 5561
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29629170687557227,
      "learning_rate": 1.3895487309586097e-05,
      "loss": 0.0519,
      "step": 5562
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36481990858838603,
      "learning_rate": 1.3888713067207531e-05,
      "loss": 0.0663,
      "step": 5563
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3595021893536748,
      "learning_rate": 1.388193959806893e-05,
      "loss": 0.079,
      "step": 5564
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2825396823652111,
      "learning_rate": 1.3875166903027327e-05,
      "loss": 0.0609,
      "step": 5565
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43417501580062645,
      "learning_rate": 1.3868394982939636e-05,
      "loss": 0.054,
      "step": 5566
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2633411028149411,
      "learning_rate": 1.3861623838662694e-05,
      "loss": 0.0399,
      "step": 5567
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33130389935653864,
      "learning_rate": 1.3854853471053225e-05,
      "loss": 0.0738,
      "step": 5568
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2537712031464752,
      "learning_rate": 1.384808388096787e-05,
      "loss": 0.0431,
      "step": 5569
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33226718373923025,
      "learning_rate": 1.3841315069263146e-05,
      "loss": 0.0731,
      "step": 5570
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.350580624510904,
      "learning_rate": 1.3834547036795502e-05,
      "loss": 0.0558,
      "step": 5571
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.27665223966367936,
      "learning_rate": 1.3827779784421262e-05,
      "loss": 0.0492,
      "step": 5572
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30374478951839884,
      "learning_rate": 1.3821013312996674e-05,
      "loss": 0.0499,
      "step": 5573
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3362766151105269,
      "learning_rate": 1.3814247623377868e-05,
      "loss": 0.0765,
      "step": 5574
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2127952449514032,
      "learning_rate": 1.3807482716420887e-05,
      "loss": 0.0274,
      "step": 5575
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2785711530035179,
      "learning_rate": 1.3800718592981668e-05,
      "loss": 0.0393,
      "step": 5576
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34604194976791997,
      "learning_rate": 1.3793955253916058e-05,
      "loss": 0.0681,
      "step": 5577
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30766432870203525,
      "learning_rate": 1.3787192700079792e-05,
      "loss": 0.0459,
      "step": 5578
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.21715383097165467,
      "learning_rate": 1.3780430932328511e-05,
      "loss": 0.0338,
      "step": 5579
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.22558563199373505,
      "learning_rate": 1.377366995151777e-05,
      "loss": 0.0379,
      "step": 5580
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3231437740737838,
      "learning_rate": 1.3766909758503002e-05,
      "loss": 0.053,
      "step": 5581
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.31741061749072647,
      "learning_rate": 1.3760150354139558e-05,
      "loss": 0.0463,
      "step": 5582
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2753116548549051,
      "learning_rate": 1.3753391739282675e-05,
      "loss": 0.0431,
      "step": 5583
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2906315336210816,
      "learning_rate": 1.3746633914787504e-05,
      "loss": 0.0477,
      "step": 5584
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3389201493884306,
      "learning_rate": 1.373987688150908e-05,
      "loss": 0.0406,
      "step": 5585
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19451453278803574,
      "learning_rate": 1.3733120640302358e-05,
      "loss": 0.0295,
      "step": 5586
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3156023173160414,
      "learning_rate": 1.3726365192022173e-05,
      "loss": 0.0502,
      "step": 5587
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.26102974568877396,
      "learning_rate": 1.3719610537523274e-05,
      "loss": 0.0494,
      "step": 5588
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2747212791070434,
      "learning_rate": 1.3712856677660299e-05,
      "loss": 0.0581,
      "step": 5589
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.27562011463567315,
      "learning_rate": 1.3706103613287796e-05,
      "loss": 0.0413,
      "step": 5590
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30757008752066367,
      "learning_rate": 1.3699351345260201e-05,
      "loss": 0.0554,
      "step": 5591
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2882843756570884,
      "learning_rate": 1.369259987443186e-05,
      "loss": 0.0648,
      "step": 5592
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32449289015297467,
      "learning_rate": 1.3685849201657009e-05,
      "loss": 0.0368,
      "step": 5593
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3160807972628687,
      "learning_rate": 1.3679099327789794e-05,
      "loss": 0.0358,
      "step": 5594
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2857022219052435,
      "learning_rate": 1.3672350253684243e-05,
      "loss": 0.0508,
      "step": 5595
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2814827591645604,
      "learning_rate": 1.3665601980194297e-05,
      "loss": 0.0677,
      "step": 5596
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32651648904558583,
      "learning_rate": 1.3658854508173796e-05,
      "loss": 0.0427,
      "step": 5597
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32266546933578244,
      "learning_rate": 1.3652107838476476e-05,
      "loss": 0.0522,
      "step": 5598
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20409014546709017,
      "learning_rate": 1.3645361971955961e-05,
      "loss": 0.0292,
      "step": 5599
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2655659918597047,
      "learning_rate": 1.3638616909465791e-05,
      "loss": 0.0576,
      "step": 5600
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.29454070210151634,
      "learning_rate": 1.3631872651859386e-05,
      "loss": 0.0449,
      "step": 5601
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3701140874164158,
      "learning_rate": 1.3625129199990083e-05,
      "loss": 0.0553,
      "step": 5602
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4178331279869523,
      "learning_rate": 1.3618386554711101e-05,
      "loss": 0.0663,
      "step": 5603
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.28184870732616496,
      "learning_rate": 1.3611644716875568e-05,
      "loss": 0.0748,
      "step": 5604
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3562858538404201,
      "learning_rate": 1.3604903687336499e-05,
      "loss": 0.0589,
      "step": 5605
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.28134279574611937,
      "learning_rate": 1.3598163466946823e-05,
      "loss": 0.0351,
      "step": 5606
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.24204038951208703,
      "learning_rate": 1.3591424056559345e-05,
      "loss": 0.0418,
      "step": 5607
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3284039220151172,
      "learning_rate": 1.3584685457026789e-05,
      "loss": 0.0642,
      "step": 5608
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.299401723827682,
      "learning_rate": 1.357794766920176e-05,
      "loss": 0.051,
      "step": 5609
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3028739361399923,
      "learning_rate": 1.3571210693936774e-05,
      "loss": 0.0399,
      "step": 5610
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.24880108349520488,
      "learning_rate": 1.3564474532084226e-05,
      "loss": 0.0495,
      "step": 5611
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33811961666835144,
      "learning_rate": 1.3557739184496435e-05,
      "loss": 0.0498,
      "step": 5612
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2787868538045174,
      "learning_rate": 1.3551004652025582e-05,
      "loss": 0.0502,
      "step": 5613
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.28603466811420103,
      "learning_rate": 1.3544270935523778e-05,
      "loss": 0.0546,
      "step": 5614
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.45555747658174617,
      "learning_rate": 1.3537538035843017e-05,
      "loss": 0.0866,
      "step": 5615
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.24424289087386372,
      "learning_rate": 1.3530805953835182e-05,
      "loss": 0.0303,
      "step": 5616
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.36258914607731035,
      "learning_rate": 1.3524074690352068e-05,
      "loss": 0.0738,
      "step": 5617
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37528027550987486,
      "learning_rate": 1.351734424624535e-05,
      "loss": 0.067,
      "step": 5618
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3750212583874112,
      "learning_rate": 1.3510614622366615e-05,
      "loss": 0.0717,
      "step": 5619
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.25464443591608105,
      "learning_rate": 1.3503885819567335e-05,
      "loss": 0.0578,
      "step": 5620
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2537079550293961,
      "learning_rate": 1.3497157838698884e-05,
      "loss": 0.0523,
      "step": 5621
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.23945383099798823,
      "learning_rate": 1.3490430680612528e-05,
      "loss": 0.032,
      "step": 5622
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.25087312875096374,
      "learning_rate": 1.3483704346159435e-05,
      "loss": 0.0389,
      "step": 5623
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2695963822326179,
      "learning_rate": 1.3476978836190658e-05,
      "loss": 0.0339,
      "step": 5624
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.29095323905207143,
      "learning_rate": 1.347025415155716e-05,
      "loss": 0.0508,
      "step": 5625
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2600336341468963,
      "learning_rate": 1.3463530293109783e-05,
      "loss": 0.0373,
      "step": 5626
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.29484438351804176,
      "learning_rate": 1.3456807261699288e-05,
      "loss": 0.0503,
      "step": 5627
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2075552068225611,
      "learning_rate": 1.34500850581763e-05,
      "loss": 0.0376,
      "step": 5628
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2993788039616833,
      "learning_rate": 1.3443363683391369e-05,
      "loss": 0.0454,
      "step": 5629
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2754091567709912,
      "learning_rate": 1.3436643138194918e-05,
      "loss": 0.0706,
      "step": 5630
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3322685179325457,
      "learning_rate": 1.3429923423437277e-05,
      "loss": 0.0392,
      "step": 5631
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.39415352746356985,
      "learning_rate": 1.3423204539968677e-05,
      "loss": 0.0542,
      "step": 5632
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2923638647918311,
      "learning_rate": 1.3416486488639222e-05,
      "loss": 0.0368,
      "step": 5633
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2851589673056533,
      "learning_rate": 1.3409769270298934e-05,
      "loss": 0.0455,
      "step": 5634
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35113104730370964,
      "learning_rate": 1.340305288579771e-05,
      "loss": 0.0423,
      "step": 5635
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.30857305819516934,
      "learning_rate": 1.3396337335985361e-05,
      "loss": 0.0414,
      "step": 5636
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3351533742928012,
      "learning_rate": 1.3389622621711578e-05,
      "loss": 0.0637,
      "step": 5637
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3897978226205992,
      "learning_rate": 1.3382908743825947e-05,
      "loss": 0.0543,
      "step": 5638
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2723712021369034,
      "learning_rate": 1.337619570317796e-05,
      "loss": 0.0316,
      "step": 5639
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.302138918114539,
      "learning_rate": 1.336948350061698e-05,
      "loss": 0.0502,
      "step": 5640
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2337126273097789,
      "learning_rate": 1.3362772136992294e-05,
      "loss": 0.0287,
      "step": 5641
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.255145642631831,
      "learning_rate": 1.335606161315306e-05,
      "loss": 0.0294,
      "step": 5642
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32703431829260704,
      "learning_rate": 1.3349351929948342e-05,
      "loss": 0.057,
      "step": 5643
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22955598958829557,
      "learning_rate": 1.3342643088227085e-05,
      "loss": 0.0439,
      "step": 5644
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.29668990438422027,
      "learning_rate": 1.3335935088838145e-05,
      "loss": 0.0389,
      "step": 5645
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3153918925692883,
      "learning_rate": 1.3329227932630255e-05,
      "loss": 0.0458,
      "step": 5646
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3255786640931498,
      "learning_rate": 1.3322521620452053e-05,
      "loss": 0.0448,
      "step": 5647
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.313656384001237,
      "learning_rate": 1.3315816153152055e-05,
      "loss": 0.057,
      "step": 5648
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.28341564228774163,
      "learning_rate": 1.3309111531578692e-05,
      "loss": 0.0446,
      "step": 5649
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.25463943261886,
      "learning_rate": 1.3302407756580278e-05,
      "loss": 0.0457,
      "step": 5650
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3247302742576951,
      "learning_rate": 1.329570482900501e-05,
      "loss": 0.05,
      "step": 5651
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33704674361070186,
      "learning_rate": 1.3289002749700992e-05,
      "loss": 0.0406,
      "step": 5652
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3043891228252577,
      "learning_rate": 1.3282301519516212e-05,
      "loss": 0.0467,
      "step": 5653
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3877916000292103,
      "learning_rate": 1.3275601139298556e-05,
      "loss": 0.0626,
      "step": 5654
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.28729085311931213,
      "learning_rate": 1.3268901609895793e-05,
      "loss": 0.0622,
      "step": 5655
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32712097052664113,
      "learning_rate": 1.3262202932155602e-05,
      "loss": 0.0488,
      "step": 5656
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2905834624642349,
      "learning_rate": 1.3255505106925535e-05,
      "loss": 0.0724,
      "step": 5657
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3413329654663837,
      "learning_rate": 1.3248808135053048e-05,
      "loss": 0.0447,
      "step": 5658
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.28871752799946365,
      "learning_rate": 1.3242112017385483e-05,
      "loss": 0.0514,
      "step": 5659
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.24008771249711366,
      "learning_rate": 1.3235416754770082e-05,
      "loss": 0.0563,
      "step": 5660
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3039428831051665,
      "learning_rate": 1.3228722348053969e-05,
      "loss": 0.0527,
      "step": 5661
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47881643999592677,
      "learning_rate": 1.3222028798084165e-05,
      "loss": 0.0843,
      "step": 5662
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35023204400203517,
      "learning_rate": 1.3215336105707581e-05,
      "loss": 0.0677,
      "step": 5663
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2558142644120952,
      "learning_rate": 1.3208644271771026e-05,
      "loss": 0.0405,
      "step": 5664
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2976658476636355,
      "learning_rate": 1.3201953297121177e-05,
      "loss": 0.0603,
      "step": 5665
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3783144507793483,
      "learning_rate": 1.3195263182604638e-05,
      "loss": 0.0555,
      "step": 5666
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.30107973819391354,
      "learning_rate": 1.3188573929067885e-05,
      "loss": 0.04,
      "step": 5667
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.34039598299213053,
      "learning_rate": 1.3181885537357277e-05,
      "loss": 0.0814,
      "step": 5668
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22695233251331523,
      "learning_rate": 1.3175198008319083e-05,
      "loss": 0.0488,
      "step": 5669
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3901315051364228,
      "learning_rate": 1.3168511342799444e-05,
      "loss": 0.0487,
      "step": 5670
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.23973099763971675,
      "learning_rate": 1.3161825541644407e-05,
      "loss": 0.045,
      "step": 5671
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.26408622115993735,
      "learning_rate": 1.3155140605699894e-05,
      "loss": 0.0582,
      "step": 5672
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2486145851903617,
      "learning_rate": 1.3148456535811742e-05,
      "loss": 0.0298,
      "step": 5673
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.27015364008581044,
      "learning_rate": 1.3141773332825647e-05,
      "loss": 0.0514,
      "step": 5674
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4992184192733983,
      "learning_rate": 1.3135090997587226e-05,
      "loss": 0.0853,
      "step": 5675
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2708242885595397,
      "learning_rate": 1.3128409530941957e-05,
      "loss": 0.0394,
      "step": 5676
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35954235160892956,
      "learning_rate": 1.3121728933735231e-05,
      "loss": 0.0682,
      "step": 5677
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19512277925438481,
      "learning_rate": 1.3115049206812325e-05,
      "loss": 0.0407,
      "step": 5678
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2511889675844502,
      "learning_rate": 1.3108370351018393e-05,
      "loss": 0.0693,
      "step": 5679
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4373225806026998,
      "learning_rate": 1.3101692367198498e-05,
      "loss": 0.1273,
      "step": 5680
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.20473264021750973,
      "learning_rate": 1.3095015256197568e-05,
      "loss": 0.0399,
      "step": 5681
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.26275027242586513,
      "learning_rate": 1.3088339018860439e-05,
      "loss": 0.0572,
      "step": 5682
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.26364107305257717,
      "learning_rate": 1.3081663656031845e-05,
      "loss": 0.0368,
      "step": 5683
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5068132105088602,
      "learning_rate": 1.307498916855638e-05,
      "loss": 0.0754,
      "step": 5684
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.26044753209623883,
      "learning_rate": 1.3068315557278559e-05,
      "loss": 0.0479,
      "step": 5685
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2345875015122804,
      "learning_rate": 1.3061642823042757e-05,
      "loss": 0.0436,
      "step": 5686
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3928649765323534,
      "learning_rate": 1.3054970966693264e-05,
      "loss": 0.0612,
      "step": 5687
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.23474465465350516,
      "learning_rate": 1.3048299989074234e-05,
      "loss": 0.0502,
      "step": 5688
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2308454990250989,
      "learning_rate": 1.3041629891029732e-05,
      "loss": 0.0286,
      "step": 5689
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2263089273703307,
      "learning_rate": 1.3034960673403699e-05,
      "loss": 0.0421,
      "step": 5690
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2750013183432103,
      "learning_rate": 1.3028292337039971e-05,
      "loss": 0.0427,
      "step": 5691
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24221237885452,
      "learning_rate": 1.3021624882782262e-05,
      "loss": 0.0398,
      "step": 5692
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20854592007108078,
      "learning_rate": 1.3014958311474194e-05,
      "loss": 0.0449,
      "step": 5693
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3487098672807025,
      "learning_rate": 1.3008292623959253e-05,
      "loss": 0.068,
      "step": 5694
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3213280419354896,
      "learning_rate": 1.3001627821080835e-05,
      "loss": 0.0547,
      "step": 5695
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2351453360372637,
      "learning_rate": 1.2994963903682205e-05,
      "loss": 0.0422,
      "step": 5696
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2874713354748142,
      "learning_rate": 1.2988300872606536e-05,
      "loss": 0.0619,
      "step": 5697
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23210050499812654,
      "learning_rate": 1.2981638728696868e-05,
      "loss": 0.045,
      "step": 5698
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32298319141598664,
      "learning_rate": 1.2974977472796146e-05,
      "loss": 0.0541,
      "step": 5699
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3437758457697743,
      "learning_rate": 1.2968317105747189e-05,
      "loss": 0.0572,
      "step": 5700
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2582180561359662,
      "learning_rate": 1.2961657628392716e-05,
      "loss": 0.0441,
      "step": 5701
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3844791255810692,
      "learning_rate": 1.2954999041575331e-05,
      "loss": 0.0636,
      "step": 5702
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.29185180096050184,
      "learning_rate": 1.2948341346137513e-05,
      "loss": 0.0438,
      "step": 5703
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3191671012625628,
      "learning_rate": 1.2941684542921646e-05,
      "loss": 0.0612,
      "step": 5704
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38620789820024126,
      "learning_rate": 1.2935028632769982e-05,
      "loss": 0.0548,
      "step": 5705
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2670933027583946,
      "learning_rate": 1.2928373616524682e-05,
      "loss": 0.0269,
      "step": 5706
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.35971463781304086,
      "learning_rate": 1.292171949502777e-05,
      "loss": 0.0667,
      "step": 5707
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.29065074704059624,
      "learning_rate": 1.291506626912118e-05,
      "loss": 0.045,
      "step": 5708
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3391240710067539,
      "learning_rate": 1.2908413939646712e-05,
      "loss": 0.0522,
      "step": 5709
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2500654969249138,
      "learning_rate": 1.290176250744607e-05,
      "loss": 0.0347,
      "step": 5710
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2704660679526101,
      "learning_rate": 1.2895111973360828e-05,
      "loss": 0.0533,
      "step": 5711
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36630925193977665,
      "learning_rate": 1.2888462338232466e-05,
      "loss": 0.0466,
      "step": 5712
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18245515417702202,
      "learning_rate": 1.288181360290233e-05,
      "loss": 0.0226,
      "step": 5713
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2916379840689359,
      "learning_rate": 1.287516576821167e-05,
      "loss": 0.0355,
      "step": 5714
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.26664493194061323,
      "learning_rate": 1.2868518835001603e-05,
      "loss": 0.0458,
      "step": 5715
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.28679912698225796,
      "learning_rate": 1.2861872804113154e-05,
      "loss": 0.0409,
      "step": 5716
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3448219275472702,
      "learning_rate": 1.28552276763872e-05,
      "loss": 0.0701,
      "step": 5717
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3062918070069477,
      "learning_rate": 1.284858345266456e-05,
      "loss": 0.0458,
      "step": 5718
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.33428865369586186,
      "learning_rate": 1.284194013378588e-05,
      "loss": 0.0528,
      "step": 5719
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4147825817407804,
      "learning_rate": 1.2835297720591729e-05,
      "loss": 0.0594,
      "step": 5720
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24612488625360576,
      "learning_rate": 1.2828656213922537e-05,
      "loss": 0.0316,
      "step": 5721
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3221885126506827,
      "learning_rate": 1.282201561461864e-05,
      "loss": 0.0589,
      "step": 5722
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39136553188970963,
      "learning_rate": 1.2815375923520244e-05,
      "loss": 0.0741,
      "step": 5723
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2846913270871041,
      "learning_rate": 1.2808737141467451e-05,
      "loss": 0.0454,
      "step": 5724
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2702520931420444,
      "learning_rate": 1.2802099269300237e-05,
      "loss": 0.041,
      "step": 5725
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3761652089258179,
      "learning_rate": 1.2795462307858478e-05,
      "loss": 0.0769,
      "step": 5726
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.22729783921874094,
      "learning_rate": 1.2788826257981914e-05,
      "loss": 0.0333,
      "step": 5727
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.28279548772995067,
      "learning_rate": 1.2782191120510196e-05,
      "loss": 0.0446,
      "step": 5728
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2769705047319417,
      "learning_rate": 1.277555689628283e-05,
      "loss": 0.0496,
      "step": 5729
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24887583228852966,
      "learning_rate": 1.2768923586139232e-05,
      "loss": 0.0461,
      "step": 5730
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37740056259823923,
      "learning_rate": 1.2762291190918685e-05,
      "loss": 0.0699,
      "step": 5731
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2696253018136561,
      "learning_rate": 1.275565971146037e-05,
      "loss": 0.0417,
      "step": 5732
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1836277646148132,
      "learning_rate": 1.2749029148603339e-05,
      "loss": 0.0502,
      "step": 5733
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.272553253941592,
      "learning_rate": 1.2742399503186528e-05,
      "loss": 0.0443,
      "step": 5734
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24973518532124672,
      "learning_rate": 1.2735770776048781e-05,
      "loss": 0.0507,
      "step": 5735
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2735331776665544,
      "learning_rate": 1.2729142968028793e-05,
      "loss": 0.0427,
      "step": 5736
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31019006780176067,
      "learning_rate": 1.272251607996517e-05,
      "loss": 0.0501,
      "step": 5737
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2837000203753502,
      "learning_rate": 1.2715890112696379e-05,
      "loss": 0.0374,
      "step": 5738
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.22991461327271748,
      "learning_rate": 1.2709265067060782e-05,
      "loss": 0.0484,
      "step": 5739
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2804481945102823,
      "learning_rate": 1.2702640943896625e-05,
      "loss": 0.0353,
      "step": 5740
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4540546516060464,
      "learning_rate": 1.2696017744042037e-05,
      "loss": 0.0405,
      "step": 5741
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3727401351083858,
      "learning_rate": 1.2689395468335027e-05,
      "loss": 0.055,
      "step": 5742
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2821227678084727,
      "learning_rate": 1.2682774117613485e-05,
      "loss": 0.0435,
      "step": 5743
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.29552586078681403,
      "learning_rate": 1.2676153692715195e-05,
      "loss": 0.0382,
      "step": 5744
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24667691963920307,
      "learning_rate": 1.266953419447781e-05,
      "loss": 0.0382,
      "step": 5745
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3267198174567278,
      "learning_rate": 1.2662915623738874e-05,
      "loss": 0.0537,
      "step": 5746
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2558199437165409,
      "learning_rate": 1.2656297981335814e-05,
      "loss": 0.0499,
      "step": 5747
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.35545495813463757,
      "learning_rate": 1.2649681268105933e-05,
      "loss": 0.0604,
      "step": 5748
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3266137379508782,
      "learning_rate": 1.2643065484886428e-05,
      "loss": 0.0587,
      "step": 5749
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3277081271039763,
      "learning_rate": 1.263645063251436e-05,
      "loss": 0.0524,
      "step": 5750
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.28440048711395666,
      "learning_rate": 1.2629836711826688e-05,
      "loss": 0.0561,
      "step": 5751
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.22416114033802687,
      "learning_rate": 1.2623223723660258e-05,
      "loss": 0.0437,
      "step": 5752
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2531515643226877,
      "learning_rate": 1.261661166885178e-05,
      "loss": 0.0424,
      "step": 5753
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3143136325193017,
      "learning_rate": 1.2610000548237851e-05,
      "loss": 0.0461,
      "step": 5754
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3286135986689357,
      "learning_rate": 1.2603390362654964e-05,
      "loss": 0.0553,
      "step": 5755
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.33618499820669795,
      "learning_rate": 1.259678111293947e-05,
      "loss": 0.0577,
      "step": 5756
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3299252195412322,
      "learning_rate": 1.259017279992763e-05,
      "loss": 0.0519,
      "step": 5757
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3256650173030957,
      "learning_rate": 1.2583565424455552e-05,
      "loss": 0.0751,
      "step": 5758
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2653295191820363,
      "learning_rate": 1.2576958987359263e-05,
      "loss": 0.0512,
      "step": 5759
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20432404434109436,
      "learning_rate": 1.2570353489474637e-05,
      "loss": 0.0397,
      "step": 5760
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2987309466155845,
      "learning_rate": 1.2563748931637454e-05,
      "loss": 0.0625,
      "step": 5761
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.30501896605775447,
      "learning_rate": 1.2557145314683364e-05,
      "loss": 0.0464,
      "step": 5762
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2846207358694666,
      "learning_rate": 1.2550542639447897e-05,
      "loss": 0.062,
      "step": 5763
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1936883824566227,
      "learning_rate": 1.254394090676647e-05,
      "loss": 0.0248,
      "step": 5764
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24202981553166694,
      "learning_rate": 1.253734011747438e-05,
      "loss": 0.0378,
      "step": 5765
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20140512147976442,
      "learning_rate": 1.2530740272406792e-05,
      "loss": 0.029,
      "step": 5766
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2325749974782116,
      "learning_rate": 1.2524141372398772e-05,
      "loss": 0.0442,
      "step": 5767
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.340700932009995,
      "learning_rate": 1.2517543418285247e-05,
      "loss": 0.0464,
      "step": 5768
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31303451839285734,
      "learning_rate": 1.2510946410901038e-05,
      "loss": 0.0469,
      "step": 5769
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31230670195902593,
      "learning_rate": 1.2504350351080845e-05,
      "loss": 0.053,
      "step": 5770
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23782032803347783,
      "learning_rate": 1.2497755239659239e-05,
      "loss": 0.0257,
      "step": 5771
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3863448445064645,
      "learning_rate": 1.2491161077470682e-05,
      "loss": 0.0755,
      "step": 5772
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.27114383004076587,
      "learning_rate": 1.2484567865349503e-05,
      "loss": 0.0518,
      "step": 5773
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24584206320461685,
      "learning_rate": 1.2477975604129929e-05,
      "loss": 0.0386,
      "step": 5774
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20498180852400374,
      "learning_rate": 1.2471384294646044e-05,
      "loss": 0.0309,
      "step": 5775
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2307205279665543,
      "learning_rate": 1.2464793937731831e-05,
      "loss": 0.049,
      "step": 5776
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3048937661783259,
      "learning_rate": 1.2458204534221144e-05,
      "loss": 0.0634,
      "step": 5777
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4014768613048012,
      "learning_rate": 1.2451616084947714e-05,
      "loss": 0.0769,
      "step": 5778
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.25776815032970624,
      "learning_rate": 1.2445028590745159e-05,
      "loss": 0.0483,
      "step": 5779
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.22807011009267175,
      "learning_rate": 1.243844205244697e-05,
      "loss": 0.0305,
      "step": 5780
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3265058669122508,
      "learning_rate": 1.2431856470886517e-05,
      "loss": 0.0677,
      "step": 5781
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.30663722334025834,
      "learning_rate": 1.2425271846897053e-05,
      "loss": 0.0549,
      "step": 5782
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.29927762201887903,
      "learning_rate": 1.2418688181311706e-05,
      "loss": 0.054,
      "step": 5783
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37099138152967415,
      "learning_rate": 1.2412105474963491e-05,
      "loss": 0.072,
      "step": 5784
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4296392933940413,
      "learning_rate": 1.2405523728685285e-05,
      "loss": 0.0914,
      "step": 5785
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.23974581405716563,
      "learning_rate": 1.2398942943309855e-05,
      "loss": 0.0395,
      "step": 5786
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2947184385237971,
      "learning_rate": 1.2392363119669854e-05,
      "loss": 0.0537,
      "step": 5787
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2661452527920139,
      "learning_rate": 1.2385784258597796e-05,
      "loss": 0.0402,
      "step": 5788
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3350297951614678,
      "learning_rate": 1.2379206360926088e-05,
      "loss": 0.0609,
      "step": 5789
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3529668732450119,
      "learning_rate": 1.2372629427487e-05,
      "loss": 0.0728,
      "step": 5790
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.29138840730669946,
      "learning_rate": 1.23660534591127e-05,
      "loss": 0.0421,
      "step": 5791
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2680121472865356,
      "learning_rate": 1.235947845663521e-05,
      "loss": 0.031,
      "step": 5792
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.26456766901930184,
      "learning_rate": 1.235290442088645e-05,
      "loss": 0.0404,
      "step": 5793
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.26997056348041215,
      "learning_rate": 1.2346331352698206e-05,
      "loss": 0.055,
      "step": 5794
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.20719950514602206,
      "learning_rate": 1.2339759252902149e-05,
      "loss": 0.0445,
      "step": 5795
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.25988945841695127,
      "learning_rate": 1.2333188122329824e-05,
      "loss": 0.0466,
      "step": 5796
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.33331206621434734,
      "learning_rate": 1.232661796181265e-05,
      "loss": 0.0549,
      "step": 5797
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.26637819036115395,
      "learning_rate": 1.2320048772181932e-05,
      "loss": 0.0514,
      "step": 5798
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31014116043935575,
      "learning_rate": 1.2313480554268842e-05,
      "loss": 0.0362,
      "step": 5799
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2613991308753473,
      "learning_rate": 1.2306913308904435e-05,
      "loss": 0.0352,
      "step": 5800
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32717414868794154,
      "learning_rate": 1.2300347036919642e-05,
      "loss": 0.0365,
      "step": 5801
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35574824485426265,
      "learning_rate": 1.2293781739145274e-05,
      "loss": 0.0603,
      "step": 5802
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.23301050835302559,
      "learning_rate": 1.2287217416412003e-05,
      "loss": 0.0338,
      "step": 5803
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.28191890909543665,
      "learning_rate": 1.2280654069550404e-05,
      "loss": 0.0513,
      "step": 5804
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.34215926441422495,
      "learning_rate": 1.227409169939091e-05,
      "loss": 0.0442,
      "step": 5805
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30860399881075634,
      "learning_rate": 1.2267530306763837e-05,
      "loss": 0.0428,
      "step": 5806
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2699669343516614,
      "learning_rate": 1.2260969892499372e-05,
      "loss": 0.0347,
      "step": 5807
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.18205267816903203,
      "learning_rate": 1.2254410457427581e-05,
      "loss": 0.0333,
      "step": 5808
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.27495044673606595,
      "learning_rate": 1.2247852002378413e-05,
      "loss": 0.034,
      "step": 5809
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3350322080366671,
      "learning_rate": 1.2241294528181678e-05,
      "loss": 0.0525,
      "step": 5810
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.24830204311852275,
      "learning_rate": 1.2234738035667077e-05,
      "loss": 0.0389,
      "step": 5811
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.22728106398085093,
      "learning_rate": 1.2228182525664175e-05,
      "loss": 0.0242,
      "step": 5812
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.28524541114138385,
      "learning_rate": 1.222162799900242e-05,
      "loss": 0.0402,
      "step": 5813
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2821618768289605,
      "learning_rate": 1.2215074456511136e-05,
      "loss": 0.0439,
      "step": 5814
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32805239918748236,
      "learning_rate": 1.2208521899019519e-05,
      "loss": 0.0475,
      "step": 5815
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2506211283615422,
      "learning_rate": 1.2201970327356639e-05,
      "loss": 0.0426,
      "step": 5816
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.28078780343331805,
      "learning_rate": 1.2195419742351446e-05,
      "loss": 0.0538,
      "step": 5817
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3444429695361017,
      "learning_rate": 1.2188870144832758e-05,
      "loss": 0.055,
      "step": 5818
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4386728947424649,
      "learning_rate": 1.2182321535629283e-05,
      "loss": 0.0672,
      "step": 5819
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2577119978169022,
      "learning_rate": 1.217577391556958e-05,
      "loss": 0.0413,
      "step": 5820
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.25646959583509726,
      "learning_rate": 1.2169227285482107e-05,
      "loss": 0.0338,
      "step": 5821
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.27901527240100915,
      "learning_rate": 1.2162681646195187e-05,
      "loss": 0.0352,
      "step": 5822
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36635788067556735,
      "learning_rate": 1.2156136998537011e-05,
      "loss": 0.0544,
      "step": 5823
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.25938737908662424,
      "learning_rate": 1.2149593343335658e-05,
      "loss": 0.0456,
      "step": 5824
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.28311264960612514,
      "learning_rate": 1.2143050681419064e-05,
      "loss": 0.0484,
      "step": 5825
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41016900178977517,
      "learning_rate": 1.2136509013615063e-05,
      "loss": 0.0569,
      "step": 5826
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.301530857600272,
      "learning_rate": 1.2129968340751334e-05,
      "loss": 0.0412,
      "step": 5827
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2512769719492511,
      "learning_rate": 1.2123428663655457e-05,
      "loss": 0.0476,
      "step": 5828
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3771855324707294,
      "learning_rate": 1.211688998315487e-05,
      "loss": 0.0526,
      "step": 5829
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3591906447862843,
      "learning_rate": 1.211035230007689e-05,
      "loss": 0.0486,
      "step": 5830
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.22484123867453631,
      "learning_rate": 1.2103815615248707e-05,
      "loss": 0.0387,
      "step": 5831
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3046539116954684,
      "learning_rate": 1.209727992949739e-05,
      "loss": 0.0504,
      "step": 5832
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.27224374133133045,
      "learning_rate": 1.2090745243649867e-05,
      "loss": 0.0315,
      "step": 5833
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.33649721757495143,
      "learning_rate": 1.2084211558532958e-05,
      "loss": 0.0441,
      "step": 5834
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40559580287859104,
      "learning_rate": 1.2077678874973345e-05,
      "loss": 0.0722,
      "step": 5835
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2173574091762297,
      "learning_rate": 1.2071147193797578e-05,
      "loss": 0.0383,
      "step": 5836
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31055949443392256,
      "learning_rate": 1.2064616515832106e-05,
      "loss": 0.0498,
      "step": 5837
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2579892448541876,
      "learning_rate": 1.2058086841903211e-05,
      "loss": 0.04,
      "step": 5838
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3041065864907497,
      "learning_rate": 1.2051558172837085e-05,
      "loss": 0.0353,
      "step": 5839
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31064776814947015,
      "learning_rate": 1.204503050945978e-05,
      "loss": 0.0567,
      "step": 5840
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.26751106702781635,
      "learning_rate": 1.2038503852597208e-05,
      "loss": 0.0316,
      "step": 5841
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.20446986866347197,
      "learning_rate": 1.2031978203075172e-05,
      "loss": 0.0424,
      "step": 5842
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2280298596392424,
      "learning_rate": 1.2025453561719335e-05,
      "loss": 0.0316,
      "step": 5843
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2075055375710355,
      "learning_rate": 1.2018929929355241e-05,
      "loss": 0.0189,
      "step": 5844
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4836078228851798,
      "learning_rate": 1.2012407306808298e-05,
      "loss": 0.0836,
      "step": 5845
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.29411732909416793,
      "learning_rate": 1.2005885694903796e-05,
      "loss": 0.0435,
      "step": 5846
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2879381997125005,
      "learning_rate": 1.1999365094466888e-05,
      "loss": 0.0382,
      "step": 5847
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2391479473494685,
      "learning_rate": 1.1992845506322607e-05,
      "loss": 0.0386,
      "step": 5848
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32471280200390734,
      "learning_rate": 1.1986326931295848e-05,
      "loss": 0.0578,
      "step": 5849
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2750064388518373,
      "learning_rate": 1.1979809370211392e-05,
      "loss": 0.044,
      "step": 5850
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3279516125291141,
      "learning_rate": 1.1973292823893873e-05,
      "loss": 0.0379,
      "step": 5851
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36044388908966557,
      "learning_rate": 1.196677729316782e-05,
      "loss": 0.064,
      "step": 5852
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3009939308774455,
      "learning_rate": 1.1960262778857606e-05,
      "loss": 0.0449,
      "step": 5853
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2874112520901466,
      "learning_rate": 1.1953749281787502e-05,
      "loss": 0.0498,
      "step": 5854
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32790016236794195,
      "learning_rate": 1.1947236802781625e-05,
      "loss": 0.0617,
      "step": 5855
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35839585971165094,
      "learning_rate": 1.194072534266399e-05,
      "loss": 0.05,
      "step": 5856
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3069293127146736,
      "learning_rate": 1.1934214902258467e-05,
      "loss": 0.0632,
      "step": 5857
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2107956727383603,
      "learning_rate": 1.1927705482388794e-05,
      "loss": 0.0316,
      "step": 5858
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3554874876344733,
      "learning_rate": 1.192119708387859e-05,
      "loss": 0.0515,
      "step": 5859
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2545166601094057,
      "learning_rate": 1.1914689707551337e-05,
      "loss": 0.0488,
      "step": 5860
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2803068639513982,
      "learning_rate": 1.1908183354230394e-05,
      "loss": 0.049,
      "step": 5861
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30098499481845514,
      "learning_rate": 1.1901678024738983e-05,
      "loss": 0.0367,
      "step": 5862
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39952902783196215,
      "learning_rate": 1.1895173719900206e-05,
      "loss": 0.0557,
      "step": 5863
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30311120483626397,
      "learning_rate": 1.1888670440537025e-05,
      "loss": 0.0346,
      "step": 5864
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37709947818455974,
      "learning_rate": 1.1882168187472281e-05,
      "loss": 0.0394,
      "step": 5865
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.33292902576596056,
      "learning_rate": 1.1875666961528679e-05,
      "loss": 0.0477,
      "step": 5866
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.26485830340163874,
      "learning_rate": 1.18691667635288e-05,
      "loss": 0.0447,
      "step": 5867
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3499069226343576,
      "learning_rate": 1.1862667594295086e-05,
      "loss": 0.0617,
      "step": 5868
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.24081491864947885,
      "learning_rate": 1.1856169454649864e-05,
      "loss": 0.0483,
      "step": 5869
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.25141724071104193,
      "learning_rate": 1.1849672345415306e-05,
      "loss": 0.0533,
      "step": 5870
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.27222725202385795,
      "learning_rate": 1.1843176267413488e-05,
      "loss": 0.0428,
      "step": 5871
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3824310543217218,
      "learning_rate": 1.1836681221466308e-05,
      "loss": 0.0532,
      "step": 5872
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2893092545660855,
      "learning_rate": 1.1830187208395594e-05,
      "loss": 0.0474,
      "step": 5873
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.28144731487803426,
      "learning_rate": 1.1823694229022995e-05,
      "loss": 0.0465,
      "step": 5874
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3432496287141752,
      "learning_rate": 1.1817202284170048e-05,
      "loss": 0.0497,
      "step": 5875
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.24584147222968614,
      "learning_rate": 1.181071137465815e-05,
      "loss": 0.0409,
      "step": 5876
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2918134927166957,
      "learning_rate": 1.1804221501308586e-05,
      "loss": 0.0607,
      "step": 5877
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43358907611149267,
      "learning_rate": 1.1797732664942481e-05,
      "loss": 0.0827,
      "step": 5878
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.22070516737271434,
      "learning_rate": 1.1791244866380862e-05,
      "loss": 0.0279,
      "step": 5879
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3062101610701623,
      "learning_rate": 1.1784758106444594e-05,
      "loss": 0.068,
      "step": 5880
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.21821269395759207,
      "learning_rate": 1.1778272385954432e-05,
      "loss": 0.0278,
      "step": 5881
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2992721201259001,
      "learning_rate": 1.1771787705730983e-05,
      "loss": 0.0442,
      "step": 5882
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2628429380927485,
      "learning_rate": 1.1765304066594747e-05,
      "loss": 0.0524,
      "step": 5883
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3784453351369725,
      "learning_rate": 1.175882146936606e-05,
      "loss": 0.0522,
      "step": 5884
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3227230409300173,
      "learning_rate": 1.1752339914865154e-05,
      "loss": 0.0377,
      "step": 5885
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.34260040654413293,
      "learning_rate": 1.1745859403912108e-05,
      "loss": 0.0507,
      "step": 5886
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.25826616542034253,
      "learning_rate": 1.1739379937326889e-05,
      "loss": 0.0575,
      "step": 5887
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2738917256501338,
      "learning_rate": 1.1732901515929312e-05,
      "loss": 0.036,
      "step": 5888
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31425453217515,
      "learning_rate": 1.1726424140539067e-05,
      "loss": 0.042,
      "step": 5889
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2844722984827675,
      "learning_rate": 1.1719947811975732e-05,
      "loss": 0.0541,
      "step": 5890
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.32665089592687435,
      "learning_rate": 1.1713472531058713e-05,
      "loss": 0.0517,
      "step": 5891
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.25860551766994555,
      "learning_rate": 1.1706998298607325e-05,
      "loss": 0.0503,
      "step": 5892
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3909095871899017,
      "learning_rate": 1.170052511544071e-05,
      "loss": 0.0608,
      "step": 5893
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3291160965152386,
      "learning_rate": 1.1694052982377915e-05,
      "loss": 0.0615,
      "step": 5894
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2655292647005712,
      "learning_rate": 1.1687581900237824e-05,
      "loss": 0.0513,
      "step": 5895
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.22471081474796067,
      "learning_rate": 1.1681111869839209e-05,
      "loss": 0.0364,
      "step": 5896
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.24770364518430704,
      "learning_rate": 1.1674642892000696e-05,
      "loss": 0.0407,
      "step": 5897
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.23060477816824915,
      "learning_rate": 1.166817496754078e-05,
      "loss": 0.0296,
      "step": 5898
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8201518310148757,
      "learning_rate": 1.1661708097277827e-05,
      "loss": 0.1434,
      "step": 5899
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.26472483302983024,
      "learning_rate": 1.1655242282030068e-05,
      "loss": 0.054,
      "step": 5900
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2857093933018677,
      "learning_rate": 1.1648777522615613e-05,
      "loss": 0.0478,
      "step": 5901
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2881781220441794,
      "learning_rate": 1.1642313819852405e-05,
      "loss": 0.0618,
      "step": 5902
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.23736697819741387,
      "learning_rate": 1.1635851174558282e-05,
      "loss": 0.0596,
      "step": 5903
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.30040539133514066,
      "learning_rate": 1.1629389587550939e-05,
      "loss": 0.0538,
      "step": 5904
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20434464576292674,
      "learning_rate": 1.162292905964795e-05,
      "loss": 0.0352,
      "step": 5905
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36130137988934835,
      "learning_rate": 1.1616469591666725e-05,
      "loss": 0.0552,
      "step": 5906
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.22679862510224041,
      "learning_rate": 1.1610011184424557e-05,
      "loss": 0.0382,
      "step": 5907
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.23509873070106485,
      "learning_rate": 1.1603553838738635e-05,
      "loss": 0.0301,
      "step": 5908
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3098879006382944,
      "learning_rate": 1.1597097555425954e-05,
      "loss": 0.0521,
      "step": 5909
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.24644970506449995,
      "learning_rate": 1.1590642335303417e-05,
      "loss": 0.0368,
      "step": 5910
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2397530859648159,
      "learning_rate": 1.1584188179187779e-05,
      "loss": 0.0302,
      "step": 5911
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2979598802067519,
      "learning_rate": 1.1577735087895664e-05,
      "loss": 0.0414,
      "step": 5912
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.25457638557545514,
      "learning_rate": 1.1571283062243563e-05,
      "loss": 0.0603,
      "step": 5913
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2361072895653942,
      "learning_rate": 1.1564832103047818e-05,
      "loss": 0.0248,
      "step": 5914
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.23804867447366063,
      "learning_rate": 1.155838221112465e-05,
      "loss": 0.0419,
      "step": 5915
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31250493522561706,
      "learning_rate": 1.1551933387290149e-05,
      "loss": 0.0471,
      "step": 5916
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3056899111607218,
      "learning_rate": 1.154548563236026e-05,
      "loss": 0.0528,
      "step": 5917
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.29692255442364485,
      "learning_rate": 1.1539038947150783e-05,
      "loss": 0.0679,
      "step": 5918
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2418046124257207,
      "learning_rate": 1.1532593332477406e-05,
      "loss": 0.0301,
      "step": 5919
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.34222985227632313,
      "learning_rate": 1.152614878915567e-05,
      "loss": 0.0537,
      "step": 5920
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.26048600545293615,
      "learning_rate": 1.1519705318000983e-05,
      "loss": 0.0373,
      "step": 5921
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2517852519658182,
      "learning_rate": 1.1513262919828603e-05,
      "loss": 0.0383,
      "step": 5922
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.26953854758293416,
      "learning_rate": 1.1506821595453674e-05,
      "loss": 0.0354,
      "step": 5923
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3077603809169237,
      "learning_rate": 1.1500381345691192e-05,
      "loss": 0.0527,
      "step": 5924
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2790589419498001,
      "learning_rate": 1.149394217135602e-05,
      "loss": 0.0486,
      "step": 5925
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37936917764007394,
      "learning_rate": 1.1487504073262886e-05,
      "loss": 0.0649,
      "step": 5926
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2314357433674794,
      "learning_rate": 1.1481067052226375e-05,
      "loss": 0.0354,
      "step": 5927
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.25326601150862577,
      "learning_rate": 1.1474631109060957e-05,
      "loss": 0.0411,
      "step": 5928
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2530355636928128,
      "learning_rate": 1.1468196244580927e-05,
      "loss": 0.0579,
      "step": 5929
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.32552842957852324,
      "learning_rate": 1.1461762459600476e-05,
      "loss": 0.0492,
      "step": 5930
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.25920532145792047,
      "learning_rate": 1.145532975493365e-05,
      "loss": 0.0352,
      "step": 5931
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3155234938457192,
      "learning_rate": 1.1448898131394364e-05,
      "loss": 0.0376,
      "step": 5932
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.23968142163553205,
      "learning_rate": 1.1442467589796374e-05,
      "loss": 0.0279,
      "step": 5933
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3115155807987206,
      "learning_rate": 1.1436038130953317e-05,
      "loss": 0.0614,
      "step": 5934
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.236974054148654,
      "learning_rate": 1.1429609755678697e-05,
      "loss": 0.0368,
      "step": 5935
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.23465142638232006,
      "learning_rate": 1.142318246478588e-05,
      "loss": 0.0358,
      "step": 5936
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.33548327721464166,
      "learning_rate": 1.1416756259088073e-05,
      "loss": 0.0664,
      "step": 5937
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.206655124667588,
      "learning_rate": 1.1410331139398365e-05,
      "loss": 0.0365,
      "step": 5938
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2631028853113967,
      "learning_rate": 1.1403907106529711e-05,
      "loss": 0.04,
      "step": 5939
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3378557176143698,
      "learning_rate": 1.1397484161294924e-05,
      "loss": 0.048,
      "step": 5940
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1746179446308563,
      "learning_rate": 1.1391062304506657e-05,
      "loss": 0.0292,
      "step": 5941
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31362214080142764,
      "learning_rate": 1.138464153697747e-05,
      "loss": 0.0537,
      "step": 5942
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3240398809803086,
      "learning_rate": 1.1378221859519756e-05,
      "loss": 0.0433,
      "step": 5943
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3175915427851454,
      "learning_rate": 1.1371803272945759e-05,
      "loss": 0.0608,
      "step": 5944
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.42304530121430417,
      "learning_rate": 1.1365385778067613e-05,
      "loss": 0.0779,
      "step": 5945
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.299659693942527,
      "learning_rate": 1.1358969375697297e-05,
      "loss": 0.0344,
      "step": 5946
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2857525873801302,
      "learning_rate": 1.1352554066646666e-05,
      "loss": 0.0418,
      "step": 5947
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3107186088511816,
      "learning_rate": 1.1346139851727412e-05,
      "loss": 0.0427,
      "step": 5948
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2548818917093416,
      "learning_rate": 1.1339726731751108e-05,
      "loss": 0.0375,
      "step": 5949
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3206887013099308,
      "learning_rate": 1.1333314707529188e-05,
      "loss": 0.0411,
      "step": 5950
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.24924795463950997,
      "learning_rate": 1.1326903779872951e-05,
      "loss": 0.0296,
      "step": 5951
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.22906372980721823,
      "learning_rate": 1.1320493949593528e-05,
      "loss": 0.0368,
      "step": 5952
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.21703269951146137,
      "learning_rate": 1.1314085217501947e-05,
      "loss": 0.0302,
      "step": 5953
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.21923790742634122,
      "learning_rate": 1.1307677584409076e-05,
      "loss": 0.0422,
      "step": 5954
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2632336825406231,
      "learning_rate": 1.1301271051125657e-05,
      "loss": 0.0432,
      "step": 5955
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.26538511271081955,
      "learning_rate": 1.1294865618462294e-05,
      "loss": 0.0292,
      "step": 5956
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2897215238464507,
      "learning_rate": 1.1288461287229423e-05,
      "loss": 0.032,
      "step": 5957
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.26655991896243997,
      "learning_rate": 1.128205805823737e-05,
      "loss": 0.0409,
      "step": 5958
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.28997380713858717,
      "learning_rate": 1.1275655932296327e-05,
      "loss": 0.049,
      "step": 5959
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2914115465516345,
      "learning_rate": 1.1269254910216316e-05,
      "loss": 0.0416,
      "step": 5960
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2507637249536318,
      "learning_rate": 1.1262854992807244e-05,
      "loss": 0.0453,
      "step": 5961
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4887002596703626,
      "learning_rate": 1.1256456180878867e-05,
      "loss": 0.0761,
      "step": 5962
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3180842702906934,
      "learning_rate": 1.1250058475240814e-05,
      "loss": 0.0688,
      "step": 5963
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3189294426842358,
      "learning_rate": 1.1243661876702552e-05,
      "loss": 0.0575,
      "step": 5964
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26943351521617026,
      "learning_rate": 1.1237266386073425e-05,
      "loss": 0.069,
      "step": 5965
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24653087123010536,
      "learning_rate": 1.1230872004162631e-05,
      "loss": 0.0326,
      "step": 5966
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3085977336771278,
      "learning_rate": 1.1224478731779238e-05,
      "loss": 0.0452,
      "step": 5967
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.22599576078876168,
      "learning_rate": 1.1218086569732152e-05,
      "loss": 0.0378,
      "step": 5968
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20021666436190477,
      "learning_rate": 1.1211695518830155e-05,
      "loss": 0.0433,
      "step": 5969
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.23652433494441713,
      "learning_rate": 1.1205305579881883e-05,
      "loss": 0.0379,
      "step": 5970
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.34138950594906714,
      "learning_rate": 1.1198916753695848e-05,
      "loss": 0.0427,
      "step": 5971
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20129466707321142,
      "learning_rate": 1.1192529041080382e-05,
      "loss": 0.0243,
      "step": 5972
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26716980791912504,
      "learning_rate": 1.1186142442843715e-05,
      "loss": 0.0383,
      "step": 5973
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3114285937541426,
      "learning_rate": 1.1179756959793918e-05,
      "loss": 0.0418,
      "step": 5974
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3405408287167941,
      "learning_rate": 1.1173372592738932e-05,
      "loss": 0.053,
      "step": 5975
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3566482593266611,
      "learning_rate": 1.1166989342486524e-05,
      "loss": 0.0669,
      "step": 5976
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2612097616099329,
      "learning_rate": 1.116060720984437e-05,
      "loss": 0.0287,
      "step": 5977
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.28235227431584947,
      "learning_rate": 1.1154226195619979e-05,
      "loss": 0.0393,
      "step": 5978
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2656755819965642,
      "learning_rate": 1.1147846300620702e-05,
      "loss": 0.0445,
      "step": 5979
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.33628503432571477,
      "learning_rate": 1.1141467525653773e-05,
      "loss": 0.0394,
      "step": 5980
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24418922703485943,
      "learning_rate": 1.1135089871526277e-05,
      "loss": 0.0246,
      "step": 5981
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20344604283808893,
      "learning_rate": 1.1128713339045162e-05,
      "loss": 0.0366,
      "step": 5982
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19663259705807523,
      "learning_rate": 1.112233792901722e-05,
      "loss": 0.0266,
      "step": 5983
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.503668174392788,
      "learning_rate": 1.1115963642249107e-05,
      "loss": 0.0832,
      "step": 5984
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.28452399458791766,
      "learning_rate": 1.1109590479547348e-05,
      "loss": 0.0445,
      "step": 5985
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20611582206201798,
      "learning_rate": 1.110321844171832e-05,
      "loss": 0.0483,
      "step": 5986
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.23977643071619714,
      "learning_rate": 1.1096847529568242e-05,
      "loss": 0.0397,
      "step": 5987
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35746460539326547,
      "learning_rate": 1.1090477743903212e-05,
      "loss": 0.0682,
      "step": 5988
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25062887250614924,
      "learning_rate": 1.1084109085529171e-05,
      "loss": 0.0277,
      "step": 5989
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.27691535369318027,
      "learning_rate": 1.1077741555251938e-05,
      "loss": 0.0547,
      "step": 5990
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2473799382507669,
      "learning_rate": 1.1071375153877156e-05,
      "loss": 0.0487,
      "step": 5991
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25727371791571674,
      "learning_rate": 1.1065009882210352e-05,
      "loss": 0.0435,
      "step": 5992
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24937420244973743,
      "learning_rate": 1.1058645741056892e-05,
      "loss": 0.0361,
      "step": 5993
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25467790573865035,
      "learning_rate": 1.1052282731222035e-05,
      "loss": 0.0456,
      "step": 5994
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.22825345054830226,
      "learning_rate": 1.1045920853510847e-05,
      "loss": 0.0405,
      "step": 5995
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.22472557710842322,
      "learning_rate": 1.1039560108728277e-05,
      "loss": 0.0368,
      "step": 5996
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3134589383408793,
      "learning_rate": 1.1033200497679138e-05,
      "loss": 0.0486,
      "step": 5997
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2563658934872525,
      "learning_rate": 1.1026842021168088e-05,
      "loss": 0.0357,
      "step": 5998
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3642069599562277,
      "learning_rate": 1.1020484679999632e-05,
      "loss": 0.0567,
      "step": 5999
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.27108506125561893,
      "learning_rate": 1.101412847497815e-05,
      "loss": 0.0503,
      "step": 6000
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3811401115016491,
      "learning_rate": 1.1007773406907866e-05,
      "loss": 0.0681,
      "step": 6001
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4295941858403881,
      "learning_rate": 1.100141947659288e-05,
      "loss": 0.059,
      "step": 6002
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26865020440489834,
      "learning_rate": 1.0995066684837113e-05,
      "loss": 0.0453,
      "step": 6003
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24963255044867685,
      "learning_rate": 1.0988715032444369e-05,
      "loss": 0.039,
      "step": 6004
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2642275570891078,
      "learning_rate": 1.09823645202183e-05,
      "loss": 0.0367,
      "step": 6005
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1542661189559467,
      "learning_rate": 1.0976015148962427e-05,
      "loss": 0.0245,
      "step": 6006
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37846384078067175,
      "learning_rate": 1.0969666919480093e-05,
      "loss": 0.0733,
      "step": 6007
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24245947517653435,
      "learning_rate": 1.0963319832574528e-05,
      "loss": 0.0385,
      "step": 6008
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24390120401484094,
      "learning_rate": 1.0956973889048807e-05,
      "loss": 0.0316,
      "step": 6009
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.23976926050695865,
      "learning_rate": 1.0950629089705857e-05,
      "loss": 0.0401,
      "step": 6010
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2545572298770189,
      "learning_rate": 1.0944285435348467e-05,
      "loss": 0.0394,
      "step": 6011
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.21670667816842418,
      "learning_rate": 1.0937942926779279e-05,
      "loss": 0.0386,
      "step": 6012
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.27973334346969486,
      "learning_rate": 1.0931601564800793e-05,
      "loss": 0.0548,
      "step": 6013
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19799726605996337,
      "learning_rate": 1.0925261350215344e-05,
      "loss": 0.0252,
      "step": 6014
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.27526029925432594,
      "learning_rate": 1.0918922283825147e-05,
      "loss": 0.0596,
      "step": 6015
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25491240757095157,
      "learning_rate": 1.091258436643226e-05,
      "loss": 0.0428,
      "step": 6016
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2977992903070403,
      "learning_rate": 1.0906247598838608e-05,
      "loss": 0.0334,
      "step": 6017
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26489547516956985,
      "learning_rate": 1.0899911981845946e-05,
      "loss": 0.0625,
      "step": 6018
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25002376622243755,
      "learning_rate": 1.08935775162559e-05,
      "loss": 0.0466,
      "step": 6019
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.22687852950017315,
      "learning_rate": 1.0887244202869951e-05,
      "loss": 0.037,
      "step": 6020
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24401706620678856,
      "learning_rate": 1.0880912042489438e-05,
      "loss": 0.0295,
      "step": 6021
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.28970854965652504,
      "learning_rate": 1.0874581035915534e-05,
      "loss": 0.0468,
      "step": 6022
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2742529019121395,
      "learning_rate": 1.0868251183949288e-05,
      "loss": 0.0442,
      "step": 6023
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2666533423363255,
      "learning_rate": 1.0861922487391588e-05,
      "loss": 0.0357,
      "step": 6024
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.23264806624161652,
      "learning_rate": 1.0855594947043198e-05,
      "loss": 0.0465,
      "step": 6025
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30432607537586237,
      "learning_rate": 1.0849268563704696e-05,
      "loss": 0.0422,
      "step": 6026
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.29694109733006746,
      "learning_rate": 1.0842943338176544e-05,
      "loss": 0.0496,
      "step": 6027
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.31908391637782896,
      "learning_rate": 1.0836619271259072e-05,
      "loss": 0.0554,
      "step": 6028
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2911529217143191,
      "learning_rate": 1.0830296363752417e-05,
      "loss": 0.0437,
      "step": 6029
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.22858501261181086,
      "learning_rate": 1.0823974616456607e-05,
      "loss": 0.0363,
      "step": 6030
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2868621696146127,
      "learning_rate": 1.0817654030171506e-05,
      "loss": 0.0447,
      "step": 6031
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30021178846248325,
      "learning_rate": 1.0811334605696837e-05,
      "loss": 0.0456,
      "step": 6032
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2845580999703231,
      "learning_rate": 1.0805016343832184e-05,
      "loss": 0.0345,
      "step": 6033
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2239709155666085,
      "learning_rate": 1.0798699245376959e-05,
      "loss": 0.032,
      "step": 6034
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2709629347851086,
      "learning_rate": 1.079238331113045e-05,
      "loss": 0.0475,
      "step": 6035
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2776401037867901,
      "learning_rate": 1.078606854189179e-05,
      "loss": 0.0451,
      "step": 6036
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2825026696003045,
      "learning_rate": 1.0779754938459975e-05,
      "loss": 0.0372,
      "step": 6037
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3195655067566507,
      "learning_rate": 1.0773442501633822e-05,
      "loss": 0.0563,
      "step": 6038
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.31929603817360847,
      "learning_rate": 1.0767131232212036e-05,
      "loss": 0.058,
      "step": 6039
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3237118147456888,
      "learning_rate": 1.0760821130993157e-05,
      "loss": 0.0812,
      "step": 6040
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3474861436077829,
      "learning_rate": 1.0754512198775586e-05,
      "loss": 0.0559,
      "step": 6041
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.22026692385715754,
      "learning_rate": 1.0748204436357562e-05,
      "loss": 0.0435,
      "step": 6042
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3939665539470694,
      "learning_rate": 1.0741897844537181e-05,
      "loss": 0.0638,
      "step": 6043
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2867646256003692,
      "learning_rate": 1.0735592424112404e-05,
      "loss": 0.0498,
      "step": 6044
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26499093334864254,
      "learning_rate": 1.0729288175881031e-05,
      "loss": 0.0531,
      "step": 6045
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2618291109752536,
      "learning_rate": 1.0722985100640717e-05,
      "loss": 0.0595,
      "step": 6046
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37681371453083634,
      "learning_rate": 1.0716683199188966e-05,
      "loss": 0.1102,
      "step": 6047
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.47743476711689026,
      "learning_rate": 1.0710382472323145e-05,
      "loss": 0.1082,
      "step": 6048
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20858470632829204,
      "learning_rate": 1.0704082920840448e-05,
      "loss": 0.0331,
      "step": 6049
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.308109869827298,
      "learning_rate": 1.0697784545537943e-05,
      "loss": 0.0586,
      "step": 6050
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3265372303033474,
      "learning_rate": 1.0691487347212541e-05,
      "loss": 0.0468,
      "step": 6051
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2388340993376687,
      "learning_rate": 1.0685191326661015e-05,
      "loss": 0.0573,
      "step": 6052
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37637518026499006,
      "learning_rate": 1.067889648467996e-05,
      "loss": 0.0681,
      "step": 6053
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.22860898450148412,
      "learning_rate": 1.0672602822065845e-05,
      "loss": 0.0334,
      "step": 6054
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20897343865416104,
      "learning_rate": 1.0666310339614996e-05,
      "loss": 0.0329,
      "step": 6055
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.22447446835167412,
      "learning_rate": 1.0660019038123577e-05,
      "loss": 0.0359,
      "step": 6056
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24142589334856104,
      "learning_rate": 1.0653728918387593e-05,
      "loss": 0.0321,
      "step": 6057
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.331008840441416,
      "learning_rate": 1.0647439981202918e-05,
      "loss": 0.0498,
      "step": 6058
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24544591252320738,
      "learning_rate": 1.0641152227365272e-05,
      "loss": 0.0437,
      "step": 6059
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.27342036057298125,
      "learning_rate": 1.0634865657670227e-05,
      "loss": 0.0658,
      "step": 6060
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2538071952684424,
      "learning_rate": 1.0628580272913184e-05,
      "loss": 0.0491,
      "step": 6061
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21203572872213403,
      "learning_rate": 1.0622296073889417e-05,
      "loss": 0.0415,
      "step": 6062
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20325100181832403,
      "learning_rate": 1.0616013061394063e-05,
      "loss": 0.0377,
      "step": 6063
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2242427487439591,
      "learning_rate": 1.0609731236222069e-05,
      "loss": 0.0393,
      "step": 6064
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.22696259129228208,
      "learning_rate": 1.0603450599168257e-05,
      "loss": 0.0434,
      "step": 6065
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.23051309967054032,
      "learning_rate": 1.0597171151027297e-05,
      "loss": 0.0438,
      "step": 6066
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.29261640341596007,
      "learning_rate": 1.059089289259371e-05,
      "loss": 0.0451,
      "step": 6067
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2515033433200905,
      "learning_rate": 1.058461582466185e-05,
      "loss": 0.0387,
      "step": 6068
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.28326375904149376,
      "learning_rate": 1.0578339948025943e-05,
      "loss": 0.0507,
      "step": 6069
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2577434360591503,
      "learning_rate": 1.0572065263480046e-05,
      "loss": 0.0349,
      "step": 6070
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2781431872604126,
      "learning_rate": 1.0565791771818082e-05,
      "loss": 0.0369,
      "step": 6071
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24526459414449023,
      "learning_rate": 1.0559519473833815e-05,
      "loss": 0.0228,
      "step": 6072
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.31440816285347983,
      "learning_rate": 1.0553248370320845e-05,
      "loss": 0.0407,
      "step": 6073
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.278018024316525,
      "learning_rate": 1.0546978462072642e-05,
      "loss": 0.0474,
      "step": 6074
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21992385483929613,
      "learning_rate": 1.0540709749882512e-05,
      "loss": 0.0425,
      "step": 6075
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.30762405387844527,
      "learning_rate": 1.0534442234543623e-05,
      "loss": 0.0633,
      "step": 6076
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.28382732398594485,
      "learning_rate": 1.0528175916848969e-05,
      "loss": 0.048,
      "step": 6077
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3165438906028394,
      "learning_rate": 1.0521910797591408e-05,
      "loss": 0.0459,
      "step": 6078
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3586557696197077,
      "learning_rate": 1.0515646877563646e-05,
      "loss": 0.0686,
      "step": 6079
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19745539045230392,
      "learning_rate": 1.0509384157558236e-05,
      "loss": 0.0378,
      "step": 6080
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2975245447211584,
      "learning_rate": 1.050312263836758e-05,
      "loss": 0.0578,
      "step": 6081
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2979477773885956,
      "learning_rate": 1.0496862320783926e-05,
      "loss": 0.0459,
      "step": 6082
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.27621262882293707,
      "learning_rate": 1.0490603205599373e-05,
      "loss": 0.0502,
      "step": 6083
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2763698025097821,
      "learning_rate": 1.0484345293605853e-05,
      "loss": 0.039,
      "step": 6084
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.27465241150577735,
      "learning_rate": 1.0478088585595167e-05,
      "loss": 0.0322,
      "step": 6085
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2622917365245004,
      "learning_rate": 1.0471833082358954e-05,
      "loss": 0.0411,
      "step": 6086
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3270997082302538,
      "learning_rate": 1.0465578784688705e-05,
      "loss": 0.0592,
      "step": 6087
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2739984480542414,
      "learning_rate": 1.0459325693375746e-05,
      "loss": 0.0405,
      "step": 6088
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3606096909379189,
      "learning_rate": 1.0453073809211262e-05,
      "loss": 0.0406,
      "step": 6089
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24210470076286697,
      "learning_rate": 1.0446823132986283e-05,
      "loss": 0.0426,
      "step": 6090
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.31585836658982,
      "learning_rate": 1.0440573665491693e-05,
      "loss": 0.0435,
      "step": 6091
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.30742888969745763,
      "learning_rate": 1.0434325407518204e-05,
      "loss": 0.0394,
      "step": 6092
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.26774327548582566,
      "learning_rate": 1.0428078359856389e-05,
      "loss": 0.03,
      "step": 6093
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2887744241350912,
      "learning_rate": 1.0421832523296665e-05,
      "loss": 0.0407,
      "step": 6094
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.22982731151598934,
      "learning_rate": 1.0415587898629307e-05,
      "loss": 0.0446,
      "step": 6095
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3673355940185556,
      "learning_rate": 1.04093444866444e-05,
      "loss": 0.0547,
      "step": 6096
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2920379220109524,
      "learning_rate": 1.0403102288131927e-05,
      "loss": 0.0337,
      "step": 6097
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.32639057571006813,
      "learning_rate": 1.0396861303881691e-05,
      "loss": 0.0383,
      "step": 6098
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2881678707186084,
      "learning_rate": 1.0390621534683327e-05,
      "loss": 0.0515,
      "step": 6099
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2887769783925161,
      "learning_rate": 1.0384382981326336e-05,
      "loss": 0.0371,
      "step": 6100
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3894467321730163,
      "learning_rate": 1.0378145644600063e-05,
      "loss": 0.0632,
      "step": 6101
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2946381245992665,
      "learning_rate": 1.0371909525293709e-05,
      "loss": 0.0619,
      "step": 6102
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.26255283959225645,
      "learning_rate": 1.0365674624196282e-05,
      "loss": 0.0343,
      "step": 6103
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2564118368550188,
      "learning_rate": 1.0359440942096682e-05,
      "loss": 0.0396,
      "step": 6104
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2441874192215508,
      "learning_rate": 1.0353208479783627e-05,
      "loss": 0.0585,
      "step": 6105
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.33642832806291767,
      "learning_rate": 1.0346977238045699e-05,
      "loss": 0.0324,
      "step": 6106
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3199003056298752,
      "learning_rate": 1.03407472176713e-05,
      "loss": 0.0482,
      "step": 6107
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.31635107336373197,
      "learning_rate": 1.0334518419448703e-05,
      "loss": 0.047,
      "step": 6108
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19773847731694738,
      "learning_rate": 1.0328290844166013e-05,
      "loss": 0.0408,
      "step": 6109
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2414813205818314,
      "learning_rate": 1.0322064492611195e-05,
      "loss": 0.0353,
      "step": 6110
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2547668690257741,
      "learning_rate": 1.0315839365572024e-05,
      "loss": 0.0389,
      "step": 6111
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.28245965116701477,
      "learning_rate": 1.0309615463836162e-05,
      "loss": 0.0278,
      "step": 6112
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.34706402817480203,
      "learning_rate": 1.0303392788191091e-05,
      "loss": 0.0493,
      "step": 6113
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.28962226744306535,
      "learning_rate": 1.0297171339424148e-05,
      "loss": 0.0601,
      "step": 6114
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3091420240469378,
      "learning_rate": 1.029095111832251e-05,
      "loss": 0.0483,
      "step": 6115
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20687722828693572,
      "learning_rate": 1.0284732125673198e-05,
      "loss": 0.0348,
      "step": 6116
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1967439457496283,
      "learning_rate": 1.0278514362263081e-05,
      "loss": 0.0247,
      "step": 6117
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.26030188256050835,
      "learning_rate": 1.0272297828878881e-05,
      "loss": 0.0347,
      "step": 6118
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2192649485250686,
      "learning_rate": 1.0266082526307138e-05,
      "loss": 0.033,
      "step": 6119
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2731743499987614,
      "learning_rate": 1.0259868455334259e-05,
      "loss": 0.0393,
      "step": 6120
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2501315575397304,
      "learning_rate": 1.025365561674649e-05,
      "loss": 0.038,
      "step": 6121
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.27388589061967406,
      "learning_rate": 1.0247444011329928e-05,
      "loss": 0.0485,
      "step": 6122
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2659913089746765,
      "learning_rate": 1.0241233639870487e-05,
      "loss": 0.0443,
      "step": 6123
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.27856050801137305,
      "learning_rate": 1.0235024503153956e-05,
      "loss": 0.0432,
      "step": 6124
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3500222054317568,
      "learning_rate": 1.0228816601965954e-05,
      "loss": 0.065,
      "step": 6125
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2888938308613042,
      "learning_rate": 1.0222609937091952e-05,
      "loss": 0.0663,
      "step": 6126
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36075980365991034,
      "learning_rate": 1.0216404509317245e-05,
      "loss": 0.0437,
      "step": 6127
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.26729182777538685,
      "learning_rate": 1.0210200319426988e-05,
      "loss": 0.0333,
      "step": 6128
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.18930065272486218,
      "learning_rate": 1.020399736820618e-05,
      "loss": 0.0232,
      "step": 6129
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24159397184248027,
      "learning_rate": 1.0197795656439662e-05,
      "loss": 0.0488,
      "step": 6130
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2852171088295083,
      "learning_rate": 1.0191595184912097e-05,
      "loss": 0.0467,
      "step": 6131
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.26951688921721084,
      "learning_rate": 1.0185395954408031e-05,
      "loss": 0.0403,
      "step": 6132
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2151563624185727,
      "learning_rate": 1.0179197965711825e-05,
      "loss": 0.0372,
      "step": 6133
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.27346715766287094,
      "learning_rate": 1.0173001219607683e-05,
      "loss": 0.0734,
      "step": 6134
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21720613907345318,
      "learning_rate": 1.016680571687966e-05,
      "loss": 0.0359,
      "step": 6135
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35256847987981105,
      "learning_rate": 1.0160611458311651e-05,
      "loss": 0.0506,
      "step": 6136
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2180379279230598,
      "learning_rate": 1.0154418444687405e-05,
      "loss": 0.0379,
      "step": 6137
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.43781994292571597,
      "learning_rate": 1.0148226676790482e-05,
      "loss": 0.0629,
      "step": 6138
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5429451203178703,
      "learning_rate": 1.0142036155404322e-05,
      "loss": 0.0842,
      "step": 6139
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3344424825827209,
      "learning_rate": 1.013584688131218e-05,
      "loss": 0.052,
      "step": 6140
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.26353648913176536,
      "learning_rate": 1.0129658855297177e-05,
      "loss": 0.0439,
      "step": 6141
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.25333014934561293,
      "learning_rate": 1.0123472078142248e-05,
      "loss": 0.0263,
      "step": 6142
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2446776016104054,
      "learning_rate": 1.0117286550630187e-05,
      "loss": 0.0454,
      "step": 6143
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37146994946827605,
      "learning_rate": 1.011110227354363e-05,
      "loss": 0.063,
      "step": 6144
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20723293485200756,
      "learning_rate": 1.0104919247665062e-05,
      "loss": 0.0336,
      "step": 6145
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3012933087973953,
      "learning_rate": 1.0098737473776781e-05,
      "loss": 0.0616,
      "step": 6146
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21922432190223268,
      "learning_rate": 1.0092556952660954e-05,
      "loss": 0.0324,
      "step": 6147
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21508036501792208,
      "learning_rate": 1.0086377685099578e-05,
      "loss": 0.0242,
      "step": 6148
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.31327340266038384,
      "learning_rate": 1.0080199671874509e-05,
      "loss": 0.0527,
      "step": 6149
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3166295899887907,
      "learning_rate": 1.0074022913767411e-05,
      "loss": 0.0533,
      "step": 6150
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23835244444077175,
      "learning_rate": 1.0067847411559816e-05,
      "loss": 0.031,
      "step": 6151
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.27785410064305965,
      "learning_rate": 1.006167316603309e-05,
      "loss": 0.0631,
      "step": 6152
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.29026907332210367,
      "learning_rate": 1.0055500177968445e-05,
      "loss": 0.0529,
      "step": 6153
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2585250803652672,
      "learning_rate": 1.0049328448146908e-05,
      "loss": 0.0382,
      "step": 6154
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.25937390729375265,
      "learning_rate": 1.0043157977349383e-05,
      "loss": 0.0349,
      "step": 6155
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3640380278126735,
      "learning_rate": 1.0036988766356592e-05,
      "loss": 0.0417,
      "step": 6156
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21365837459001652,
      "learning_rate": 1.0030820815949114e-05,
      "loss": 0.0381,
      "step": 6157
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.366832683468777,
      "learning_rate": 1.0024654126907343e-05,
      "loss": 0.0719,
      "step": 6158
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23757749347295498,
      "learning_rate": 1.0018488700011536e-05,
      "loss": 0.0401,
      "step": 6159
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.25865233066083226,
      "learning_rate": 1.0012324536041781e-05,
      "loss": 0.0388,
      "step": 6160
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2741600707433722,
      "learning_rate": 1.000616163577802e-05,
      "loss": 0.0414,
      "step": 6161
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21818814328311095,
      "learning_rate": 1.0000000000000006e-05,
      "loss": 0.0384,
      "step": 6162
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2668007187126608,
      "learning_rate": 9.993839629487357e-06,
      "loss": 0.043,
      "step": 6163
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3032125818418757,
      "learning_rate": 9.987680525019521e-06,
      "loss": 0.0546,
      "step": 6164
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.20366737507076754,
      "learning_rate": 9.981522687375793e-06,
      "loss": 0.0434,
      "step": 6165
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.41513302805212776,
      "learning_rate": 9.975366117335301e-06,
      "loss": 0.0825,
      "step": 6166
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.18871187927225874,
      "learning_rate": 9.969210815677016e-06,
      "loss": 0.0311,
      "step": 6167
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2552559120565852,
      "learning_rate": 9.96305678317975e-06,
      "loss": 0.041,
      "step": 6168
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.20703961697351564,
      "learning_rate": 9.956904020622137e-06,
      "loss": 0.0288,
      "step": 6169
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21144496839471708,
      "learning_rate": 9.950752528782679e-06,
      "loss": 0.0392,
      "step": 6170
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2645127205349823,
      "learning_rate": 9.944602308439696e-06,
      "loss": 0.0419,
      "step": 6171
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2682123672755254,
      "learning_rate": 9.938453360371363e-06,
      "loss": 0.0515,
      "step": 6172
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2751075290564487,
      "learning_rate": 9.932305685355672e-06,
      "loss": 0.0512,
      "step": 6173
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.31168776097582196,
      "learning_rate": 9.926159284170471e-06,
      "loss": 0.0542,
      "step": 6174
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.26094750025425906,
      "learning_rate": 9.92001415759345e-06,
      "loss": 0.0435,
      "step": 6175
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.26129550302568794,
      "learning_rate": 9.913870306402129e-06,
      "loss": 0.0493,
      "step": 6176
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21577135454231355,
      "learning_rate": 9.90772773137386e-06,
      "loss": 0.0335,
      "step": 6177
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2364096540318172,
      "learning_rate": 9.901586433285845e-06,
      "loss": 0.0268,
      "step": 6178
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.36304646770470045,
      "learning_rate": 9.895446412915124e-06,
      "loss": 0.0426,
      "step": 6179
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2856261027333615,
      "learning_rate": 9.889307671038579e-06,
      "loss": 0.0472,
      "step": 6180
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23325696352471895,
      "learning_rate": 9.88317020843291e-06,
      "loss": 0.0365,
      "step": 6181
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.31977235795833003,
      "learning_rate": 9.877034025874675e-06,
      "loss": 0.0406,
      "step": 6182
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2558078859660114,
      "learning_rate": 9.87089912414026e-06,
      "loss": 0.0474,
      "step": 6183
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.25894290022360145,
      "learning_rate": 9.864765504005901e-06,
      "loss": 0.0463,
      "step": 6184
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3540219848225874,
      "learning_rate": 9.858633166247659e-06,
      "loss": 0.0508,
      "step": 6185
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2029467038610122,
      "learning_rate": 9.852502111641438e-06,
      "loss": 0.0466,
      "step": 6186
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2655416526179896,
      "learning_rate": 9.846372340962987e-06,
      "loss": 0.0504,
      "step": 6187
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.22159925337566252,
      "learning_rate": 9.840243854987868e-06,
      "loss": 0.034,
      "step": 6188
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.25748234618882093,
      "learning_rate": 9.834116654491506e-06,
      "loss": 0.0381,
      "step": 6189
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23850988144367538,
      "learning_rate": 9.827990740249156e-06,
      "loss": 0.0311,
      "step": 6190
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.22772029486391687,
      "learning_rate": 9.821866113035905e-06,
      "loss": 0.0351,
      "step": 6191
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21421351326928012,
      "learning_rate": 9.815742773626693e-06,
      "loss": 0.0477,
      "step": 6192
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2757065766470592,
      "learning_rate": 9.809620722796265e-06,
      "loss": 0.0277,
      "step": 6193
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.22013983115002259,
      "learning_rate": 9.803499961319234e-06,
      "loss": 0.0276,
      "step": 6194
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.25999750099448354,
      "learning_rate": 9.79738048997004e-06,
      "loss": 0.0389,
      "step": 6195
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3546731030067183,
      "learning_rate": 9.791262309522959e-06,
      "loss": 0.0785,
      "step": 6196
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.20879912856185506,
      "learning_rate": 9.785145420752096e-06,
      "loss": 0.0355,
      "step": 6197
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2584543474741684,
      "learning_rate": 9.779029824431403e-06,
      "loss": 0.0402,
      "step": 6198
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4245585462929971,
      "learning_rate": 9.772915521334664e-06,
      "loss": 0.1176,
      "step": 6199
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23702502156362043,
      "learning_rate": 9.766802512235507e-06,
      "loss": 0.0426,
      "step": 6200
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2706143000912401,
      "learning_rate": 9.760690797907383e-06,
      "loss": 0.0328,
      "step": 6201
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2615778814371287,
      "learning_rate": 9.75458037912359e-06,
      "loss": 0.0471,
      "step": 6202
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2757278433896372,
      "learning_rate": 9.748471256657264e-06,
      "loss": 0.0568,
      "step": 6203
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2978884564672695,
      "learning_rate": 9.742363431281356e-06,
      "loss": 0.0446,
      "step": 6204
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.20370410933347816,
      "learning_rate": 9.736256903768678e-06,
      "loss": 0.0338,
      "step": 6205
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.25345471654097956,
      "learning_rate": 9.73015167489186e-06,
      "loss": 0.0248,
      "step": 6206
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34774019535141376,
      "learning_rate": 9.724047745423396e-06,
      "loss": 0.0428,
      "step": 6207
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32560227964655303,
      "learning_rate": 9.717945116135568e-06,
      "loss": 0.0554,
      "step": 6208
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3232524098050541,
      "learning_rate": 9.711843787800535e-06,
      "loss": 0.0595,
      "step": 6209
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.27333508344609464,
      "learning_rate": 9.705743761190273e-06,
      "loss": 0.0341,
      "step": 6210
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2773933903501465,
      "learning_rate": 9.699645037076608e-06,
      "loss": 0.0321,
      "step": 6211
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2615605060952595,
      "learning_rate": 9.693547616231173e-06,
      "loss": 0.0653,
      "step": 6212
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.290546923065797,
      "learning_rate": 9.687451499425465e-06,
      "loss": 0.0371,
      "step": 6213
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.427013940135805,
      "learning_rate": 9.681356687430798e-06,
      "loss": 0.069,
      "step": 6214
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21506138961979826,
      "learning_rate": 9.67526318101834e-06,
      "loss": 0.0361,
      "step": 6215
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2641080709783023,
      "learning_rate": 9.669170980959063e-06,
      "loss": 0.0407,
      "step": 6216
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2623568769105091,
      "learning_rate": 9.663080088023795e-06,
      "loss": 0.0265,
      "step": 6217
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34069439329886586,
      "learning_rate": 9.656990502983216e-06,
      "loss": 0.0487,
      "step": 6218
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32778804371449066,
      "learning_rate": 9.650902226607802e-06,
      "loss": 0.0585,
      "step": 6219
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.241926490973222,
      "learning_rate": 9.644815259667881e-06,
      "loss": 0.0361,
      "step": 6220
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.26820518639536545,
      "learning_rate": 9.638729602933621e-06,
      "loss": 0.0394,
      "step": 6221
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23553373480436743,
      "learning_rate": 9.632645257175027e-06,
      "loss": 0.0484,
      "step": 6222
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2943329900177595,
      "learning_rate": 9.626562223161916e-06,
      "loss": 0.041,
      "step": 6223
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23571131740363835,
      "learning_rate": 9.620480501663954e-06,
      "loss": 0.0337,
      "step": 6224
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.28536693419321146,
      "learning_rate": 9.614400093450649e-06,
      "loss": 0.0471,
      "step": 6225
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.24616376698227518,
      "learning_rate": 9.608320999291333e-06,
      "loss": 0.0287,
      "step": 6226
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3335323509513379,
      "learning_rate": 9.602243219955168e-06,
      "loss": 0.0551,
      "step": 6227
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2093054364067025,
      "learning_rate": 9.59616675621115e-06,
      "loss": 0.0299,
      "step": 6228
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23734024943067492,
      "learning_rate": 9.590091608828123e-06,
      "loss": 0.0238,
      "step": 6229
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.28447029487318604,
      "learning_rate": 9.58401777857475e-06,
      "loss": 0.0421,
      "step": 6230
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23408007344432577,
      "learning_rate": 9.577945266219536e-06,
      "loss": 0.0388,
      "step": 6231
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32076666199095033,
      "learning_rate": 9.571874072530809e-06,
      "loss": 0.0554,
      "step": 6232
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3434619129865472,
      "learning_rate": 9.565804198276733e-06,
      "loss": 0.0454,
      "step": 6233
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23862960241284548,
      "learning_rate": 9.559735644225316e-06,
      "loss": 0.0264,
      "step": 6234
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23075814585395765,
      "learning_rate": 9.553668411144387e-06,
      "loss": 0.0266,
      "step": 6235
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23590381615550915,
      "learning_rate": 9.547602499801616e-06,
      "loss": 0.0393,
      "step": 6236
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.28383059214184636,
      "learning_rate": 9.541537910964495e-06,
      "loss": 0.0417,
      "step": 6237
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2748857293266264,
      "learning_rate": 9.53547464540037e-06,
      "loss": 0.0565,
      "step": 6238
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2277131294822672,
      "learning_rate": 9.529412703876387e-06,
      "loss": 0.0465,
      "step": 6239
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24707902734637896,
      "learning_rate": 9.523352087159548e-06,
      "loss": 0.0327,
      "step": 6240
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24524278956253664,
      "learning_rate": 9.517292796016684e-06,
      "loss": 0.0446,
      "step": 6241
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.26185331159198083,
      "learning_rate": 9.511234831214464e-06,
      "loss": 0.034,
      "step": 6242
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24203865030105287,
      "learning_rate": 9.505178193519366e-06,
      "loss": 0.0372,
      "step": 6243
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23152872184127937,
      "learning_rate": 9.499122883697724e-06,
      "loss": 0.0395,
      "step": 6244
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2965330362908359,
      "learning_rate": 9.493068902515692e-06,
      "loss": 0.0503,
      "step": 6245
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2977029898884496,
      "learning_rate": 9.487016250739269e-06,
      "loss": 0.0421,
      "step": 6246
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3416215610484597,
      "learning_rate": 9.480964929134263e-06,
      "loss": 0.0375,
      "step": 6247
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2732846377747829,
      "learning_rate": 9.474914938466328e-06,
      "loss": 0.0352,
      "step": 6248
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.26970197274105745,
      "learning_rate": 9.468866279500956e-06,
      "loss": 0.0488,
      "step": 6249
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2600905028881518,
      "learning_rate": 9.462818953003465e-06,
      "loss": 0.0309,
      "step": 6250
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2737487928372596,
      "learning_rate": 9.45677295973899e-06,
      "loss": 0.0381,
      "step": 6251
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.26136220302113367,
      "learning_rate": 9.45072830047251e-06,
      "loss": 0.0638,
      "step": 6252
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.28952782782782804,
      "learning_rate": 9.444684975968852e-06,
      "loss": 0.0477,
      "step": 6253
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.30617854053418464,
      "learning_rate": 9.438642986992641e-06,
      "loss": 0.0475,
      "step": 6254
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.234407231975574,
      "learning_rate": 9.432602334308353e-06,
      "loss": 0.0333,
      "step": 6255
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24852066147661583,
      "learning_rate": 9.426563018680293e-06,
      "loss": 0.0274,
      "step": 6256
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3931757993883918,
      "learning_rate": 9.420525040872602e-06,
      "loss": 0.0518,
      "step": 6257
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.22685950641429906,
      "learning_rate": 9.414488401649227e-06,
      "loss": 0.0472,
      "step": 6258
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2941157965040893,
      "learning_rate": 9.408453101773971e-06,
      "loss": 0.0386,
      "step": 6259
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.31684167359391574,
      "learning_rate": 9.40241914201046e-06,
      "loss": 0.0494,
      "step": 6260
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4020733341232103,
      "learning_rate": 9.396386523122158e-06,
      "loss": 0.0773,
      "step": 6261
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.26485523716593634,
      "learning_rate": 9.390355245872337e-06,
      "loss": 0.0363,
      "step": 6262
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2767078066427714,
      "learning_rate": 9.384325311024123e-06,
      "loss": 0.0421,
      "step": 6263
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2760967643499841,
      "learning_rate": 9.378296719340459e-06,
      "loss": 0.0384,
      "step": 6264
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20478149032404033,
      "learning_rate": 9.372269471584128e-06,
      "loss": 0.043,
      "step": 6265
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2691958040055307,
      "learning_rate": 9.366243568517726e-06,
      "loss": 0.0493,
      "step": 6266
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23869043143850346,
      "learning_rate": 9.360219010903695e-06,
      "loss": 0.0287,
      "step": 6267
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20597591973733836,
      "learning_rate": 9.354195799504305e-06,
      "loss": 0.0376,
      "step": 6268
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.209775496806426,
      "learning_rate": 9.348173935081645e-06,
      "loss": 0.0381,
      "step": 6269
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2688287619572404,
      "learning_rate": 9.342153418397647e-06,
      "loss": 0.0486,
      "step": 6270
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2902534796894734,
      "learning_rate": 9.336134250214061e-06,
      "loss": 0.0445,
      "step": 6271
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24861511712883994,
      "learning_rate": 9.330116431292478e-06,
      "loss": 0.049,
      "step": 6272
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2843546115986259,
      "learning_rate": 9.324099962394313e-06,
      "loss": 0.0598,
      "step": 6273
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2661283716245321,
      "learning_rate": 9.318084844280798e-06,
      "loss": 0.0457,
      "step": 6274
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2794483547640721,
      "learning_rate": 9.31207107771301e-06,
      "loss": 0.0284,
      "step": 6275
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23227527547035146,
      "learning_rate": 9.306058663451852e-06,
      "loss": 0.0399,
      "step": 6276
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.25572760297267416,
      "learning_rate": 9.30004760225806e-06,
      "loss": 0.0422,
      "step": 6277
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2265836442749071,
      "learning_rate": 9.294037894892178e-06,
      "loss": 0.0466,
      "step": 6278
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38933889383896975,
      "learning_rate": 9.288029542114602e-06,
      "loss": 0.0622,
      "step": 6279
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20071796948873447,
      "learning_rate": 9.28202254468555e-06,
      "loss": 0.0225,
      "step": 6280
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.31054400796586457,
      "learning_rate": 9.276016903365068e-06,
      "loss": 0.0523,
      "step": 6281
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.27328691423237594,
      "learning_rate": 9.270012618913018e-06,
      "loss": 0.0473,
      "step": 6282
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2255796621531451,
      "learning_rate": 9.26400969208911e-06,
      "loss": 0.0467,
      "step": 6283
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3017817737732413,
      "learning_rate": 9.258008123652868e-06,
      "loss": 0.0534,
      "step": 6284
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40936924952188236,
      "learning_rate": 9.252007914363664e-06,
      "loss": 0.0751,
      "step": 6285
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23785788807231603,
      "learning_rate": 9.246009064980657e-06,
      "loss": 0.0377,
      "step": 6286
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.27980544730699336,
      "learning_rate": 9.240011576262887e-06,
      "loss": 0.0452,
      "step": 6287
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2829410787583162,
      "learning_rate": 9.23401544896919e-06,
      "loss": 0.0435,
      "step": 6288
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2584916042720561,
      "learning_rate": 9.228020683858223e-06,
      "loss": 0.0481,
      "step": 6289
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24872340659220568,
      "learning_rate": 9.22202728168849e-06,
      "loss": 0.0384,
      "step": 6290
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.25803977889566404,
      "learning_rate": 9.216035243218318e-06,
      "loss": 0.0372,
      "step": 6291
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2000381038548197,
      "learning_rate": 9.210044569205863e-06,
      "loss": 0.0246,
      "step": 6292
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20023316746532552,
      "learning_rate": 9.204055260409091e-06,
      "loss": 0.041,
      "step": 6293
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24988703857874645,
      "learning_rate": 9.198067317585816e-06,
      "loss": 0.0375,
      "step": 6294
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2536094544789971,
      "learning_rate": 9.19208074149367e-06,
      "loss": 0.0512,
      "step": 6295
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23277586738353942,
      "learning_rate": 9.186095532890121e-06,
      "loss": 0.0464,
      "step": 6296
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23146955950597894,
      "learning_rate": 9.180111692532446e-06,
      "loss": 0.0473,
      "step": 6297
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20974903502827713,
      "learning_rate": 9.174129221177762e-06,
      "loss": 0.0328,
      "step": 6298
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2503838869994431,
      "learning_rate": 9.168148119583011e-06,
      "loss": 0.0317,
      "step": 6299
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2928620462015423,
      "learning_rate": 9.162168388504972e-06,
      "loss": 0.0395,
      "step": 6300
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2828001377923487,
      "learning_rate": 9.156190028700226e-06,
      "loss": 0.0385,
      "step": 6301
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.21102328675309784,
      "learning_rate": 9.150213040925193e-06,
      "loss": 0.0382,
      "step": 6302
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2844854590968907,
      "learning_rate": 9.144237425936129e-06,
      "loss": 0.0518,
      "step": 6303
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4165599646637808,
      "learning_rate": 9.138263184489104e-06,
      "loss": 0.0648,
      "step": 6304
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2724789302449723,
      "learning_rate": 9.132290317340018e-06,
      "loss": 0.0393,
      "step": 6305
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3027393094552803,
      "learning_rate": 9.1263188252446e-06,
      "loss": 0.0499,
      "step": 6306
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2533256789016248,
      "learning_rate": 9.120348708958399e-06,
      "loss": 0.0465,
      "step": 6307
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2664072914886253,
      "learning_rate": 9.114379969236802e-06,
      "loss": 0.0305,
      "step": 6308
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2437821694077894,
      "learning_rate": 9.108412606834998e-06,
      "loss": 0.0377,
      "step": 6309
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2879293371748821,
      "learning_rate": 9.102446622508025e-06,
      "loss": 0.0502,
      "step": 6310
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2503264203528756,
      "learning_rate": 9.096482017010737e-06,
      "loss": 0.0295,
      "step": 6311
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.25013961469818086,
      "learning_rate": 9.090518791097822e-06,
      "loss": 0.0444,
      "step": 6312
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2477486011025846,
      "learning_rate": 9.084556945523772e-06,
      "loss": 0.0471,
      "step": 6313
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4334807893165917,
      "learning_rate": 9.078596481042927e-06,
      "loss": 0.0818,
      "step": 6314
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23353296656458178,
      "learning_rate": 9.072637398409444e-06,
      "loss": 0.0189,
      "step": 6315
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2954942567770665,
      "learning_rate": 9.066679698377311e-06,
      "loss": 0.0332,
      "step": 6316
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23530170498235112,
      "learning_rate": 9.060723381700323e-06,
      "loss": 0.0408,
      "step": 6317
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24766727747869663,
      "learning_rate": 9.054768449132115e-06,
      "loss": 0.0323,
      "step": 6318
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24632032500021417,
      "learning_rate": 9.048814901426146e-06,
      "loss": 0.0426,
      "step": 6319
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.28128099270764245,
      "learning_rate": 9.042862739335707e-06,
      "loss": 0.0341,
      "step": 6320
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24277883295031283,
      "learning_rate": 9.036911963613881e-06,
      "loss": 0.0393,
      "step": 6321
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23686681781114505,
      "learning_rate": 9.030962575013622e-06,
      "loss": 0.0443,
      "step": 6322
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.26004526675001527,
      "learning_rate": 9.025014574287685e-06,
      "loss": 0.0284,
      "step": 6323
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2446766652525605,
      "learning_rate": 9.019067962188634e-06,
      "loss": 0.0409,
      "step": 6324
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.22322480074775033,
      "learning_rate": 9.013122739468882e-06,
      "loss": 0.038,
      "step": 6325
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.21102602302103093,
      "learning_rate": 9.007178906880655e-06,
      "loss": 0.0264,
      "step": 6326
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2739814252996459,
      "learning_rate": 9.001236465176017e-06,
      "loss": 0.045,
      "step": 6327
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2962305200988842,
      "learning_rate": 8.995295415106829e-06,
      "loss": 0.0426,
      "step": 6328
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2513516069280048,
      "learning_rate": 8.989355757424796e-06,
      "loss": 0.0295,
      "step": 6329
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.23622794431841201,
      "learning_rate": 8.983417492881443e-06,
      "loss": 0.0447,
      "step": 6330
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20425040918306897,
      "learning_rate": 8.977480622228128e-06,
      "loss": 0.0236,
      "step": 6331
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.211285339649907,
      "learning_rate": 8.971545146216005e-06,
      "loss": 0.0359,
      "step": 6332
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.272476291565687,
      "learning_rate": 8.965611065596077e-06,
      "loss": 0.0427,
      "step": 6333
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21597569253476873,
      "learning_rate": 8.959678381119166e-06,
      "loss": 0.0366,
      "step": 6334
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3310963539752032,
      "learning_rate": 8.953747093535918e-06,
      "loss": 0.037,
      "step": 6335
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2095192366163649,
      "learning_rate": 8.947817203596785e-06,
      "loss": 0.021,
      "step": 6336
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2470472724635825,
      "learning_rate": 8.94188871205206e-06,
      "loss": 0.0338,
      "step": 6337
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20715595484212052,
      "learning_rate": 8.935961619651859e-06,
      "loss": 0.0407,
      "step": 6338
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22768347581955933,
      "learning_rate": 8.930035927146114e-06,
      "loss": 0.0345,
      "step": 6339
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3099342517386408,
      "learning_rate": 8.924111635284582e-06,
      "loss": 0.0488,
      "step": 6340
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.43779650926487895,
      "learning_rate": 8.918188744816844e-06,
      "loss": 0.087,
      "step": 6341
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2332687975555119,
      "learning_rate": 8.91226725649231e-06,
      "loss": 0.0425,
      "step": 6342
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19567928207038063,
      "learning_rate": 8.906347171060191e-06,
      "loss": 0.0229,
      "step": 6343
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.31602631823126964,
      "learning_rate": 8.900428489269541e-06,
      "loss": 0.0489,
      "step": 6344
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.23726831950245672,
      "learning_rate": 8.89451121186923e-06,
      "loss": 0.0343,
      "step": 6345
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.31267871514804374,
      "learning_rate": 8.888595339607961e-06,
      "loss": 0.0524,
      "step": 6346
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2462707291872784,
      "learning_rate": 8.882680873234233e-06,
      "loss": 0.0358,
      "step": 6347
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22789185315397206,
      "learning_rate": 8.876767813496388e-06,
      "loss": 0.0307,
      "step": 6348
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2658545259744762,
      "learning_rate": 8.870856161142587e-06,
      "loss": 0.04,
      "step": 6349
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34409919255540783,
      "learning_rate": 8.86494591692081e-06,
      "loss": 0.0492,
      "step": 6350
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2679736145289774,
      "learning_rate": 8.859037081578867e-06,
      "loss": 0.0489,
      "step": 6351
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2568584475787984,
      "learning_rate": 8.85312965586437e-06,
      "loss": 0.044,
      "step": 6352
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2876786537293762,
      "learning_rate": 8.84722364052477e-06,
      "loss": 0.0575,
      "step": 6353
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.23660125397348275,
      "learning_rate": 8.841319036307334e-06,
      "loss": 0.0409,
      "step": 6354
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2962576571010167,
      "learning_rate": 8.835415843959154e-06,
      "loss": 0.0576,
      "step": 6355
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3155991184840565,
      "learning_rate": 8.829514064227138e-06,
      "loss": 0.065,
      "step": 6356
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20948942601939022,
      "learning_rate": 8.82361369785802e-06,
      "loss": 0.0306,
      "step": 6357
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3294762535742318,
      "learning_rate": 8.817714745598358e-06,
      "loss": 0.0462,
      "step": 6358
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25829241618292653,
      "learning_rate": 8.811817208194512e-06,
      "loss": 0.0466,
      "step": 6359
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2077765006958192,
      "learning_rate": 8.805921086392686e-06,
      "loss": 0.0253,
      "step": 6360
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25343540256634217,
      "learning_rate": 8.800026380938895e-06,
      "loss": 0.0291,
      "step": 6361
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25178844777491177,
      "learning_rate": 8.79413309257898e-06,
      "loss": 0.04,
      "step": 6362
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20695157587285948,
      "learning_rate": 8.788241222058588e-06,
      "loss": 0.0331,
      "step": 6363
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.27255576886387817,
      "learning_rate": 8.782350770123202e-06,
      "loss": 0.0479,
      "step": 6364
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25812254671311835,
      "learning_rate": 8.776461737518123e-06,
      "loss": 0.0433,
      "step": 6365
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22198138104472048,
      "learning_rate": 8.770574124988474e-06,
      "loss": 0.0369,
      "step": 6366
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.27744670085426953,
      "learning_rate": 8.764687933279179e-06,
      "loss": 0.0459,
      "step": 6367
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.28767703503924624,
      "learning_rate": 8.758803163135008e-06,
      "loss": 0.0489,
      "step": 6368
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2544070367064107,
      "learning_rate": 8.752919815300541e-06,
      "loss": 0.0314,
      "step": 6369
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22781994064435185,
      "learning_rate": 8.74703789052018e-06,
      "loss": 0.0423,
      "step": 6370
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2631358030733747,
      "learning_rate": 8.741157389538132e-06,
      "loss": 0.0474,
      "step": 6371
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22057342938330163,
      "learning_rate": 8.73527831309844e-06,
      "loss": 0.0319,
      "step": 6372
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21460137133260163,
      "learning_rate": 8.729400661944984e-06,
      "loss": 0.0261,
      "step": 6373
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4747864820535669,
      "learning_rate": 8.723524436821418e-06,
      "loss": 0.0668,
      "step": 6374
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2662109055942396,
      "learning_rate": 8.717649638471249e-06,
      "loss": 0.0374,
      "step": 6375
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3418576431750124,
      "learning_rate": 8.711776267637794e-06,
      "loss": 0.0402,
      "step": 6376
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22465527107035038,
      "learning_rate": 8.705904325064201e-06,
      "loss": 0.0398,
      "step": 6377
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3423718439272161,
      "learning_rate": 8.700033811493407e-06,
      "loss": 0.0462,
      "step": 6378
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33873568135084464,
      "learning_rate": 8.694164727668199e-06,
      "loss": 0.0808,
      "step": 6379
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.30990123209805315,
      "learning_rate": 8.688297074331171e-06,
      "loss": 0.0386,
      "step": 6380
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37951634634947395,
      "learning_rate": 8.682430852224743e-06,
      "loss": 0.0377,
      "step": 6381
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.23587397610735772,
      "learning_rate": 8.676566062091135e-06,
      "loss": 0.0288,
      "step": 6382
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.28234982026440475,
      "learning_rate": 8.670702704672403e-06,
      "loss": 0.0344,
      "step": 6383
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.26419354860571054,
      "learning_rate": 8.66484078071042e-06,
      "loss": 0.043,
      "step": 6384
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.27265589585595507,
      "learning_rate": 8.658980290946881e-06,
      "loss": 0.0465,
      "step": 6385
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24813252569743952,
      "learning_rate": 8.653121236123278e-06,
      "loss": 0.0286,
      "step": 6386
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2670370325184495,
      "learning_rate": 8.647263616980948e-06,
      "loss": 0.0443,
      "step": 6387
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.26468644098027216,
      "learning_rate": 8.641407434261031e-06,
      "loss": 0.0357,
      "step": 6388
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22262329560397953,
      "learning_rate": 8.635552688704491e-06,
      "loss": 0.0307,
      "step": 6389
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33227491974359313,
      "learning_rate": 8.62969938105211e-06,
      "loss": 0.0605,
      "step": 6390
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.17559031607289025,
      "learning_rate": 8.623847512044485e-06,
      "loss": 0.0195,
      "step": 6391
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2930631612606321,
      "learning_rate": 8.617997082422031e-06,
      "loss": 0.0521,
      "step": 6392
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2828701824565216,
      "learning_rate": 8.612148092924994e-06,
      "loss": 0.0571,
      "step": 6393
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22517277918334888,
      "learning_rate": 8.606300544293412e-06,
      "loss": 0.0282,
      "step": 6394
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2827000589408449,
      "learning_rate": 8.600454437267158e-06,
      "loss": 0.0421,
      "step": 6395
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3838056798545468,
      "learning_rate": 8.594609772585922e-06,
      "loss": 0.0664,
      "step": 6396
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.28354324437547307,
      "learning_rate": 8.588766550989218e-06,
      "loss": 0.0401,
      "step": 6397
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24436393443223262,
      "learning_rate": 8.582924773216353e-06,
      "loss": 0.0245,
      "step": 6398
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.23143113169303037,
      "learning_rate": 8.577084440006471e-06,
      "loss": 0.0372,
      "step": 6399
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34290995691316134,
      "learning_rate": 8.571245552098533e-06,
      "loss": 0.0557,
      "step": 6400
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2562108137825738,
      "learning_rate": 8.565408110231319e-06,
      "loss": 0.0302,
      "step": 6401
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3028696311777119,
      "learning_rate": 8.559572115143406e-06,
      "loss": 0.0328,
      "step": 6402
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2720516507881614,
      "learning_rate": 8.55373756757321e-06,
      "loss": 0.041,
      "step": 6403
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.26955753695798945,
      "learning_rate": 8.547904468258957e-06,
      "loss": 0.0369,
      "step": 6404
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.205862136112929,
      "learning_rate": 8.542072817938693e-06,
      "loss": 0.032,
      "step": 6405
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.26788001189925775,
      "learning_rate": 8.536242617350265e-06,
      "loss": 0.0511,
      "step": 6406
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25751294448408285,
      "learning_rate": 8.530413867231347e-06,
      "loss": 0.0468,
      "step": 6407
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24201039021822887,
      "learning_rate": 8.524586568319451e-06,
      "loss": 0.0411,
      "step": 6408
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25891186661461163,
      "learning_rate": 8.518760721351865e-06,
      "loss": 0.034,
      "step": 6409
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2302191239125216,
      "learning_rate": 8.51293632706572e-06,
      "loss": 0.0334,
      "step": 6410
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.28851454580948577,
      "learning_rate": 8.507113386197956e-06,
      "loss": 0.0474,
      "step": 6411
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5054794711312898,
      "learning_rate": 8.501291899485337e-06,
      "loss": 0.0753,
      "step": 6412
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4117383688029116,
      "learning_rate": 8.495471867664423e-06,
      "loss": 0.0789,
      "step": 6413
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2344748681602388,
      "learning_rate": 8.489653291471607e-06,
      "loss": 0.0333,
      "step": 6414
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2514175963227083,
      "learning_rate": 8.483836171643094e-06,
      "loss": 0.0308,
      "step": 6415
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24590569834635775,
      "learning_rate": 8.47802050891491e-06,
      "loss": 0.0477,
      "step": 6416
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2268210366467165,
      "learning_rate": 8.472206304022881e-06,
      "loss": 0.0397,
      "step": 6417
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.39791710846894723,
      "learning_rate": 8.466393557702659e-06,
      "loss": 0.0534,
      "step": 6418
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.26604481787571554,
      "learning_rate": 8.460582270689716e-06,
      "loss": 0.0486,
      "step": 6419
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3581197002944131,
      "learning_rate": 8.454772443719339e-06,
      "loss": 0.0609,
      "step": 6420
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2837691340107946,
      "learning_rate": 8.44896407752661e-06,
      "loss": 0.0449,
      "step": 6421
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20723475045468773,
      "learning_rate": 8.443157172846448e-06,
      "loss": 0.03,
      "step": 6422
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.24840566263190597,
      "learning_rate": 8.437351730413586e-06,
      "loss": 0.0408,
      "step": 6423
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2877292817592264,
      "learning_rate": 8.43154775096256e-06,
      "loss": 0.0443,
      "step": 6424
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.235031036295703,
      "learning_rate": 8.425745235227729e-06,
      "loss": 0.0282,
      "step": 6425
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.240622555732698,
      "learning_rate": 8.419944183943266e-06,
      "loss": 0.049,
      "step": 6426
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.220243025031694,
      "learning_rate": 8.41414459784316e-06,
      "loss": 0.0352,
      "step": 6427
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3090907814269771,
      "learning_rate": 8.408346477661218e-06,
      "loss": 0.0382,
      "step": 6428
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23984505345490492,
      "learning_rate": 8.402549824131042e-06,
      "loss": 0.0229,
      "step": 6429
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.27724245397550806,
      "learning_rate": 8.39675463798607e-06,
      "loss": 0.0503,
      "step": 6430
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20537202755964704,
      "learning_rate": 8.390960919959543e-06,
      "loss": 0.034,
      "step": 6431
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3095653425232888,
      "learning_rate": 8.385168670784532e-06,
      "loss": 0.0478,
      "step": 6432
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.32619957035373787,
      "learning_rate": 8.379377891193894e-06,
      "loss": 0.0387,
      "step": 6433
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.28788377813222676,
      "learning_rate": 8.373588581920325e-06,
      "loss": 0.0487,
      "step": 6434
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23606485883055905,
      "learning_rate": 8.367800743696322e-06,
      "loss": 0.0368,
      "step": 6435
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.24119394837137248,
      "learning_rate": 8.362014377254213e-06,
      "loss": 0.0315,
      "step": 6436
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3046419158484017,
      "learning_rate": 8.356229483326108e-06,
      "loss": 0.0517,
      "step": 6437
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.25592941315652334,
      "learning_rate": 8.35044606264396e-06,
      "loss": 0.0396,
      "step": 6438
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4289129037809963,
      "learning_rate": 8.344664115939523e-06,
      "loss": 0.0621,
      "step": 6439
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2876814896525271,
      "learning_rate": 8.338883643944375e-06,
      "loss": 0.0401,
      "step": 6440
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2288342154237088,
      "learning_rate": 8.333104647389876e-06,
      "loss": 0.0265,
      "step": 6441
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19262957359629862,
      "learning_rate": 8.327327127007247e-06,
      "loss": 0.0323,
      "step": 6442
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2714254191760533,
      "learning_rate": 8.321551083527495e-06,
      "loss": 0.029,
      "step": 6443
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.28344231074665666,
      "learning_rate": 8.315776517681428e-06,
      "loss": 0.0499,
      "step": 6444
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22648521453449175,
      "learning_rate": 8.31000343019969e-06,
      "loss": 0.0366,
      "step": 6445
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.28209833842440313,
      "learning_rate": 8.304231821812733e-06,
      "loss": 0.043,
      "step": 6446
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3154077788028386,
      "learning_rate": 8.298461693250821e-06,
      "loss": 0.0513,
      "step": 6447
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2245812191108761,
      "learning_rate": 8.292693045244016e-06,
      "loss": 0.0337,
      "step": 6448
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2644923971251656,
      "learning_rate": 8.286925878522212e-06,
      "loss": 0.0348,
      "step": 6449
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5011667110090846,
      "learning_rate": 8.281160193815108e-06,
      "loss": 0.0692,
      "step": 6450
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43526896347352445,
      "learning_rate": 8.275395991852224e-06,
      "loss": 0.0622,
      "step": 6451
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35474573741177284,
      "learning_rate": 8.269633273362872e-06,
      "loss": 0.0588,
      "step": 6452
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.24254991732867046,
      "learning_rate": 8.26387203907619e-06,
      "loss": 0.0247,
      "step": 6453
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2382054130066473,
      "learning_rate": 8.258112289721134e-06,
      "loss": 0.0364,
      "step": 6454
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.25477298109471,
      "learning_rate": 8.25235402602647e-06,
      "loss": 0.04,
      "step": 6455
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2646705224088017,
      "learning_rate": 8.246597248720756e-06,
      "loss": 0.0485,
      "step": 6456
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23004118002617988,
      "learning_rate": 8.240841958532385e-06,
      "loss": 0.0235,
      "step": 6457
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1825263159375156,
      "learning_rate": 8.23508815618955e-06,
      "loss": 0.0314,
      "step": 6458
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.290648901376989,
      "learning_rate": 8.229335842420265e-06,
      "loss": 0.0647,
      "step": 6459
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3316200739735159,
      "learning_rate": 8.22358501795235e-06,
      "loss": 0.0365,
      "step": 6460
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2714120233302748,
      "learning_rate": 8.217835683513434e-06,
      "loss": 0.0318,
      "step": 6461
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.24189913409970287,
      "learning_rate": 8.212087839830968e-06,
      "loss": 0.028,
      "step": 6462
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22447675853390506,
      "learning_rate": 8.206341487632194e-06,
      "loss": 0.0385,
      "step": 6463
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22658546923256995,
      "learning_rate": 8.200596627644187e-06,
      "loss": 0.0437,
      "step": 6464
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.26287797157704945,
      "learning_rate": 8.19485326059382e-06,
      "loss": 0.0554,
      "step": 6465
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.25696982313523764,
      "learning_rate": 8.189111387207782e-06,
      "loss": 0.0363,
      "step": 6466
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23969556505194853,
      "learning_rate": 8.183371008212582e-06,
      "loss": 0.0482,
      "step": 6467
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.27502408897492436,
      "learning_rate": 8.177632124334513e-06,
      "loss": 0.0642,
      "step": 6468
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2700727662075413,
      "learning_rate": 8.171894736299706e-06,
      "loss": 0.0535,
      "step": 6469
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.24773505277850721,
      "learning_rate": 8.16615884483409e-06,
      "loss": 0.0307,
      "step": 6470
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2185423154627355,
      "learning_rate": 8.16042445066342e-06,
      "loss": 0.0308,
      "step": 6471
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.25625908184240126,
      "learning_rate": 8.154691554513228e-06,
      "loss": 0.039,
      "step": 6472
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2664218478516857,
      "learning_rate": 8.148960157108889e-06,
      "loss": 0.0339,
      "step": 6473
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3309385833146156,
      "learning_rate": 8.143230259175574e-06,
      "loss": 0.0517,
      "step": 6474
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20773226436706252,
      "learning_rate": 8.137501861438278e-06,
      "loss": 0.0292,
      "step": 6475
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2076986280531425,
      "learning_rate": 8.13177496462177e-06,
      "loss": 0.0323,
      "step": 6476
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3011683289434168,
      "learning_rate": 8.126049569450678e-06,
      "loss": 0.0448,
      "step": 6477
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23462786544153602,
      "learning_rate": 8.120325676649416e-06,
      "loss": 0.043,
      "step": 6478
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30795756083096393,
      "learning_rate": 8.114603286942196e-06,
      "loss": 0.0381,
      "step": 6479
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.26037141247776246,
      "learning_rate": 8.108882401053055e-06,
      "loss": 0.025,
      "step": 6480
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35760622533792163,
      "learning_rate": 8.10316301970584e-06,
      "loss": 0.0793,
      "step": 6481
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.21957358227045928,
      "learning_rate": 8.09744514362421e-06,
      "loss": 0.0321,
      "step": 6482
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20876935789158357,
      "learning_rate": 8.091728773531615e-06,
      "loss": 0.0385,
      "step": 6483
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2484817747334928,
      "learning_rate": 8.086013910151334e-06,
      "loss": 0.0355,
      "step": 6484
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19710067424402405,
      "learning_rate": 8.08030055420645e-06,
      "loss": 0.0301,
      "step": 6485
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19765443537184443,
      "learning_rate": 8.07458870641986e-06,
      "loss": 0.0349,
      "step": 6486
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.33794631940008796,
      "learning_rate": 8.06887836751425e-06,
      "loss": 0.0492,
      "step": 6487
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.21348078759314743,
      "learning_rate": 8.063169538212139e-06,
      "loss": 0.0457,
      "step": 6488
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2353299712728007,
      "learning_rate": 8.057462219235842e-06,
      "loss": 0.0392,
      "step": 6489
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2131401751728623,
      "learning_rate": 8.051756411307494e-06,
      "loss": 0.0278,
      "step": 6490
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18141132155433656,
      "learning_rate": 8.04605211514902e-06,
      "loss": 0.0315,
      "step": 6491
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2265805944254647,
      "learning_rate": 8.040349331482167e-06,
      "loss": 0.0386,
      "step": 6492
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2466687267221202,
      "learning_rate": 8.034648061028492e-06,
      "loss": 0.0563,
      "step": 6493
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.21133378562467695,
      "learning_rate": 8.028948304509356e-06,
      "loss": 0.0242,
      "step": 6494
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.21124663482897502,
      "learning_rate": 8.023250062645931e-06,
      "loss": 0.0334,
      "step": 6495
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41337780752921405,
      "learning_rate": 8.017553336159192e-06,
      "loss": 0.0576,
      "step": 6496
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2783136333534162,
      "learning_rate": 8.011858125769938e-06,
      "loss": 0.0588,
      "step": 6497
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19622494233881813,
      "learning_rate": 8.006164432198747e-06,
      "loss": 0.0257,
      "step": 6498
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20460929869147498,
      "learning_rate": 8.00047225616603e-06,
      "loss": 0.023,
      "step": 6499
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2801765779608232,
      "learning_rate": 7.994781598391995e-06,
      "loss": 0.0469,
      "step": 6500
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2924463319041974,
      "learning_rate": 7.989092459596676e-06,
      "loss": 0.0551,
      "step": 6501
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3092880406169501,
      "learning_rate": 7.983404840499882e-06,
      "loss": 0.0536,
      "step": 6502
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.21091489317515238,
      "learning_rate": 7.977718741821253e-06,
      "loss": 0.0339,
      "step": 6503
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2772214646802612,
      "learning_rate": 7.972034164280231e-06,
      "loss": 0.0504,
      "step": 6504
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.26462391532095425,
      "learning_rate": 7.96635110859607e-06,
      "loss": 0.055,
      "step": 6505
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20219168453809466,
      "learning_rate": 7.96066957548783e-06,
      "loss": 0.026,
      "step": 6506
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23814090910905186,
      "learning_rate": 7.954989565674365e-06,
      "loss": 0.0424,
      "step": 6507
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.24045474327049027,
      "learning_rate": 7.949311079874352e-06,
      "loss": 0.0342,
      "step": 6508
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.25521322005647196,
      "learning_rate": 7.943634118806272e-06,
      "loss": 0.0268,
      "step": 6509
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2618900990742312,
      "learning_rate": 7.937958683188407e-06,
      "loss": 0.0493,
      "step": 6510
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.22104491715613434,
      "learning_rate": 7.932284773738856e-06,
      "loss": 0.0371,
      "step": 6511
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.26975350269574855,
      "learning_rate": 7.926612391175516e-06,
      "loss": 0.0415,
      "step": 6512
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20594752268504055,
      "learning_rate": 7.9209415362161e-06,
      "loss": 0.0288,
      "step": 6513
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34855123040864916,
      "learning_rate": 7.915272209578112e-06,
      "loss": 0.0562,
      "step": 6514
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2424700304197419,
      "learning_rate": 7.909604411978872e-06,
      "loss": 0.0329,
      "step": 6515
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.22326849534622184,
      "learning_rate": 7.903938144135515e-06,
      "loss": 0.029,
      "step": 6516
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25573753775720287,
      "learning_rate": 7.898273406764974e-06,
      "loss": 0.0429,
      "step": 6517
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23229788029204876,
      "learning_rate": 7.892610200583979e-06,
      "loss": 0.0242,
      "step": 6518
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2321852272661782,
      "learning_rate": 7.886948526309082e-06,
      "loss": 0.0268,
      "step": 6519
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2636226466812287,
      "learning_rate": 7.881288384656634e-06,
      "loss": 0.0272,
      "step": 6520
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2777781241467754,
      "learning_rate": 7.875629776342802e-06,
      "loss": 0.0346,
      "step": 6521
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3214410115286993,
      "learning_rate": 7.869972702083532e-06,
      "loss": 0.043,
      "step": 6522
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3582476264824408,
      "learning_rate": 7.864317162594608e-06,
      "loss": 0.0468,
      "step": 6523
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.22799082755682637,
      "learning_rate": 7.8586631585916e-06,
      "loss": 0.0357,
      "step": 6524
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2627168100629604,
      "learning_rate": 7.853010690789897e-06,
      "loss": 0.0346,
      "step": 6525
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23385006611435474,
      "learning_rate": 7.847359759904675e-06,
      "loss": 0.0371,
      "step": 6526
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25229807460993714,
      "learning_rate": 7.841710366650932e-06,
      "loss": 0.0318,
      "step": 6527
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25261015573592,
      "learning_rate": 7.836062511743468e-06,
      "loss": 0.0373,
      "step": 6528
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3223993197464688,
      "learning_rate": 7.830416195896882e-06,
      "loss": 0.068,
      "step": 6529
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.29124753787264757,
      "learning_rate": 7.824771419825588e-06,
      "loss": 0.0524,
      "step": 6530
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36407316707979276,
      "learning_rate": 7.819128184243796e-06,
      "loss": 0.0478,
      "step": 6531
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.26657014878312896,
      "learning_rate": 7.813486489865534e-06,
      "loss": 0.0479,
      "step": 6532
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2749290385590782,
      "learning_rate": 7.807846337404611e-06,
      "loss": 0.0278,
      "step": 6533
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20985298930298352,
      "learning_rate": 7.802207727574665e-06,
      "loss": 0.0321,
      "step": 6534
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2291186762903262,
      "learning_rate": 7.79657066108913e-06,
      "loss": 0.0359,
      "step": 6535
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24471351546134185,
      "learning_rate": 7.790935138661246e-06,
      "loss": 0.05,
      "step": 6536
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23810186043396592,
      "learning_rate": 7.785301161004049e-06,
      "loss": 0.0358,
      "step": 6537
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2642028689447798,
      "learning_rate": 7.779668728830389e-06,
      "loss": 0.0371,
      "step": 6538
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23555028606114825,
      "learning_rate": 7.774037842852921e-06,
      "loss": 0.0448,
      "step": 6539
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2626707214417275,
      "learning_rate": 7.768408503784108e-06,
      "loss": 0.0266,
      "step": 6540
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.32157064696317006,
      "learning_rate": 7.762780712336196e-06,
      "loss": 0.0463,
      "step": 6541
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20668172121173925,
      "learning_rate": 7.757154469221257e-06,
      "loss": 0.0392,
      "step": 6542
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23665624492045323,
      "learning_rate": 7.751529775151161e-06,
      "loss": 0.0376,
      "step": 6543
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.28505605740282963,
      "learning_rate": 7.745906630837586e-06,
      "loss": 0.0417,
      "step": 6544
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23061561754054719,
      "learning_rate": 7.740285036991987e-06,
      "loss": 0.0334,
      "step": 6545
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2701965634663251,
      "learning_rate": 7.734664994325672e-06,
      "loss": 0.0418,
      "step": 6546
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2672412728971369,
      "learning_rate": 7.729046503549712e-06,
      "loss": 0.0442,
      "step": 6547
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34885610914596044,
      "learning_rate": 7.723429565375006e-06,
      "loss": 0.0618,
      "step": 6548
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24358799759371624,
      "learning_rate": 7.71781418051223e-06,
      "loss": 0.0245,
      "step": 6549
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3072416283659649,
      "learning_rate": 7.71220034967189e-06,
      "loss": 0.0533,
      "step": 6550
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3403584102001735,
      "learning_rate": 7.706588073564278e-06,
      "loss": 0.056,
      "step": 6551
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3367970138899929,
      "learning_rate": 7.700977352899506e-06,
      "loss": 0.056,
      "step": 6552
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24168692663512917,
      "learning_rate": 7.695368188387466e-06,
      "loss": 0.032,
      "step": 6553
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25406383612587574,
      "learning_rate": 7.68976058073787e-06,
      "loss": 0.0376,
      "step": 6554
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2043610889471732,
      "learning_rate": 7.684154530660235e-06,
      "loss": 0.0361,
      "step": 6555
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.31421248190236667,
      "learning_rate": 7.678550038863877e-06,
      "loss": 0.0444,
      "step": 6556
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2467697713012407,
      "learning_rate": 7.672947106057903e-06,
      "loss": 0.0455,
      "step": 6557
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20496925007078223,
      "learning_rate": 7.667345732951233e-06,
      "loss": 0.0344,
      "step": 6558
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2825699930916705,
      "learning_rate": 7.661745920252594e-06,
      "loss": 0.0407,
      "step": 6559
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24181142957326718,
      "learning_rate": 7.656147668670519e-06,
      "loss": 0.0411,
      "step": 6560
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20664482220603836,
      "learning_rate": 7.65055097891332e-06,
      "loss": 0.0226,
      "step": 6561
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24109204253990774,
      "learning_rate": 7.644955851689129e-06,
      "loss": 0.0497,
      "step": 6562
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23486936406686915,
      "learning_rate": 7.639362287705894e-06,
      "loss": 0.0349,
      "step": 6563
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.17092372162293284,
      "learning_rate": 7.63377028767133e-06,
      "loss": 0.0288,
      "step": 6564
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20596855858440993,
      "learning_rate": 7.628179852292983e-06,
      "loss": 0.0235,
      "step": 6565
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3337479920451879,
      "learning_rate": 7.622590982278189e-06,
      "loss": 0.0583,
      "step": 6566
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2431587540380298,
      "learning_rate": 7.617003678334096e-06,
      "loss": 0.0489,
      "step": 6567
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19104538200146343,
      "learning_rate": 7.611417941167634e-06,
      "loss": 0.0399,
      "step": 6568
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2554741645962459,
      "learning_rate": 7.605833771485549e-06,
      "loss": 0.0394,
      "step": 6569
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1954194443691002,
      "learning_rate": 7.600251169994392e-06,
      "loss": 0.0261,
      "step": 6570
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2536431343096896,
      "learning_rate": 7.594670137400517e-06,
      "loss": 0.0514,
      "step": 6571
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37128636230257456,
      "learning_rate": 7.589090674410056e-06,
      "loss": 0.0494,
      "step": 6572
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2536575860362757,
      "learning_rate": 7.583512781728966e-06,
      "loss": 0.0488,
      "step": 6573
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.294649895567449,
      "learning_rate": 7.577936460063e-06,
      "loss": 0.0573,
      "step": 6574
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23970868970666664,
      "learning_rate": 7.572361710117717e-06,
      "loss": 0.0283,
      "step": 6575
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.15715059376108945,
      "learning_rate": 7.566788532598457e-06,
      "loss": 0.0258,
      "step": 6576
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4482228892042876,
      "learning_rate": 7.561216928210382e-06,
      "loss": 0.0725,
      "step": 6577
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20514763153753457,
      "learning_rate": 7.555646897658448e-06,
      "loss": 0.0278,
      "step": 6578
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24031197449883396,
      "learning_rate": 7.55007844164741e-06,
      "loss": 0.0548,
      "step": 6579
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2700967523067453,
      "learning_rate": 7.544511560881829e-06,
      "loss": 0.0566,
      "step": 6580
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3832098786267362,
      "learning_rate": 7.53894625606606e-06,
      "loss": 0.0515,
      "step": 6581
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23901507461962287,
      "learning_rate": 7.533382527904263e-06,
      "loss": 0.0398,
      "step": 6582
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.28280219275306995,
      "learning_rate": 7.527820377100403e-06,
      "loss": 0.0451,
      "step": 6583
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.266661724526584,
      "learning_rate": 7.5222598043582274e-06,
      "loss": 0.0417,
      "step": 6584
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2635961494494878,
      "learning_rate": 7.516700810381301e-06,
      "loss": 0.0355,
      "step": 6585
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23455403959952692,
      "learning_rate": 7.511143395872986e-06,
      "loss": 0.0386,
      "step": 6586
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2830660521170048,
      "learning_rate": 7.5055875615364495e-06,
      "loss": 0.033,
      "step": 6587
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3227446261873834,
      "learning_rate": 7.500033308074639e-06,
      "loss": 0.0327,
      "step": 6588
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2744162481363485,
      "learning_rate": 7.49448063619032e-06,
      "loss": 0.031,
      "step": 6589
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2482165334608095,
      "learning_rate": 7.488929546586053e-06,
      "loss": 0.0459,
      "step": 6590
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.22046721349082088,
      "learning_rate": 7.483380039964205e-06,
      "loss": 0.0342,
      "step": 6591
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.27028440208153975,
      "learning_rate": 7.477832117026924e-06,
      "loss": 0.0294,
      "step": 6592
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2463165062203717,
      "learning_rate": 7.4722857784761734e-06,
      "loss": 0.0309,
      "step": 6593
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.27237289811120324,
      "learning_rate": 7.466741025013715e-06,
      "loss": 0.0412,
      "step": 6594
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.18011673163854972,
      "learning_rate": 7.461197857341114e-06,
      "loss": 0.0258,
      "step": 6595
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2445382656516951,
      "learning_rate": 7.455656276159713e-06,
      "loss": 0.0374,
      "step": 6596
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21940658158610174,
      "learning_rate": 7.450116282170667e-06,
      "loss": 0.0287,
      "step": 6597
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3439944871693606,
      "learning_rate": 7.444577876074956e-06,
      "loss": 0.0411,
      "step": 6598
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2766455337101841,
      "learning_rate": 7.4390410585733176e-06,
      "loss": 0.0364,
      "step": 6599
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.18934257467358104,
      "learning_rate": 7.4335058303663056e-06,
      "loss": 0.0334,
      "step": 6600
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2542107003924037,
      "learning_rate": 7.42797219215428e-06,
      "loss": 0.0317,
      "step": 6601
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2762423796271572,
      "learning_rate": 7.422440144637395e-06,
      "loss": 0.067,
      "step": 6602
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2952446087499235,
      "learning_rate": 7.41690968851559e-06,
      "loss": 0.0456,
      "step": 6603
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.27888347468628955,
      "learning_rate": 7.411380824488621e-06,
      "loss": 0.0391,
      "step": 6604
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2180002340186464,
      "learning_rate": 7.405853553256035e-06,
      "loss": 0.0358,
      "step": 6605
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2389220158431419,
      "learning_rate": 7.400327875517188e-06,
      "loss": 0.0349,
      "step": 6606
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.26144262237509924,
      "learning_rate": 7.394803791971208e-06,
      "loss": 0.0417,
      "step": 6607
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3388880522622967,
      "learning_rate": 7.389281303317046e-06,
      "loss": 0.053,
      "step": 6608
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.24191057616311354,
      "learning_rate": 7.383760410253446e-06,
      "loss": 0.0387,
      "step": 6609
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19632820194666395,
      "learning_rate": 7.37824111347895e-06,
      "loss": 0.0405,
      "step": 6610
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.25926263052650766,
      "learning_rate": 7.372723413691884e-06,
      "loss": 0.0474,
      "step": 6611
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21710414418950477,
      "learning_rate": 7.367207311590392e-06,
      "loss": 0.0343,
      "step": 6612
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22550321275945098,
      "learning_rate": 7.361692807872405e-06,
      "loss": 0.0485,
      "step": 6613
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22521029817028132,
      "learning_rate": 7.356179903235654e-06,
      "loss": 0.039,
      "step": 6614
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.23056507522218203,
      "learning_rate": 7.350668598377668e-06,
      "loss": 0.0351,
      "step": 6615
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2473592312824255,
      "learning_rate": 7.345158893995774e-06,
      "loss": 0.0327,
      "step": 6616
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3047008389218534,
      "learning_rate": 7.339650790787103e-06,
      "loss": 0.05,
      "step": 6617
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2251726137418741,
      "learning_rate": 7.33414428944856e-06,
      "loss": 0.0303,
      "step": 6618
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3037834203646682,
      "learning_rate": 7.328639390676873e-06,
      "loss": 0.068,
      "step": 6619
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.18046847917280476,
      "learning_rate": 7.3231360951685574e-06,
      "loss": 0.0311,
      "step": 6620
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2331037936012605,
      "learning_rate": 7.317634403619931e-06,
      "loss": 0.0285,
      "step": 6621
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22343730292945121,
      "learning_rate": 7.312134316727093e-06,
      "loss": 0.0301,
      "step": 6622
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.23224822537675274,
      "learning_rate": 7.306635835185956e-06,
      "loss": 0.0424,
      "step": 6623
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22277578692106198,
      "learning_rate": 7.301138959692225e-06,
      "loss": 0.0339,
      "step": 6624
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21664888468698934,
      "learning_rate": 7.295643690941397e-06,
      "loss": 0.0271,
      "step": 6625
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19762131755469559,
      "learning_rate": 7.290150029628777e-06,
      "loss": 0.0389,
      "step": 6626
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3171774214284935,
      "learning_rate": 7.28465797644945e-06,
      "loss": 0.0418,
      "step": 6627
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2399029112053458,
      "learning_rate": 7.2791675320983076e-06,
      "loss": 0.0572,
      "step": 6628
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2441456298315135,
      "learning_rate": 7.273678697270039e-06,
      "loss": 0.0285,
      "step": 6629
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3210788491311601,
      "learning_rate": 7.268191472659136e-06,
      "loss": 0.0447,
      "step": 6630
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19610518767903182,
      "learning_rate": 7.262705858959855e-06,
      "loss": 0.0304,
      "step": 6631
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2463955920326748,
      "learning_rate": 7.257221856866295e-06,
      "loss": 0.0386,
      "step": 6632
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22082828247544065,
      "learning_rate": 7.251739467072323e-06,
      "loss": 0.0393,
      "step": 6633
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19626924784842306,
      "learning_rate": 7.246258690271599e-06,
      "loss": 0.028,
      "step": 6634
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.18274363051937373,
      "learning_rate": 7.240779527157589e-06,
      "loss": 0.027,
      "step": 6635
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2504980074237907,
      "learning_rate": 7.235301978423555e-06,
      "loss": 0.0444,
      "step": 6636
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2484751181042505,
      "learning_rate": 7.229826044762558e-06,
      "loss": 0.0434,
      "step": 6637
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.24730890988975338,
      "learning_rate": 7.224351726867433e-06,
      "loss": 0.0334,
      "step": 6638
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2100350448766171,
      "learning_rate": 7.218879025430838e-06,
      "loss": 0.0181,
      "step": 6639
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2383164082838716,
      "learning_rate": 7.213407941145214e-06,
      "loss": 0.0284,
      "step": 6640
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.20707655357018465,
      "learning_rate": 7.207938474702802e-06,
      "loss": 0.0248,
      "step": 6641
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21129910945014121,
      "learning_rate": 7.202470626795626e-06,
      "loss": 0.0339,
      "step": 6642
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.246992460809062,
      "learning_rate": 7.197004398115515e-06,
      "loss": 0.0343,
      "step": 6643
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2819418741773533,
      "learning_rate": 7.191539789354096e-06,
      "loss": 0.0424,
      "step": 6644
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2255049556020603,
      "learning_rate": 7.18607680120279e-06,
      "loss": 0.0244,
      "step": 6645
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22799028834774265,
      "learning_rate": 7.180615434352802e-06,
      "loss": 0.0291,
      "step": 6646
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.18657914173720364,
      "learning_rate": 7.175155689495142e-06,
      "loss": 0.0321,
      "step": 6647
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.25142358237693113,
      "learning_rate": 7.1696975673206125e-06,
      "loss": 0.039,
      "step": 6648
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2955624526716143,
      "learning_rate": 7.164241068519815e-06,
      "loss": 0.0397,
      "step": 6649
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22012736746290956,
      "learning_rate": 7.158786193783138e-06,
      "loss": 0.0309,
      "step": 6650
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2442214221771545,
      "learning_rate": 7.153332943800768e-06,
      "loss": 0.0343,
      "step": 6651
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22565001861522319,
      "learning_rate": 7.147881319262695e-06,
      "loss": 0.0509,
      "step": 6652
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.18746004076181033,
      "learning_rate": 7.142431320858676e-06,
      "loss": 0.0349,
      "step": 6653
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.23662973499145099,
      "learning_rate": 7.136982949278293e-06,
      "loss": 0.028,
      "step": 6654
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2341850623753995,
      "learning_rate": 7.131536205210905e-06,
      "loss": 0.032,
      "step": 6655
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.24560601478955826,
      "learning_rate": 7.126091089345679e-06,
      "loss": 0.0427,
      "step": 6656
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32064811058112347,
      "learning_rate": 7.120647602371551e-06,
      "loss": 0.0453,
      "step": 6657
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22087303305604478,
      "learning_rate": 7.115205744977276e-06,
      "loss": 0.034,
      "step": 6658
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.26221100930263325,
      "learning_rate": 7.10976551785139e-06,
      "loss": 0.0303,
      "step": 6659
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.25698065216249494,
      "learning_rate": 7.104326921682236e-06,
      "loss": 0.0288,
      "step": 6660
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3253034921511151,
      "learning_rate": 7.0988899571579266e-06,
      "loss": 0.0393,
      "step": 6661
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.20542203132737769,
      "learning_rate": 7.093454624966387e-06,
      "loss": 0.0414,
      "step": 6662
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33082687506549724,
      "learning_rate": 7.088020925795334e-06,
      "loss": 0.0388,
      "step": 6663
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21913092518655058,
      "learning_rate": 7.082588860332271e-06,
      "loss": 0.0352,
      "step": 6664
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30516152310442196,
      "learning_rate": 7.077158429264508e-06,
      "loss": 0.0392,
      "step": 6665
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.223187457073168,
      "learning_rate": 7.071729633279118e-06,
      "loss": 0.0282,
      "step": 6666
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2642827340136583,
      "learning_rate": 7.066302473063007e-06,
      "loss": 0.0319,
      "step": 6667
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.23745255121783146,
      "learning_rate": 7.060876949302855e-06,
      "loss": 0.0287,
      "step": 6668
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22469789170599044,
      "learning_rate": 7.055453062685122e-06,
      "loss": 0.0331,
      "step": 6669
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.275362189157255,
      "learning_rate": 7.050030813896078e-06,
      "loss": 0.0329,
      "step": 6670
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.286495889744892,
      "learning_rate": 7.044610203621787e-06,
      "loss": 0.0367,
      "step": 6671
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2955555077466014,
      "learning_rate": 7.0391912325481e-06,
      "loss": 0.045,
      "step": 6672
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3717115780433946,
      "learning_rate": 7.033773901360652e-06,
      "loss": 0.0589,
      "step": 6673
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2947748209158103,
      "learning_rate": 7.028358210744881e-06,
      "loss": 0.0533,
      "step": 6674
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21997114344477642,
      "learning_rate": 7.022944161386023e-06,
      "loss": 0.03,
      "step": 6675
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2279474988579471,
      "learning_rate": 7.017531753969098e-06,
      "loss": 0.0332,
      "step": 6676
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2267305310091527,
      "learning_rate": 7.01212098917891e-06,
      "loss": 0.0322,
      "step": 6677
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.34418955789521966,
      "learning_rate": 7.006711867700069e-06,
      "loss": 0.0489,
      "step": 6678
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21267668447198487,
      "learning_rate": 7.0013043902169745e-06,
      "loss": 0.0366,
      "step": 6679
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2698602741353036,
      "learning_rate": 6.995898557413823e-06,
      "loss": 0.0341,
      "step": 6680
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22272471998062027,
      "learning_rate": 6.990494369974579e-06,
      "loss": 0.0356,
      "step": 6681
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2769559378541649,
      "learning_rate": 6.985091828583024e-06,
      "loss": 0.0474,
      "step": 6682
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2394360948514919,
      "learning_rate": 6.979690933922725e-06,
      "loss": 0.0461,
      "step": 6683
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2510101355276847,
      "learning_rate": 6.974291686677035e-06,
      "loss": 0.0471,
      "step": 6684
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30603822226865846,
      "learning_rate": 6.968894087529103e-06,
      "loss": 0.0511,
      "step": 6685
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22525735349538595,
      "learning_rate": 6.96349813716187e-06,
      "loss": 0.0283,
      "step": 6686
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.27284032461279195,
      "learning_rate": 6.958103836258074e-06,
      "loss": 0.0339,
      "step": 6687
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.17913651993611,
      "learning_rate": 6.952711185500223e-06,
      "loss": 0.0324,
      "step": 6688
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21817188846550117,
      "learning_rate": 6.947320185570634e-06,
      "loss": 0.0245,
      "step": 6689
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.25154878270914904,
      "learning_rate": 6.941930837151416e-06,
      "loss": 0.0282,
      "step": 6690
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22785372117810418,
      "learning_rate": 6.93654314092447e-06,
      "loss": 0.0356,
      "step": 6691
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1968834545954086,
      "learning_rate": 6.931157097571468e-06,
      "loss": 0.033,
      "step": 6692
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21011434761279868,
      "learning_rate": 6.925772707773894e-06,
      "loss": 0.0316,
      "step": 6693
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.27499520882854395,
      "learning_rate": 6.920389972213017e-06,
      "loss": 0.0492,
      "step": 6694
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33057321290709657,
      "learning_rate": 6.915008891569906e-06,
      "loss": 0.0487,
      "step": 6695
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.17008143639484036,
      "learning_rate": 6.909629466525389e-06,
      "loss": 0.0315,
      "step": 6696
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24153922594375032,
      "learning_rate": 6.90425169776012e-06,
      "loss": 0.033,
      "step": 6697
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3175985805966503,
      "learning_rate": 6.898875585954527e-06,
      "loss": 0.0411,
      "step": 6698
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.25412963430642016,
      "learning_rate": 6.893501131788838e-06,
      "loss": 0.0466,
      "step": 6699
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4069227736709644,
      "learning_rate": 6.888128335943041e-06,
      "loss": 0.0697,
      "step": 6700
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33502762689473636,
      "learning_rate": 6.882757199096961e-06,
      "loss": 0.0566,
      "step": 6701
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3009942526696348,
      "learning_rate": 6.877387721930182e-06,
      "loss": 0.0299,
      "step": 6702
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3082821999606284,
      "learning_rate": 6.8720199051220914e-06,
      "loss": 0.0327,
      "step": 6703
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.32758929027304423,
      "learning_rate": 6.866653749351846e-06,
      "loss": 0.0611,
      "step": 6704
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2775734635146077,
      "learning_rate": 6.861289255298416e-06,
      "loss": 0.0418,
      "step": 6705
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23643962342158065,
      "learning_rate": 6.855926423640549e-06,
      "loss": 0.0424,
      "step": 6706
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2334515180042888,
      "learning_rate": 6.850565255056796e-06,
      "loss": 0.0482,
      "step": 6707
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.19592635011965912,
      "learning_rate": 6.84520575022547e-06,
      "loss": 0.0333,
      "step": 6708
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24017503827474826,
      "learning_rate": 6.839847909824699e-06,
      "loss": 0.057,
      "step": 6709
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21520028441469727,
      "learning_rate": 6.8344917345323935e-06,
      "loss": 0.0346,
      "step": 6710
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20818196201660227,
      "learning_rate": 6.829137225026259e-06,
      "loss": 0.032,
      "step": 6711
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2753011544625325,
      "learning_rate": 6.823784381983764e-06,
      "loss": 0.0521,
      "step": 6712
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23919578838216948,
      "learning_rate": 6.818433206082198e-06,
      "loss": 0.0505,
      "step": 6713
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.26167725476075,
      "learning_rate": 6.8130836979986236e-06,
      "loss": 0.0461,
      "step": 6714
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.232901966902673,
      "learning_rate": 6.807735858409903e-06,
      "loss": 0.0417,
      "step": 6715
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2567725302747838,
      "learning_rate": 6.802389687992666e-06,
      "loss": 0.0353,
      "step": 6716
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.29739894559114705,
      "learning_rate": 6.797045187423345e-06,
      "loss": 0.0342,
      "step": 6717
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2604807425188423,
      "learning_rate": 6.791702357378185e-06,
      "loss": 0.0457,
      "step": 6718
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20898025722546792,
      "learning_rate": 6.786361198533169e-06,
      "loss": 0.0407,
      "step": 6719
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3361953588497254,
      "learning_rate": 6.781021711564107e-06,
      "loss": 0.0513,
      "step": 6720
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2077066363393191,
      "learning_rate": 6.775683897146583e-06,
      "loss": 0.0369,
      "step": 6721
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.26035007897059514,
      "learning_rate": 6.770347755955982e-06,
      "loss": 0.036,
      "step": 6722
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2468894251097828,
      "learning_rate": 6.7650132886674505e-06,
      "loss": 0.0468,
      "step": 6723
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.242480331709689,
      "learning_rate": 6.7596804959559494e-06,
      "loss": 0.0248,
      "step": 6724
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20534873546840743,
      "learning_rate": 6.754349378496219e-06,
      "loss": 0.036,
      "step": 6725
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2319250165457605,
      "learning_rate": 6.749019936962791e-06,
      "loss": 0.0481,
      "step": 6726
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23197156896398202,
      "learning_rate": 6.7436921720299675e-06,
      "loss": 0.039,
      "step": 6727
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2435435508990161,
      "learning_rate": 6.7383660843718635e-06,
      "loss": 0.0441,
      "step": 6728
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2929532492033752,
      "learning_rate": 6.733041674662366e-06,
      "loss": 0.0494,
      "step": 6729
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23325549421044114,
      "learning_rate": 6.727718943575161e-06,
      "loss": 0.0354,
      "step": 6730
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18902550445706776,
      "learning_rate": 6.722397891783705e-06,
      "loss": 0.0254,
      "step": 6731
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2602503563386528,
      "learning_rate": 6.717078519961257e-06,
      "loss": 0.027,
      "step": 6732
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.28259134971968575,
      "learning_rate": 6.711760828780858e-06,
      "loss": 0.0279,
      "step": 6733
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18558818959495804,
      "learning_rate": 6.706444818915345e-06,
      "loss": 0.0303,
      "step": 6734
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24799531115340948,
      "learning_rate": 6.701130491037315e-06,
      "loss": 0.047,
      "step": 6735
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.26283223721012183,
      "learning_rate": 6.695817845819188e-06,
      "loss": 0.0449,
      "step": 6736
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35873455532735216,
      "learning_rate": 6.690506883933159e-06,
      "loss": 0.0441,
      "step": 6737
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20649669119756392,
      "learning_rate": 6.68519760605119e-06,
      "loss": 0.028,
      "step": 6738
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2489254029227338,
      "learning_rate": 6.679890012845052e-06,
      "loss": 0.0219,
      "step": 6739
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2496134929287456,
      "learning_rate": 6.674584104986295e-06,
      "loss": 0.0548,
      "step": 6740
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3020056290275606,
      "learning_rate": 6.669279883146262e-06,
      "loss": 0.0451,
      "step": 6741
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37669951055533923,
      "learning_rate": 6.66397734799608e-06,
      "loss": 0.066,
      "step": 6742
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18733059661161258,
      "learning_rate": 6.658676500206649e-06,
      "loss": 0.0221,
      "step": 6743
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2660315991953571,
      "learning_rate": 6.653377340448673e-06,
      "loss": 0.035,
      "step": 6744
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.22917323572685044,
      "learning_rate": 6.648079869392634e-06,
      "loss": 0.0299,
      "step": 6745
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23852382881957337,
      "learning_rate": 6.642784087708814e-06,
      "loss": 0.0337,
      "step": 6746
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.19873770435854676,
      "learning_rate": 6.6374899960672545e-06,
      "loss": 0.0299,
      "step": 6747
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24036038203233998,
      "learning_rate": 6.6321975951378034e-06,
      "loss": 0.039,
      "step": 6748
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20760325818422112,
      "learning_rate": 6.626906885590094e-06,
      "loss": 0.025,
      "step": 6749
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20016482981771522,
      "learning_rate": 6.621617868093544e-06,
      "loss": 0.034,
      "step": 6750
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2665290035117795,
      "learning_rate": 6.616330543317344e-06,
      "loss": 0.0349,
      "step": 6751
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24545157371854967,
      "learning_rate": 6.611044911930477e-06,
      "loss": 0.0471,
      "step": 6752
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2778263728365057,
      "learning_rate": 6.605760974601743e-06,
      "loss": 0.0453,
      "step": 6753
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2038267923518429,
      "learning_rate": 6.6004787319996714e-06,
      "loss": 0.0315,
      "step": 6754
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1910679805906707,
      "learning_rate": 6.59519818479262e-06,
      "loss": 0.0316,
      "step": 6755
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37476717158339,
      "learning_rate": 6.589919333648711e-06,
      "loss": 0.0565,
      "step": 6756
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24757010529067153,
      "learning_rate": 6.5846421792358735e-06,
      "loss": 0.0385,
      "step": 6757
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23392257734099883,
      "learning_rate": 6.579366722221789e-06,
      "loss": 0.053,
      "step": 6758
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3667734733810828,
      "learning_rate": 6.574092963273953e-06,
      "loss": 0.0543,
      "step": 6759
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24177087819697402,
      "learning_rate": 6.568820903059632e-06,
      "loss": 0.0354,
      "step": 6760
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2415961922593796,
      "learning_rate": 6.563550542245891e-06,
      "loss": 0.0379,
      "step": 6761
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.25252289628093993,
      "learning_rate": 6.558281881499556e-06,
      "loss": 0.0516,
      "step": 6762
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2411511922505879,
      "learning_rate": 6.55301492148726e-06,
      "loss": 0.0393,
      "step": 6763
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2495150033992153,
      "learning_rate": 6.547749662875411e-06,
      "loss": 0.0448,
      "step": 6764
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.27672934644885144,
      "learning_rate": 6.542486106330215e-06,
      "loss": 0.0311,
      "step": 6765
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20908389885757378,
      "learning_rate": 6.537224252517633e-06,
      "loss": 0.0296,
      "step": 6766
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2125216357352454,
      "learning_rate": 6.53196410210344e-06,
      "loss": 0.0328,
      "step": 6767
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.26519493742436007,
      "learning_rate": 6.526705655753183e-06,
      "loss": 0.05,
      "step": 6768
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24269239802431805,
      "learning_rate": 6.521448914132193e-06,
      "loss": 0.029,
      "step": 6769
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20822451948565476,
      "learning_rate": 6.516193877905592e-06,
      "loss": 0.0396,
      "step": 6770
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37164807328015104,
      "learning_rate": 6.510940547738278e-06,
      "loss": 0.0538,
      "step": 6771
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.238176815410413,
      "learning_rate": 6.505688924294944e-06,
      "loss": 0.031,
      "step": 6772
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38653026187340817,
      "learning_rate": 6.500439008240047e-06,
      "loss": 0.0547,
      "step": 6773
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2887829124387146,
      "learning_rate": 6.495190800237845e-06,
      "loss": 0.0268,
      "step": 6774
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.25954762252710073,
      "learning_rate": 6.48994430095238e-06,
      "loss": 0.029,
      "step": 6775
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1981215744294018,
      "learning_rate": 6.484699511047474e-06,
      "loss": 0.0288,
      "step": 6776
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21034891521803428,
      "learning_rate": 6.4794564311867254e-06,
      "loss": 0.0257,
      "step": 6777
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23240921242464563,
      "learning_rate": 6.474215062033527e-06,
      "loss": 0.0302,
      "step": 6778
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23992311270520944,
      "learning_rate": 6.468975404251047e-06,
      "loss": 0.0427,
      "step": 6779
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21893798890741215,
      "learning_rate": 6.463737458502255e-06,
      "loss": 0.0299,
      "step": 6780
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23098809515220292,
      "learning_rate": 6.458501225449872e-06,
      "loss": 0.0331,
      "step": 6781
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21917623478765833,
      "learning_rate": 6.453266705756427e-06,
      "loss": 0.0354,
      "step": 6782
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23833755703386872,
      "learning_rate": 6.448033900084228e-06,
      "loss": 0.0295,
      "step": 6783
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20428044634201836,
      "learning_rate": 6.442802809095363e-06,
      "loss": 0.046,
      "step": 6784
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2711294447515628,
      "learning_rate": 6.43757343345171e-06,
      "loss": 0.0364,
      "step": 6785
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.215917970820392,
      "learning_rate": 6.4323457738149034e-06,
      "loss": 0.0269,
      "step": 6786
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3150409275101422,
      "learning_rate": 6.427119830846402e-06,
      "loss": 0.0355,
      "step": 6787
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19288600429409625,
      "learning_rate": 6.421895605207427e-06,
      "loss": 0.0254,
      "step": 6788
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1740954732739422,
      "learning_rate": 6.4166730975589656e-06,
      "loss": 0.0247,
      "step": 6789
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21777521654437285,
      "learning_rate": 6.41145230856181e-06,
      "loss": 0.0265,
      "step": 6790
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2654947073529259,
      "learning_rate": 6.406233238876534e-06,
      "loss": 0.0491,
      "step": 6791
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.316390778905072,
      "learning_rate": 6.401015889163489e-06,
      "loss": 0.0511,
      "step": 6792
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2797450091732419,
      "learning_rate": 6.395800260082796e-06,
      "loss": 0.0324,
      "step": 6793
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.22057211202829743,
      "learning_rate": 6.3905863522943786e-06,
      "loss": 0.0354,
      "step": 6794
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.27117342259111876,
      "learning_rate": 6.385374166457936e-06,
      "loss": 0.051,
      "step": 6795
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2841852456299777,
      "learning_rate": 6.380163703232953e-06,
      "loss": 0.0573,
      "step": 6796
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.24613588971428838,
      "learning_rate": 6.3749549632786765e-06,
      "loss": 0.0433,
      "step": 6797
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23222062670386093,
      "learning_rate": 6.369747947254159e-06,
      "loss": 0.0335,
      "step": 6798
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23216652667635165,
      "learning_rate": 6.364542655818226e-06,
      "loss": 0.0324,
      "step": 6799
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.242540202654977,
      "learning_rate": 6.35933908962949e-06,
      "loss": 0.0406,
      "step": 6800
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20443812824387148,
      "learning_rate": 6.35413724934633e-06,
      "loss": 0.0263,
      "step": 6801
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.24624244799385803,
      "learning_rate": 6.348937135626922e-06,
      "loss": 0.0423,
      "step": 6802
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21915284597009768,
      "learning_rate": 6.343738749129218e-06,
      "loss": 0.0289,
      "step": 6803
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3349522309204238,
      "learning_rate": 6.338542090510951e-06,
      "loss": 0.0497,
      "step": 6804
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.24035537587746944,
      "learning_rate": 6.333347160429637e-06,
      "loss": 0.0286,
      "step": 6805
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.24749989914169568,
      "learning_rate": 6.328153959542573e-06,
      "loss": 0.041,
      "step": 6806
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31219472279178767,
      "learning_rate": 6.322962488506843e-06,
      "loss": 0.0535,
      "step": 6807
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.22861377531875743,
      "learning_rate": 6.3177727479792914e-06,
      "loss": 0.0344,
      "step": 6808
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2128419557987818,
      "learning_rate": 6.312584738616567e-06,
      "loss": 0.041,
      "step": 6809
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.25656225032538693,
      "learning_rate": 6.307398461075091e-06,
      "loss": 0.0296,
      "step": 6810
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21255100353415055,
      "learning_rate": 6.302213916011069e-06,
      "loss": 0.0204,
      "step": 6811
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.26928668052383264,
      "learning_rate": 6.297031104080471e-06,
      "loss": 0.0387,
      "step": 6812
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.24163960193623427,
      "learning_rate": 6.291850025939068e-06,
      "loss": 0.052,
      "step": 6813
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21072960251219236,
      "learning_rate": 6.286670682242404e-06,
      "loss": 0.0196,
      "step": 6814
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29053011339935075,
      "learning_rate": 6.28149307364581e-06,
      "loss": 0.041,
      "step": 6815
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2452281229729745,
      "learning_rate": 6.276317200804376e-06,
      "loss": 0.0301,
      "step": 6816
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3008340566279486,
      "learning_rate": 6.2711430643729955e-06,
      "loss": 0.0473,
      "step": 6817
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3223829344889388,
      "learning_rate": 6.265970665006334e-06,
      "loss": 0.0498,
      "step": 6818
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23962565580366854,
      "learning_rate": 6.26080000335884e-06,
      "loss": 0.0324,
      "step": 6819
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.16584767092333566,
      "learning_rate": 6.255631080084735e-06,
      "loss": 0.0258,
      "step": 6820
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.25293478353072396,
      "learning_rate": 6.250463895838015e-06,
      "loss": 0.0307,
      "step": 6821
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3810558888051284,
      "learning_rate": 6.245298451272486e-06,
      "loss": 0.0626,
      "step": 6822
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.28154471638666695,
      "learning_rate": 6.240134747041711e-06,
      "loss": 0.042,
      "step": 6823
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3325632209476083,
      "learning_rate": 6.234972783799023e-06,
      "loss": 0.0441,
      "step": 6824
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.22231711108091096,
      "learning_rate": 6.229812562197552e-06,
      "loss": 0.0411,
      "step": 6825
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.27645581445262735,
      "learning_rate": 6.224654082890207e-06,
      "loss": 0.0353,
      "step": 6826
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23175364691345224,
      "learning_rate": 6.2194973465296775e-06,
      "loss": 0.0297,
      "step": 6827
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2889149263357233,
      "learning_rate": 6.214342353768412e-06,
      "loss": 0.0455,
      "step": 6828
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.26965546164605725,
      "learning_rate": 6.209189105258661e-06,
      "loss": 0.0543,
      "step": 6829
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31113666931941986,
      "learning_rate": 6.2040376016524506e-06,
      "loss": 0.0655,
      "step": 6830
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2593064085336173,
      "learning_rate": 6.198887843601587e-06,
      "loss": 0.0444,
      "step": 6831
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2496085231091882,
      "learning_rate": 6.193739831757637e-06,
      "loss": 0.0301,
      "step": 6832
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34749893315240676,
      "learning_rate": 6.18859356677197e-06,
      "loss": 0.0312,
      "step": 6833
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.26292284786456743,
      "learning_rate": 6.183449049295722e-06,
      "loss": 0.0463,
      "step": 6834
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21210002211152915,
      "learning_rate": 6.17830627997982e-06,
      "loss": 0.0331,
      "step": 6835
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21704679981329525,
      "learning_rate": 6.1731652594749465e-06,
      "loss": 0.033,
      "step": 6836
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2112713573595891,
      "learning_rate": 6.168025988431585e-06,
      "loss": 0.0355,
      "step": 6837
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21845159442873738,
      "learning_rate": 6.162888467499988e-06,
      "loss": 0.0312,
      "step": 6838
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.26620694533200334,
      "learning_rate": 6.157752697330186e-06,
      "loss": 0.0297,
      "step": 6839
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2821739966202227,
      "learning_rate": 6.152618678571996e-06,
      "loss": 0.0319,
      "step": 6840
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.22109456079017598,
      "learning_rate": 6.147486411875004e-06,
      "loss": 0.0282,
      "step": 6841
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.616477412961729,
      "learning_rate": 6.1423558978885836e-06,
      "loss": 0.1456,
      "step": 6842
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.22913896689479998,
      "learning_rate": 6.1372271372618695e-06,
      "loss": 0.0325,
      "step": 6843
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2623221856683833,
      "learning_rate": 6.1321001306437946e-06,
      "loss": 0.0369,
      "step": 6844
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.24793242269557578,
      "learning_rate": 6.126974878683056e-06,
      "loss": 0.0336,
      "step": 6845
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21941552930290348,
      "learning_rate": 6.121851382028146e-06,
      "loss": 0.0374,
      "step": 6846
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21167935501075336,
      "learning_rate": 6.1167296413273084e-06,
      "loss": 0.0394,
      "step": 6847
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2235205589317875,
      "learning_rate": 6.111609657228581e-06,
      "loss": 0.044,
      "step": 6848
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2676865156290542,
      "learning_rate": 6.106491430379784e-06,
      "loss": 0.0431,
      "step": 6849
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3225944113413735,
      "learning_rate": 6.101374961428512e-06,
      "loss": 0.0325,
      "step": 6850
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21392748257104954,
      "learning_rate": 6.096260251022124e-06,
      "loss": 0.034,
      "step": 6851
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2518758729299559,
      "learning_rate": 6.091147299807769e-06,
      "loss": 0.0389,
      "step": 6852
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4328833594312295,
      "learning_rate": 6.086036108432376e-06,
      "loss": 0.0781,
      "step": 6853
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23521230859553557,
      "learning_rate": 6.080926677542646e-06,
      "loss": 0.0431,
      "step": 6854
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2257676816777061,
      "learning_rate": 6.075819007785047e-06,
      "loss": 0.0319,
      "step": 6855
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21971528766103066,
      "learning_rate": 6.070713099805845e-06,
      "loss": 0.0391,
      "step": 6856
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.288846902089892,
      "learning_rate": 6.065608954251074e-06,
      "loss": 0.0421,
      "step": 6857
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19815137502584007,
      "learning_rate": 6.0605065717665445e-06,
      "loss": 0.0249,
      "step": 6858
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2565631505653599,
      "learning_rate": 6.0554059529978345e-06,
      "loss": 0.0342,
      "step": 6859
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35445553088280773,
      "learning_rate": 6.050307098590311e-06,
      "loss": 0.0559,
      "step": 6860
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19227802501685104,
      "learning_rate": 6.045210009189117e-06,
      "loss": 0.0388,
      "step": 6861
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2040952106564328,
      "learning_rate": 6.040114685439175e-06,
      "loss": 0.0231,
      "step": 6862
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.24997459222905485,
      "learning_rate": 6.035021127985165e-06,
      "loss": 0.0363,
      "step": 6863
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23336590655917344,
      "learning_rate": 6.029929337471565e-06,
      "loss": 0.0215,
      "step": 6864
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.26287510898368344,
      "learning_rate": 6.024839314542621e-06,
      "loss": 0.0547,
      "step": 6865
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19023282561562238,
      "learning_rate": 6.019751059842362e-06,
      "loss": 0.025,
      "step": 6866
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.22636640217787196,
      "learning_rate": 6.014664574014575e-06,
      "loss": 0.029,
      "step": 6867
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3853410784169599,
      "learning_rate": 6.009579857702843e-06,
      "loss": 0.0381,
      "step": 6868
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2615681827205863,
      "learning_rate": 6.004496911550512e-06,
      "loss": 0.037,
      "step": 6869
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23539204071307943,
      "learning_rate": 5.999415736200724e-06,
      "loss": 0.0368,
      "step": 6870
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.24217109470794276,
      "learning_rate": 5.9943363322963646e-06,
      "loss": 0.0376,
      "step": 6871
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3303204611808623,
      "learning_rate": 5.98925870048012e-06,
      "loss": 0.0419,
      "step": 6872
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29619485341829044,
      "learning_rate": 5.984182841394448e-06,
      "loss": 0.0384,
      "step": 6873
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4046545980032232,
      "learning_rate": 5.979108755681575e-06,
      "loss": 0.0628,
      "step": 6874
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23116407731271563,
      "learning_rate": 5.974036443983513e-06,
      "loss": 0.0447,
      "step": 6875
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.32092446507816974,
      "learning_rate": 5.968965906942039e-06,
      "loss": 0.0434,
      "step": 6876
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24333309135740694,
      "learning_rate": 5.963897145198721e-06,
      "loss": 0.0349,
      "step": 6877
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2118458088595964,
      "learning_rate": 5.958830159394875e-06,
      "loss": 0.0367,
      "step": 6878
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21557485854374153,
      "learning_rate": 5.953764950171619e-06,
      "loss": 0.029,
      "step": 6879
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21661123283407868,
      "learning_rate": 5.948701518169835e-06,
      "loss": 0.0505,
      "step": 6880
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20379084796353694,
      "learning_rate": 5.943639864030188e-06,
      "loss": 0.049,
      "step": 6881
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2704629964233051,
      "learning_rate": 5.938579988393099e-06,
      "loss": 0.0367,
      "step": 6882
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22877641615224523,
      "learning_rate": 5.9335218918987815e-06,
      "loss": 0.0415,
      "step": 6883
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1880715718234911,
      "learning_rate": 5.928465575187221e-06,
      "loss": 0.035,
      "step": 6884
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.26952019267966726,
      "learning_rate": 5.923411038898181e-06,
      "loss": 0.0394,
      "step": 6885
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24459876752361198,
      "learning_rate": 5.918358283671182e-06,
      "loss": 0.0607,
      "step": 6886
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.23928127563026055,
      "learning_rate": 5.9133073101455375e-06,
      "loss": 0.046,
      "step": 6887
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21356373862816366,
      "learning_rate": 5.90825811896033e-06,
      "loss": 0.0413,
      "step": 6888
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.36327952722940804,
      "learning_rate": 5.903210710754423e-06,
      "loss": 0.0465,
      "step": 6889
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3333922597338471,
      "learning_rate": 5.89816508616643e-06,
      "loss": 0.0549,
      "step": 6890
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24586662529072634,
      "learning_rate": 5.8931212458347716e-06,
      "loss": 0.0618,
      "step": 6891
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2233143917471241,
      "learning_rate": 5.888079190397628e-06,
      "loss": 0.0312,
      "step": 6892
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22086511422989913,
      "learning_rate": 5.883038920492945e-06,
      "loss": 0.0378,
      "step": 6893
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22993504425919062,
      "learning_rate": 5.878000436758453e-06,
      "loss": 0.0201,
      "step": 6894
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21778264047652549,
      "learning_rate": 5.8729637398316544e-06,
      "loss": 0.0316,
      "step": 6895
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21002404794773494,
      "learning_rate": 5.86792883034983e-06,
      "loss": 0.0217,
      "step": 6896
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2520829982334989,
      "learning_rate": 5.862895708950021e-06,
      "loss": 0.0382,
      "step": 6897
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21394714184742136,
      "learning_rate": 5.857864376269051e-06,
      "loss": 0.0279,
      "step": 6898
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29743431758681066,
      "learning_rate": 5.852834832943524e-06,
      "loss": 0.0396,
      "step": 6899
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21811396229315153,
      "learning_rate": 5.847807079609804e-06,
      "loss": 0.0343,
      "step": 6900
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20649080993174362,
      "learning_rate": 5.842781116904046e-06,
      "loss": 0.0395,
      "step": 6901
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2129150341634506,
      "learning_rate": 5.837756945462154e-06,
      "loss": 0.0295,
      "step": 6902
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.25851569942635844,
      "learning_rate": 5.832734565919826e-06,
      "loss": 0.0325,
      "step": 6903
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2281988018009029,
      "learning_rate": 5.827713978912524e-06,
      "loss": 0.0268,
      "step": 6904
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21300050542104332,
      "learning_rate": 5.8226951850754935e-06,
      "loss": 0.0536,
      "step": 6905
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.16063120244626866,
      "learning_rate": 5.817678185043733e-06,
      "loss": 0.0302,
      "step": 6906
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2316560067896362,
      "learning_rate": 5.812662979452024e-06,
      "loss": 0.0327,
      "step": 6907
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21599583802474104,
      "learning_rate": 5.807649568934945e-06,
      "loss": 0.0401,
      "step": 6908
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2311583643501086,
      "learning_rate": 5.802637954126804e-06,
      "loss": 0.0275,
      "step": 6909
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2563924841059302,
      "learning_rate": 5.79762813566171e-06,
      "loss": 0.0435,
      "step": 6910
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2608357379551629,
      "learning_rate": 5.792620114173539e-06,
      "loss": 0.03,
      "step": 6911
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.23836058718542835,
      "learning_rate": 5.7876138902959445e-06,
      "loss": 0.0402,
      "step": 6912
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22964789607059966,
      "learning_rate": 5.782609464662334e-06,
      "loss": 0.0364,
      "step": 6913
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.30840909237662617,
      "learning_rate": 5.777606837905905e-06,
      "loss": 0.0553,
      "step": 6914
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21881703643868794,
      "learning_rate": 5.772606010659625e-06,
      "loss": 0.0404,
      "step": 6915
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17035636490086253,
      "learning_rate": 5.767606983556237e-06,
      "loss": 0.0309,
      "step": 6916
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21178796418067145,
      "learning_rate": 5.762609757228237e-06,
      "loss": 0.0299,
      "step": 6917
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2579718010928188,
      "learning_rate": 5.757614332307912e-06,
      "loss": 0.0375,
      "step": 6918
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22600900520997128,
      "learning_rate": 5.752620709427322e-06,
      "loss": 0.0439,
      "step": 6919
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24079720540191418,
      "learning_rate": 5.7476288892182905e-06,
      "loss": 0.0489,
      "step": 6920
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2581555944604589,
      "learning_rate": 5.742638872312407e-06,
      "loss": 0.0369,
      "step": 6921
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29413186930356994,
      "learning_rate": 5.737650659341048e-06,
      "loss": 0.0406,
      "step": 6922
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.18655045797950764,
      "learning_rate": 5.732664250935354e-06,
      "loss": 0.0298,
      "step": 6923
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.23384817040677708,
      "learning_rate": 5.7276796477262365e-06,
      "loss": 0.0449,
      "step": 6924
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2033207573643049,
      "learning_rate": 5.72269685034438e-06,
      "loss": 0.02,
      "step": 6925
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21485899957808957,
      "learning_rate": 5.717715859420246e-06,
      "loss": 0.0329,
      "step": 6926
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.15651580731972609,
      "learning_rate": 5.712736675584061e-06,
      "loss": 0.0238,
      "step": 6927
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1968726123937514,
      "learning_rate": 5.707759299465816e-06,
      "loss": 0.0205,
      "step": 6928
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24112351566933618,
      "learning_rate": 5.702783731695285e-06,
      "loss": 0.0379,
      "step": 6929
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.18789883752270922,
      "learning_rate": 5.6978099729020105e-06,
      "loss": 0.032,
      "step": 6930
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19796030082225394,
      "learning_rate": 5.692838023715313e-06,
      "loss": 0.0254,
      "step": 6931
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19897116441611334,
      "learning_rate": 5.68786788476426e-06,
      "loss": 0.031,
      "step": 6932
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2563013089380243,
      "learning_rate": 5.682899556677714e-06,
      "loss": 0.0358,
      "step": 6933
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17442135363094818,
      "learning_rate": 5.6779330400843e-06,
      "loss": 0.025,
      "step": 6934
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2752223887501325,
      "learning_rate": 5.672968335612421e-06,
      "loss": 0.0327,
      "step": 6935
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21292662534889523,
      "learning_rate": 5.66800544389023e-06,
      "loss": 0.0276,
      "step": 6936
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2636235369419687,
      "learning_rate": 5.663044365545674e-06,
      "loss": 0.063,
      "step": 6937
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.23667289868956498,
      "learning_rate": 5.658085101206456e-06,
      "loss": 0.0415,
      "step": 6938
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24502593693849398,
      "learning_rate": 5.653127651500065e-06,
      "loss": 0.0296,
      "step": 6939
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19311461707786928,
      "learning_rate": 5.648172017053737e-06,
      "loss": 0.0347,
      "step": 6940
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4783905177155054,
      "learning_rate": 5.643218198494493e-06,
      "loss": 0.0885,
      "step": 6941
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2429590241672004,
      "learning_rate": 5.638266196449123e-06,
      "loss": 0.0344,
      "step": 6942
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21213756149191748,
      "learning_rate": 5.6333160115442005e-06,
      "loss": 0.0312,
      "step": 6943
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17742884228765268,
      "learning_rate": 5.628367644406039e-06,
      "loss": 0.0207,
      "step": 6944
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2854125817324459,
      "learning_rate": 5.6234210956607435e-06,
      "loss": 0.0467,
      "step": 6945
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29586945445315455,
      "learning_rate": 5.618476365934184e-06,
      "loss": 0.0319,
      "step": 6946
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22614443782221783,
      "learning_rate": 5.613533455852005e-06,
      "loss": 0.0544,
      "step": 6947
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2528457207655395,
      "learning_rate": 5.608592366039607e-06,
      "loss": 0.0389,
      "step": 6948
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2840851039726853,
      "learning_rate": 5.60365309712217e-06,
      "loss": 0.0448,
      "step": 6949
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.16087403759079388,
      "learning_rate": 5.598715649724647e-06,
      "loss": 0.0245,
      "step": 6950
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20304726066714182,
      "learning_rate": 5.593780024471758e-06,
      "loss": 0.0289,
      "step": 6951
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22567431052381426,
      "learning_rate": 5.588846221987982e-06,
      "loss": 0.0493,
      "step": 6952
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.16442153450556243,
      "learning_rate": 5.583914242897581e-06,
      "loss": 0.0214,
      "step": 6953
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2316113612713826,
      "learning_rate": 5.578984087824581e-06,
      "loss": 0.0273,
      "step": 6954
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19527171662822493,
      "learning_rate": 5.574055757392787e-06,
      "loss": 0.0199,
      "step": 6955
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.30821959834681634,
      "learning_rate": 5.569129252225745e-06,
      "loss": 0.0423,
      "step": 6956
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21358375405932348,
      "learning_rate": 5.564204572946797e-06,
      "loss": 0.0286,
      "step": 6957
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2752489983828872,
      "learning_rate": 5.559281720179046e-06,
      "loss": 0.0276,
      "step": 6958
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21643923566597623,
      "learning_rate": 5.554360694545367e-06,
      "loss": 0.0395,
      "step": 6959
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2919627911624011,
      "learning_rate": 5.549441496668393e-06,
      "loss": 0.0476,
      "step": 6960
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22633794640117683,
      "learning_rate": 5.544524127170541e-06,
      "loss": 0.0462,
      "step": 6961
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1900146362822263,
      "learning_rate": 5.539608586673988e-06,
      "loss": 0.0258,
      "step": 6962
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20311661849702797,
      "learning_rate": 5.534694875800675e-06,
      "loss": 0.025,
      "step": 6963
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.26763514447217246,
      "learning_rate": 5.529782995172315e-06,
      "loss": 0.0348,
      "step": 6964
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3062852391579767,
      "learning_rate": 5.5248729454103975e-06,
      "loss": 0.041,
      "step": 6965
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20943061033167892,
      "learning_rate": 5.519964727136178e-06,
      "loss": 0.0326,
      "step": 6966
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.22073272911630623,
      "learning_rate": 5.515058340970665e-06,
      "loss": 0.0202,
      "step": 6967
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20566803691968882,
      "learning_rate": 5.510153787534651e-06,
      "loss": 0.0264,
      "step": 6968
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2464246946650871,
      "learning_rate": 5.505251067448694e-06,
      "loss": 0.0409,
      "step": 6969
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2604281645461818,
      "learning_rate": 5.500350181333121e-06,
      "loss": 0.0353,
      "step": 6970
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19740877528445158,
      "learning_rate": 5.495451129808016e-06,
      "loss": 0.0256,
      "step": 6971
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19082865857835518,
      "learning_rate": 5.490553913493242e-06,
      "loss": 0.025,
      "step": 6972
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2976076847416356,
      "learning_rate": 5.4856585330084315e-06,
      "loss": 0.0456,
      "step": 6973
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3182344367011818,
      "learning_rate": 5.48076498897298e-06,
      "loss": 0.0366,
      "step": 6974
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3705311576313587,
      "learning_rate": 5.475873282006041e-06,
      "loss": 0.039,
      "step": 6975
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2187700262439578,
      "learning_rate": 5.470983412726547e-06,
      "loss": 0.0277,
      "step": 6976
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.22780393974223076,
      "learning_rate": 5.466095381753207e-06,
      "loss": 0.0281,
      "step": 6977
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1866740672065089,
      "learning_rate": 5.461209189704486e-06,
      "loss": 0.0191,
      "step": 6978
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2198423631484719,
      "learning_rate": 5.4563248371986055e-06,
      "loss": 0.0273,
      "step": 6979
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2563586278111561,
      "learning_rate": 5.451442324853571e-06,
      "loss": 0.0426,
      "step": 6980
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21093825940595903,
      "learning_rate": 5.446561653287152e-06,
      "loss": 0.0277,
      "step": 6981
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2056065699106746,
      "learning_rate": 5.441682823116887e-06,
      "loss": 0.0296,
      "step": 6982
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2382169933917382,
      "learning_rate": 5.436805834960066e-06,
      "loss": 0.034,
      "step": 6983
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.24752212948076216,
      "learning_rate": 5.431930689433762e-06,
      "loss": 0.0282,
      "step": 6984
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2999169453252791,
      "learning_rate": 5.4270573871548145e-06,
      "loss": 0.0305,
      "step": 6985
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2500953939351185,
      "learning_rate": 5.422185928739827e-06,
      "loss": 0.0381,
      "step": 6986
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18438252983234843,
      "learning_rate": 5.4173163148051585e-06,
      "loss": 0.0292,
      "step": 6987
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.211526023236054,
      "learning_rate": 5.4124485459669485e-06,
      "loss": 0.0305,
      "step": 6988
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19899179583319965,
      "learning_rate": 5.407582622841103e-06,
      "loss": 0.0305,
      "step": 6989
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.25614371770600686,
      "learning_rate": 5.402718546043293e-06,
      "loss": 0.0269,
      "step": 6990
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.22575712108047447,
      "learning_rate": 5.397856316188941e-06,
      "loss": 0.0256,
      "step": 6991
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.24717336322469746,
      "learning_rate": 5.392995933893255e-06,
      "loss": 0.036,
      "step": 6992
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2919907330633065,
      "learning_rate": 5.388137399771203e-06,
      "loss": 0.0313,
      "step": 6993
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.25249117687125416,
      "learning_rate": 5.383280714437518e-06,
      "loss": 0.0311,
      "step": 6994
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.24682961300225534,
      "learning_rate": 5.378425878506699e-06,
      "loss": 0.0407,
      "step": 6995
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21672774406917658,
      "learning_rate": 5.373572892593013e-06,
      "loss": 0.0471,
      "step": 6996
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2527252166126367,
      "learning_rate": 5.368721757310498e-06,
      "loss": 0.0303,
      "step": 6997
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19257475867433962,
      "learning_rate": 5.363872473272935e-06,
      "loss": 0.0218,
      "step": 6998
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2320130463083775,
      "learning_rate": 5.359025041093897e-06,
      "loss": 0.0324,
      "step": 6999
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21244071526155064,
      "learning_rate": 5.354179461386712e-06,
      "loss": 0.0492,
      "step": 7000
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5452902037863576,
      "learning_rate": 5.349335734764483e-06,
      "loss": 0.0808,
      "step": 7001
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3080990362931693,
      "learning_rate": 5.3444938618400524e-06,
      "loss": 0.0396,
      "step": 7002
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18269649354342107,
      "learning_rate": 5.3396538432260556e-06,
      "loss": 0.0183,
      "step": 7003
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.318177504750219,
      "learning_rate": 5.334815679534882e-06,
      "loss": 0.0389,
      "step": 7004
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2037615522081898,
      "learning_rate": 5.329979371378693e-06,
      "loss": 0.0328,
      "step": 7005
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21685994579713638,
      "learning_rate": 5.325144919369398e-06,
      "loss": 0.0337,
      "step": 7006
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.26275105221825307,
      "learning_rate": 5.320312324118693e-06,
      "loss": 0.0387,
      "step": 7007
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2178134795868791,
      "learning_rate": 5.315481586238025e-06,
      "loss": 0.0208,
      "step": 7008
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.28813787735650404,
      "learning_rate": 5.310652706338619e-06,
      "loss": 0.0398,
      "step": 7009
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.257843969331823,
      "learning_rate": 5.305825685031445e-06,
      "loss": 0.0373,
      "step": 7010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2546008073665107,
      "learning_rate": 5.301000522927247e-06,
      "loss": 0.0306,
      "step": 7011
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1940067512797477,
      "learning_rate": 5.296177220636556e-06,
      "loss": 0.0292,
      "step": 7012
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21660280562681306,
      "learning_rate": 5.291355778769629e-06,
      "loss": 0.04,
      "step": 7013
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.22972912022049433,
      "learning_rate": 5.286536197936512e-06,
      "loss": 0.0436,
      "step": 7014
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20096376450609368,
      "learning_rate": 5.281718478747009e-06,
      "loss": 0.0279,
      "step": 7015
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33209048472082764,
      "learning_rate": 5.276902621810691e-06,
      "loss": 0.0487,
      "step": 7016
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2586955495339104,
      "learning_rate": 5.272088627736898e-06,
      "loss": 0.0325,
      "step": 7017
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.28198918586313926,
      "learning_rate": 5.267276497134715e-06,
      "loss": 0.0398,
      "step": 7018
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.24910528296554205,
      "learning_rate": 5.26246623061301e-06,
      "loss": 0.0323,
      "step": 7019
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.23293844886895632,
      "learning_rate": 5.257657828780409e-06,
      "loss": 0.0393,
      "step": 7020
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2570166731530166,
      "learning_rate": 5.252851292245311e-06,
      "loss": 0.0396,
      "step": 7021
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2495247376511187,
      "learning_rate": 5.2480466216158565e-06,
      "loss": 0.0458,
      "step": 7022
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.26271737725716104,
      "learning_rate": 5.243243817499968e-06,
      "loss": 0.0366,
      "step": 7023
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.26338079421427896,
      "learning_rate": 5.2384428805053325e-06,
      "loss": 0.0389,
      "step": 7024
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.23889113374020046,
      "learning_rate": 5.233643811239395e-06,
      "loss": 0.0463,
      "step": 7025
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.26312777576658714,
      "learning_rate": 5.228846610309359e-06,
      "loss": 0.0421,
      "step": 7026
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3745738429301858,
      "learning_rate": 5.224051278322204e-06,
      "loss": 0.0489,
      "step": 7027
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.28799493936556236,
      "learning_rate": 5.219257815884662e-06,
      "loss": 0.038,
      "step": 7028
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.24428743144716925,
      "learning_rate": 5.214466223603238e-06,
      "loss": 0.0309,
      "step": 7029
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.28104745671326653,
      "learning_rate": 5.209676502084191e-06,
      "loss": 0.0484,
      "step": 7030
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.26035724756740336,
      "learning_rate": 5.204888651933553e-06,
      "loss": 0.0383,
      "step": 7031
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.26880693553370694,
      "learning_rate": 5.200102673757115e-06,
      "loss": 0.0453,
      "step": 7032
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.14152478734677498,
      "learning_rate": 5.195318568160421e-06,
      "loss": 0.0137,
      "step": 7033
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.25012842992216233,
      "learning_rate": 5.190536335748792e-06,
      "loss": 0.0411,
      "step": 7034
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3170977441985976,
      "learning_rate": 5.185755977127311e-06,
      "loss": 0.0536,
      "step": 7035
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.25279435473420747,
      "learning_rate": 5.180977492900823e-06,
      "loss": 0.0277,
      "step": 7036
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2353987587106853,
      "learning_rate": 5.1762008836739185e-06,
      "loss": 0.0498,
      "step": 7037
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.24286597338142374,
      "learning_rate": 5.171426150050977e-06,
      "loss": 0.0299,
      "step": 7038
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21200146608587972,
      "learning_rate": 5.1666532926361255e-06,
      "loss": 0.0479,
      "step": 7039
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2996070160455856,
      "learning_rate": 5.161882312033264e-06,
      "loss": 0.0633,
      "step": 7040
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2505240015211575,
      "learning_rate": 5.157113208846037e-06,
      "loss": 0.0296,
      "step": 7041
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.23581899230900907,
      "learning_rate": 5.152345983677866e-06,
      "loss": 0.0226,
      "step": 7042
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2478916929325985,
      "learning_rate": 5.147580637131935e-06,
      "loss": 0.0394,
      "step": 7043
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.17358674209485858,
      "learning_rate": 5.142817169811189e-06,
      "loss": 0.0248,
      "step": 7044
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20393352801096296,
      "learning_rate": 5.13805558231832e-06,
      "loss": 0.0348,
      "step": 7045
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19399537384463117,
      "learning_rate": 5.133295875255808e-06,
      "loss": 0.0184,
      "step": 7046
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18247770398607657,
      "learning_rate": 5.128538049225884e-06,
      "loss": 0.0293,
      "step": 7047
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20700894541519846,
      "learning_rate": 5.1237821048305305e-06,
      "loss": 0.0343,
      "step": 7048
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.16088415669196585,
      "learning_rate": 5.119028042671501e-06,
      "loss": 0.0247,
      "step": 7049
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.25981119659625457,
      "learning_rate": 5.114275863350313e-06,
      "loss": 0.0336,
      "step": 7050
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3039939395926302,
      "learning_rate": 5.109525567468252e-06,
      "loss": 0.0322,
      "step": 7051
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18998088279342515,
      "learning_rate": 5.104777155626341e-06,
      "loss": 0.0291,
      "step": 7052
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2860743361990016,
      "learning_rate": 5.100030628425388e-06,
      "loss": 0.0489,
      "step": 7053
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19724893971497856,
      "learning_rate": 5.095285986465952e-06,
      "loss": 0.0282,
      "step": 7054
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2687032392763007,
      "learning_rate": 5.090543230348366e-06,
      "loss": 0.0331,
      "step": 7055
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18303929468301616,
      "learning_rate": 5.085802360672701e-06,
      "loss": 0.0259,
      "step": 7056
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.17444892996131578,
      "learning_rate": 5.081063378038806e-06,
      "loss": 0.0171,
      "step": 7057
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20331597520524242,
      "learning_rate": 5.076326283046291e-06,
      "loss": 0.031,
      "step": 7058
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.41325724771816824,
      "learning_rate": 5.0715910762945245e-06,
      "loss": 0.0599,
      "step": 7059
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20221508235962637,
      "learning_rate": 5.066857758382642e-06,
      "loss": 0.029,
      "step": 7060
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2390238026234169,
      "learning_rate": 5.062126329909518e-06,
      "loss": 0.0393,
      "step": 7061
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21206172070583498,
      "learning_rate": 5.057396791473807e-06,
      "loss": 0.0252,
      "step": 7062
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20448550019355807,
      "learning_rate": 5.05266914367394e-06,
      "loss": 0.0414,
      "step": 7063
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20824523592957186,
      "learning_rate": 5.047943387108072e-06,
      "loss": 0.0365,
      "step": 7064
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.286528785307214,
      "learning_rate": 5.043219522374138e-06,
      "loss": 0.0345,
      "step": 7065
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22294669699193048,
      "learning_rate": 5.038497550069836e-06,
      "loss": 0.0383,
      "step": 7066
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.15719723872281507,
      "learning_rate": 5.033777470792627e-06,
      "loss": 0.0132,
      "step": 7067
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.27240964618669405,
      "learning_rate": 5.029059285139715e-06,
      "loss": 0.0437,
      "step": 7068
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22839418265390438,
      "learning_rate": 5.024342993708078e-06,
      "loss": 0.0234,
      "step": 7069
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.27050998838355206,
      "learning_rate": 5.019628597094455e-06,
      "loss": 0.0303,
      "step": 7070
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23354315959465716,
      "learning_rate": 5.014916095895346e-06,
      "loss": 0.0322,
      "step": 7071
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19399194608148468,
      "learning_rate": 5.010205490706998e-06,
      "loss": 0.0284,
      "step": 7072
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23230742994774076,
      "learning_rate": 5.005496782125434e-06,
      "loss": 0.0214,
      "step": 7073
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24544607947721744,
      "learning_rate": 5.000789970746427e-06,
      "loss": 0.0285,
      "step": 7074
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22568602220193534,
      "learning_rate": 4.996085057165523e-06,
      "loss": 0.0299,
      "step": 7075
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.27522268653238274,
      "learning_rate": 4.9913820419780035e-06,
      "loss": 0.0391,
      "step": 7076
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2166246812898434,
      "learning_rate": 4.9866809257789325e-06,
      "loss": 0.0233,
      "step": 7077
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2193627291016119,
      "learning_rate": 4.981981709163126e-06,
      "loss": 0.0398,
      "step": 7078
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23855988049137086,
      "learning_rate": 4.977284392725166e-06,
      "loss": 0.0298,
      "step": 7079
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.26524348067775994,
      "learning_rate": 4.972588977059369e-06,
      "loss": 0.048,
      "step": 7080
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.216200110311705,
      "learning_rate": 4.967895462759846e-06,
      "loss": 0.0332,
      "step": 7081
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20991447289452403,
      "learning_rate": 4.963203850420455e-06,
      "loss": 0.0285,
      "step": 7082
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20960717672474216,
      "learning_rate": 4.958514140634794e-06,
      "loss": 0.0284,
      "step": 7083
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19099572005831705,
      "learning_rate": 4.953826333996243e-06,
      "loss": 0.0376,
      "step": 7084
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20503317022332884,
      "learning_rate": 4.949140431097934e-06,
      "loss": 0.036,
      "step": 7085
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3208452191050301,
      "learning_rate": 4.944456432532765e-06,
      "loss": 0.039,
      "step": 7086
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22067291761969213,
      "learning_rate": 4.93977433889337e-06,
      "loss": 0.0444,
      "step": 7087
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.274916585820955,
      "learning_rate": 4.93509415077217e-06,
      "loss": 0.0474,
      "step": 7088
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2029797878597161,
      "learning_rate": 4.930415868761332e-06,
      "loss": 0.034,
      "step": 7089
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2955677715419463,
      "learning_rate": 4.925739493452783e-06,
      "loss": 0.0474,
      "step": 7090
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22196154384094519,
      "learning_rate": 4.9210650254382005e-06,
      "loss": 0.0362,
      "step": 7091
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1972197110970643,
      "learning_rate": 4.916392465309037e-06,
      "loss": 0.0239,
      "step": 7092
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3031942873352124,
      "learning_rate": 4.911721813656494e-06,
      "loss": 0.0339,
      "step": 7093
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.28387594868557303,
      "learning_rate": 4.907053071071535e-06,
      "loss": 0.0466,
      "step": 7094
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22284934387974892,
      "learning_rate": 4.902386238144874e-06,
      "loss": 0.0329,
      "step": 7095
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23515567304217255,
      "learning_rate": 4.89772131546699e-06,
      "loss": 0.0268,
      "step": 7096
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19799529049009537,
      "learning_rate": 4.893058303628115e-06,
      "loss": 0.0323,
      "step": 7097
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19242451791810714,
      "learning_rate": 4.888397203218265e-06,
      "loss": 0.0287,
      "step": 7098
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.18424288655476692,
      "learning_rate": 4.88373801482717e-06,
      "loss": 0.0356,
      "step": 7099
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.27399806736522087,
      "learning_rate": 4.879080739044351e-06,
      "loss": 0.038,
      "step": 7100
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19106941362936872,
      "learning_rate": 4.874425376459073e-06,
      "loss": 0.0263,
      "step": 7101
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2049932031551427,
      "learning_rate": 4.869771927660371e-06,
      "loss": 0.0469,
      "step": 7102
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.12816239805573748,
      "learning_rate": 4.865120393237019e-06,
      "loss": 0.0168,
      "step": 7103
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20605942417865544,
      "learning_rate": 4.860470773777566e-06,
      "loss": 0.023,
      "step": 7104
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.26985367549628797,
      "learning_rate": 4.855823069870309e-06,
      "loss": 0.0371,
      "step": 7105
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22160236337652517,
      "learning_rate": 4.851177282103312e-06,
      "loss": 0.0414,
      "step": 7106
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2043381373965783,
      "learning_rate": 4.846533411064382e-06,
      "loss": 0.0359,
      "step": 7107
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22184569071216248,
      "learning_rate": 4.841891457341095e-06,
      "loss": 0.0332,
      "step": 7108
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.27263404788587753,
      "learning_rate": 4.837251421520781e-06,
      "loss": 0.0376,
      "step": 7109
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23986932885273013,
      "learning_rate": 4.832613304190537e-06,
      "loss": 0.0342,
      "step": 7110
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.265632348800429,
      "learning_rate": 4.827977105937194e-06,
      "loss": 0.028,
      "step": 7111
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23589280917835584,
      "learning_rate": 4.823342827347357e-06,
      "loss": 0.0474,
      "step": 7112
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24107644350900834,
      "learning_rate": 4.81871046900739e-06,
      "loss": 0.0332,
      "step": 7113
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23450582349923524,
      "learning_rate": 4.814080031503407e-06,
      "loss": 0.0288,
      "step": 7114
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3286385150987333,
      "learning_rate": 4.809451515421282e-06,
      "loss": 0.0402,
      "step": 7115
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24546769147386088,
      "learning_rate": 4.804824921346645e-06,
      "loss": 0.0481,
      "step": 7116
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.28899800380096574,
      "learning_rate": 4.800200249864888e-06,
      "loss": 0.0491,
      "step": 7117
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19867234938396502,
      "learning_rate": 4.795577501561144e-06,
      "loss": 0.0332,
      "step": 7118
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.18439832879232843,
      "learning_rate": 4.790956677020318e-06,
      "loss": 0.032,
      "step": 7119
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1806893305248692,
      "learning_rate": 4.786337776827066e-06,
      "loss": 0.0303,
      "step": 7120
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22563755386171844,
      "learning_rate": 4.78172080156581e-06,
      "loss": 0.026,
      "step": 7121
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.16889515722127768,
      "learning_rate": 4.777105751820708e-06,
      "loss": 0.024,
      "step": 7122
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17531771471900265,
      "learning_rate": 4.772492628175689e-06,
      "loss": 0.0164,
      "step": 7123
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21394301511636027,
      "learning_rate": 4.767881431214441e-06,
      "loss": 0.034,
      "step": 7124
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2237262780385221,
      "learning_rate": 4.7632721615204026e-06,
      "loss": 0.0496,
      "step": 7125
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.25818663240008555,
      "learning_rate": 4.758664819676759e-06,
      "loss": 0.0361,
      "step": 7126
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2407501081246582,
      "learning_rate": 4.754059406266469e-06,
      "loss": 0.0374,
      "step": 7127
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22687041799526456,
      "learning_rate": 4.7494559218722395e-06,
      "loss": 0.0336,
      "step": 7128
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22490714693758546,
      "learning_rate": 4.7448543670765365e-06,
      "loss": 0.0446,
      "step": 7129
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22175242369250836,
      "learning_rate": 4.74025474246157e-06,
      "loss": 0.0259,
      "step": 7130
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19588816683345786,
      "learning_rate": 4.735657048609313e-06,
      "loss": 0.0262,
      "step": 7131
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24040090789968613,
      "learning_rate": 4.7310612861015125e-06,
      "loss": 0.0373,
      "step": 7132
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.29827285738808434,
      "learning_rate": 4.726467455519639e-06,
      "loss": 0.0439,
      "step": 7133
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24358759231810329,
      "learning_rate": 4.7218755574449394e-06,
      "loss": 0.036,
      "step": 7134
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2209561593810618,
      "learning_rate": 4.717285592458409e-06,
      "loss": 0.0506,
      "step": 7135
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.227784920345592,
      "learning_rate": 4.712697561140802e-06,
      "loss": 0.0446,
      "step": 7136
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.318987945491812,
      "learning_rate": 4.708111464072631e-06,
      "loss": 0.041,
      "step": 7137
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.18720442601452222,
      "learning_rate": 4.703527301834148e-06,
      "loss": 0.0308,
      "step": 7138
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4186644424032141,
      "learning_rate": 4.698945075005376e-06,
      "loss": 0.059,
      "step": 7139
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1876109907344866,
      "learning_rate": 4.69436478416609e-06,
      "loss": 0.0286,
      "step": 7140
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1840340630239719,
      "learning_rate": 4.689786429895821e-06,
      "loss": 0.0329,
      "step": 7141
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17822269805489663,
      "learning_rate": 4.685210012773844e-06,
      "loss": 0.0262,
      "step": 7142
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2942919287533011,
      "learning_rate": 4.680635533379201e-06,
      "loss": 0.0369,
      "step": 7143
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2470827438785433,
      "learning_rate": 4.676062992290686e-06,
      "loss": 0.0255,
      "step": 7144
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2485888260186477,
      "learning_rate": 4.671492390086854e-06,
      "loss": 0.0276,
      "step": 7145
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.28644153227809366,
      "learning_rate": 4.666923727345991e-06,
      "loss": 0.0424,
      "step": 7146
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19935511615460752,
      "learning_rate": 4.662357004646165e-06,
      "loss": 0.0306,
      "step": 7147
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24604430534526286,
      "learning_rate": 4.657792222565185e-06,
      "loss": 0.0358,
      "step": 7148
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.22886739049467247,
      "learning_rate": 4.653229381680618e-06,
      "loss": 0.0588,
      "step": 7149
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2368928923429487,
      "learning_rate": 4.6486684825697845e-06,
      "loss": 0.0327,
      "step": 7150
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3019565434034133,
      "learning_rate": 4.644109525809759e-06,
      "loss": 0.0366,
      "step": 7151
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24875938777605114,
      "learning_rate": 4.639552511977374e-06,
      "loss": 0.0509,
      "step": 7152
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2015093408021646,
      "learning_rate": 4.6349974416492076e-06,
      "loss": 0.0293,
      "step": 7153
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.17503663343784281,
      "learning_rate": 4.630444315401594e-06,
      "loss": 0.0365,
      "step": 7154
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.22330631760068748,
      "learning_rate": 4.625893133810633e-06,
      "loss": 0.0302,
      "step": 7155
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.16254767374240126,
      "learning_rate": 4.621343897452169e-06,
      "loss": 0.0273,
      "step": 7156
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2216907195988581,
      "learning_rate": 4.616796606901795e-06,
      "loss": 0.0432,
      "step": 7157
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1832942040817646,
      "learning_rate": 4.612251262734864e-06,
      "loss": 0.0199,
      "step": 7158
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2108783815686001,
      "learning_rate": 4.607707865526488e-06,
      "loss": 0.0525,
      "step": 7159
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24065777443954176,
      "learning_rate": 4.603166415851527e-06,
      "loss": 0.0382,
      "step": 7160
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2681096082106456,
      "learning_rate": 4.5986269142845915e-06,
      "loss": 0.045,
      "step": 7161
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.21947794093903353,
      "learning_rate": 4.594089361400047e-06,
      "loss": 0.0337,
      "step": 7162
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.23302886944486295,
      "learning_rate": 4.5895537577720205e-06,
      "loss": 0.0213,
      "step": 7163
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19341405422930102,
      "learning_rate": 4.585020103974387e-06,
      "loss": 0.0178,
      "step": 7164
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19684455802740736,
      "learning_rate": 4.580488400580765e-06,
      "loss": 0.029,
      "step": 7165
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2486288497381265,
      "learning_rate": 4.575958648164536e-06,
      "loss": 0.0471,
      "step": 7166
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.26865002413781003,
      "learning_rate": 4.571430847298847e-06,
      "loss": 0.0431,
      "step": 7167
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19638044213537073,
      "learning_rate": 4.5669049985565735e-06,
      "loss": 0.0294,
      "step": 7168
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1851259878835393,
      "learning_rate": 4.5623811025103585e-06,
      "loss": 0.0257,
      "step": 7169
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.22322706202999607,
      "learning_rate": 4.5578591597325935e-06,
      "loss": 0.0313,
      "step": 7170
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3013621331691586,
      "learning_rate": 4.553339170795432e-06,
      "loss": 0.0442,
      "step": 7171
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18648484714609698,
      "learning_rate": 4.54882113627076e-06,
      "loss": 0.0289,
      "step": 7172
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2056967713808739,
      "learning_rate": 4.544305056730236e-06,
      "loss": 0.0252,
      "step": 7173
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2147100379307889,
      "learning_rate": 4.53979093274526e-06,
      "loss": 0.0399,
      "step": 7174
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18553464748850038,
      "learning_rate": 4.535278764886994e-06,
      "loss": 0.0197,
      "step": 7175
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24516378013079404,
      "learning_rate": 4.530768553726348e-06,
      "loss": 0.0474,
      "step": 7176
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24318379394875955,
      "learning_rate": 4.526260299833975e-06,
      "loss": 0.037,
      "step": 7177
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3943790293214147,
      "learning_rate": 4.521754003780294e-06,
      "loss": 0.0511,
      "step": 7178
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1896732663717066,
      "learning_rate": 4.517249666135468e-06,
      "loss": 0.0267,
      "step": 7179
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.215306520572789,
      "learning_rate": 4.512747287469426e-06,
      "loss": 0.0336,
      "step": 7180
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.16702007038611053,
      "learning_rate": 4.508246868351822e-06,
      "loss": 0.0285,
      "step": 7181
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24656831306596144,
      "learning_rate": 4.503748409352089e-06,
      "loss": 0.0454,
      "step": 7182
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18917855294907626,
      "learning_rate": 4.499251911039397e-06,
      "loss": 0.0266,
      "step": 7183
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.31749326691284085,
      "learning_rate": 4.494757373982674e-06,
      "loss": 0.0401,
      "step": 7184
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2008329239635991,
      "learning_rate": 4.490264798750599e-06,
      "loss": 0.0302,
      "step": 7185
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24711993627565226,
      "learning_rate": 4.4857741859116024e-06,
      "loss": 0.0332,
      "step": 7186
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1963125565925621,
      "learning_rate": 4.4812855360338684e-06,
      "loss": 0.0267,
      "step": 7187
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.30581139861781137,
      "learning_rate": 4.476798849685322e-06,
      "loss": 0.0303,
      "step": 7188
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2517301770676468,
      "learning_rate": 4.472314127433654e-06,
      "loss": 0.0295,
      "step": 7189
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2385067654290901,
      "learning_rate": 4.467831369846301e-06,
      "loss": 0.0538,
      "step": 7190
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18815070727537506,
      "learning_rate": 4.4633505774904525e-06,
      "loss": 0.0296,
      "step": 7191
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.17006587361554348,
      "learning_rate": 4.458871750933038e-06,
      "loss": 0.0225,
      "step": 7192
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18014816651484594,
      "learning_rate": 4.4543948907407565e-06,
      "loss": 0.0317,
      "step": 7193
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.29838820095209245,
      "learning_rate": 4.449919997480047e-06,
      "loss": 0.0404,
      "step": 7194
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.23245130080189177,
      "learning_rate": 4.445447071717111e-06,
      "loss": 0.0255,
      "step": 7195
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.23549077360693527,
      "learning_rate": 4.4409761140178765e-06,
      "loss": 0.0313,
      "step": 7196
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3126686356439457,
      "learning_rate": 4.4365071249480465e-06,
      "loss": 0.0408,
      "step": 7197
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2498958713517684,
      "learning_rate": 4.432040105073065e-06,
      "loss": 0.0456,
      "step": 7198
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.20521245825602638,
      "learning_rate": 4.427575054958135e-06,
      "loss": 0.0502,
      "step": 7199
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2496850891378838,
      "learning_rate": 4.4231119751681885e-06,
      "loss": 0.0462,
      "step": 7200
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.22993245200619342,
      "learning_rate": 4.418650866267939e-06,
      "loss": 0.0405,
      "step": 7201
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2232953986883259,
      "learning_rate": 4.414191728821838e-06,
      "loss": 0.041,
      "step": 7202
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19814779355074183,
      "learning_rate": 4.409734563394068e-06,
      "loss": 0.0449,
      "step": 7203
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.17990734293428287,
      "learning_rate": 4.405279370548587e-06,
      "loss": 0.025,
      "step": 7204
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18586921800523507,
      "learning_rate": 4.400826150849096e-06,
      "loss": 0.0366,
      "step": 7205
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2316034476952561,
      "learning_rate": 4.396374904859051e-06,
      "loss": 0.0416,
      "step": 7206
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.32250203795491444,
      "learning_rate": 4.391925633141638e-06,
      "loss": 0.0668,
      "step": 7207
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.244388149668675,
      "learning_rate": 4.387478336259821e-06,
      "loss": 0.0468,
      "step": 7208
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2485689691052336,
      "learning_rate": 4.383033014776295e-06,
      "loss": 0.0531,
      "step": 7209
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.28475634084880813,
      "learning_rate": 4.3785896692535165e-06,
      "loss": 0.0433,
      "step": 7210
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2130011088108475,
      "learning_rate": 4.3741483002536775e-06,
      "loss": 0.0208,
      "step": 7211
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3025118101982933,
      "learning_rate": 4.369708908338735e-06,
      "loss": 0.0449,
      "step": 7212
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24423563824992509,
      "learning_rate": 4.365271494070389e-06,
      "loss": 0.0242,
      "step": 7213
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1787324277536797,
      "learning_rate": 4.360836058010096e-06,
      "loss": 0.032,
      "step": 7214
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.27283793519818544,
      "learning_rate": 4.356402600719045e-06,
      "loss": 0.0352,
      "step": 7215
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2742632658600584,
      "learning_rate": 4.351971122758194e-06,
      "loss": 0.0275,
      "step": 7216
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2002178365550513,
      "learning_rate": 4.3475416246882385e-06,
      "loss": 0.0396,
      "step": 7217
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.270840987072905,
      "learning_rate": 4.343114107069628e-06,
      "loss": 0.0592,
      "step": 7218
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.29874445173865516,
      "learning_rate": 4.338688570462563e-06,
      "loss": 0.0514,
      "step": 7219
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.15539143235585182,
      "learning_rate": 4.334265015426993e-06,
      "loss": 0.0277,
      "step": 7220
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.25961655089396246,
      "learning_rate": 4.32984344252261e-06,
      "loss": 0.0347,
      "step": 7221
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2920618899024944,
      "learning_rate": 4.3254238523088695e-06,
      "loss": 0.0388,
      "step": 7222
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2515800286340595,
      "learning_rate": 4.3210062453449545e-06,
      "loss": 0.0376,
      "step": 7223
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2950708126892183,
      "learning_rate": 4.316590622189815e-06,
      "loss": 0.0392,
      "step": 7224
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2314437351092575,
      "learning_rate": 4.312176983402145e-06,
      "loss": 0.035,
      "step": 7225
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.28408489415997096,
      "learning_rate": 4.307765329540394e-06,
      "loss": 0.0299,
      "step": 7226
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24959465783420848,
      "learning_rate": 4.303355661162741e-06,
      "loss": 0.0483,
      "step": 7227
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2076560525053648,
      "learning_rate": 4.298947978827128e-06,
      "loss": 0.034,
      "step": 7228
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.25602230679101495,
      "learning_rate": 4.2945422830912495e-06,
      "loss": 0.0402,
      "step": 7229
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2768406670598191,
      "learning_rate": 4.290138574512546e-06,
      "loss": 0.0221,
      "step": 7230
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.219052956464382,
      "learning_rate": 4.285736853648193e-06,
      "loss": 0.0334,
      "step": 7231
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2252603054965118,
      "learning_rate": 4.2813371210551294e-06,
      "loss": 0.0363,
      "step": 7232
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2148567282484756,
      "learning_rate": 4.276939377290037e-06,
      "loss": 0.0405,
      "step": 7233
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1946691885766725,
      "learning_rate": 4.272543622909355e-06,
      "loss": 0.0214,
      "step": 7234
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.17487108045961364,
      "learning_rate": 4.268149858469246e-06,
      "loss": 0.0183,
      "step": 7235
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.21066032849155022,
      "learning_rate": 4.263758084525656e-06,
      "loss": 0.0303,
      "step": 7236
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24427570420853645,
      "learning_rate": 4.259368301634257e-06,
      "loss": 0.0381,
      "step": 7237
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2085337370363043,
      "learning_rate": 4.254980510350464e-06,
      "loss": 0.027,
      "step": 7238
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2646034878215807,
      "learning_rate": 4.250594711229454e-06,
      "loss": 0.034,
      "step": 7239
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.30299725347791207,
      "learning_rate": 4.246210904826149e-06,
      "loss": 0.0341,
      "step": 7240
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1829265890565706,
      "learning_rate": 4.24182909169522e-06,
      "loss": 0.0276,
      "step": 7241
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.26593498766706414,
      "learning_rate": 4.237449272391072e-06,
      "loss": 0.0326,
      "step": 7242
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.202994066000166,
      "learning_rate": 4.233071447467876e-06,
      "loss": 0.0269,
      "step": 7243
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23736513411904328,
      "learning_rate": 4.228695617479541e-06,
      "loss": 0.032,
      "step": 7244
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19945480348961828,
      "learning_rate": 4.22432178297973e-06,
      "loss": 0.0313,
      "step": 7245
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1747158719976968,
      "learning_rate": 4.219949944521842e-06,
      "loss": 0.0347,
      "step": 7246
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20491102768012978,
      "learning_rate": 4.215580102659031e-06,
      "loss": 0.0238,
      "step": 7247
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19409592342188592,
      "learning_rate": 4.2112122579442015e-06,
      "loss": 0.0446,
      "step": 7248
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.24526514094294566,
      "learning_rate": 4.206846410930008e-06,
      "loss": 0.0375,
      "step": 7249
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.18689115338059054,
      "learning_rate": 4.202482562168832e-06,
      "loss": 0.027,
      "step": 7250
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.22866039092195897,
      "learning_rate": 4.198120712212823e-06,
      "loss": 0.0273,
      "step": 7251
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20828330015359683,
      "learning_rate": 4.193760861613865e-06,
      "loss": 0.0347,
      "step": 7252
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23221984866409456,
      "learning_rate": 4.189403010923614e-06,
      "loss": 0.0302,
      "step": 7253
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.25478049669139174,
      "learning_rate": 4.185047160693432e-06,
      "loss": 0.0527,
      "step": 7254
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1891855729908844,
      "learning_rate": 4.18069331147446e-06,
      "loss": 0.03,
      "step": 7255
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1495294561647225,
      "learning_rate": 4.176341463817573e-06,
      "loss": 0.0217,
      "step": 7256
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2243583188324651,
      "learning_rate": 4.171991618273401e-06,
      "loss": 0.0369,
      "step": 7257
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19205187881514563,
      "learning_rate": 4.167643775392305e-06,
      "loss": 0.0236,
      "step": 7258
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.24354774966908543,
      "learning_rate": 4.1632979357244065e-06,
      "loss": 0.0401,
      "step": 7259
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.24764599282476804,
      "learning_rate": 4.1589540998195695e-06,
      "loss": 0.0408,
      "step": 7260
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3020765972482597,
      "learning_rate": 4.1546122682274114e-06,
      "loss": 0.0442,
      "step": 7261
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2176942039692882,
      "learning_rate": 4.150272441497276e-06,
      "loss": 0.0349,
      "step": 7262
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27023797744418715,
      "learning_rate": 4.145934620178276e-06,
      "loss": 0.0431,
      "step": 7263
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21292073798035252,
      "learning_rate": 4.141598804819256e-06,
      "loss": 0.045,
      "step": 7264
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.14978775341504655,
      "learning_rate": 4.137264995968821e-06,
      "loss": 0.0146,
      "step": 7265
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20463800887571834,
      "learning_rate": 4.132933194175299e-06,
      "loss": 0.0326,
      "step": 7266
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37764577526738075,
      "learning_rate": 4.128603399986786e-06,
      "loss": 0.061,
      "step": 7267
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23873546958066713,
      "learning_rate": 4.124275613951114e-06,
      "loss": 0.0421,
      "step": 7268
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33564860097567145,
      "learning_rate": 4.119949836615864e-06,
      "loss": 0.0279,
      "step": 7269
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2066706179537851,
      "learning_rate": 4.115626068528362e-06,
      "loss": 0.0269,
      "step": 7270
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.25533851722095974,
      "learning_rate": 4.111304310235677e-06,
      "loss": 0.0412,
      "step": 7271
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1839250051679114,
      "learning_rate": 4.106984562284633e-06,
      "loss": 0.0298,
      "step": 7272
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2661371902784373,
      "learning_rate": 4.102666825221782e-06,
      "loss": 0.0343,
      "step": 7273
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.162178926599613,
      "learning_rate": 4.0983510995934365e-06,
      "loss": 0.0308,
      "step": 7274
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.22450044727280885,
      "learning_rate": 4.094037385945653e-06,
      "loss": 0.0263,
      "step": 7275
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2022797623280109,
      "learning_rate": 4.089725684824235e-06,
      "loss": 0.0389,
      "step": 7276
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2673571990223701,
      "learning_rate": 4.085415996774713e-06,
      "loss": 0.0415,
      "step": 7277
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20757784753335104,
      "learning_rate": 4.081108322342389e-06,
      "loss": 0.0231,
      "step": 7278
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19718298739122775,
      "learning_rate": 4.076802662072292e-06,
      "loss": 0.0292,
      "step": 7279
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2418098273682959,
      "learning_rate": 4.07249901650921e-06,
      "loss": 0.0374,
      "step": 7280
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21103589977043397,
      "learning_rate": 4.068197386197657e-06,
      "loss": 0.0315,
      "step": 7281
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3596431312405758,
      "learning_rate": 4.0638977716819105e-06,
      "loss": 0.052,
      "step": 7282
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.17590913359270413,
      "learning_rate": 4.059600173505984e-06,
      "loss": 0.0149,
      "step": 7283
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23740582606759936,
      "learning_rate": 4.055304592213645e-06,
      "loss": 0.0297,
      "step": 7284
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27166826894693274,
      "learning_rate": 4.051011028348384e-06,
      "loss": 0.0379,
      "step": 7285
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21230503352469537,
      "learning_rate": 4.046719482453461e-06,
      "loss": 0.0262,
      "step": 7286
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23170149583942842,
      "learning_rate": 4.042429955071867e-06,
      "loss": 0.0458,
      "step": 7287
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4131956202849141,
      "learning_rate": 4.038142446746342e-06,
      "loss": 0.0699,
      "step": 7288
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2397054416352333,
      "learning_rate": 4.033856958019371e-06,
      "loss": 0.032,
      "step": 7289
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1754527253912369,
      "learning_rate": 4.029573489433179e-06,
      "loss": 0.0212,
      "step": 7290
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.16757525883960098,
      "learning_rate": 4.025292041529747e-06,
      "loss": 0.0149,
      "step": 7291
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.16581500778429892,
      "learning_rate": 4.021012614850779e-06,
      "loss": 0.0234,
      "step": 7292
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.16963062318450156,
      "learning_rate": 4.016735209937741e-06,
      "loss": 0.0317,
      "step": 7293
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1468770676325282,
      "learning_rate": 4.012459827331841e-06,
      "loss": 0.014,
      "step": 7294
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1881485590171477,
      "learning_rate": 4.008186467574027e-06,
      "loss": 0.0206,
      "step": 7295
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2596238114885833,
      "learning_rate": 4.003915131204996e-06,
      "loss": 0.0442,
      "step": 7296
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.32493470893273096,
      "learning_rate": 3.999645818765176e-06,
      "loss": 0.0449,
      "step": 7297
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3892719103551836,
      "learning_rate": 3.995378530794754e-06,
      "loss": 0.0611,
      "step": 7298
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.18195816709510693,
      "learning_rate": 3.991113267833655e-06,
      "loss": 0.0242,
      "step": 7299
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3471964678219904,
      "learning_rate": 3.986850030421554e-06,
      "loss": 0.0393,
      "step": 7300
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.343517831947537,
      "learning_rate": 3.98258881909785e-06,
      "loss": 0.0379,
      "step": 7301
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20391452012519146,
      "learning_rate": 3.97832963440171e-06,
      "loss": 0.0374,
      "step": 7302
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.22616040788744537,
      "learning_rate": 3.974072476872028e-06,
      "loss": 0.0355,
      "step": 7303
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2003873694677798,
      "learning_rate": 3.969817347047451e-06,
      "loss": 0.0274,
      "step": 7304
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20780992721456482,
      "learning_rate": 3.965564245466365e-06,
      "loss": 0.0296,
      "step": 7305
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.28346236629823574,
      "learning_rate": 3.961313172666898e-06,
      "loss": 0.0352,
      "step": 7306
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21140964085241257,
      "learning_rate": 3.957064129186934e-06,
      "loss": 0.033,
      "step": 7307
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.16740479163637847,
      "learning_rate": 3.952817115564076e-06,
      "loss": 0.0144,
      "step": 7308
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2089287514505321,
      "learning_rate": 3.948572132335688e-06,
      "loss": 0.0236,
      "step": 7309
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20734197762428194,
      "learning_rate": 3.944329180038875e-06,
      "loss": 0.0413,
      "step": 7310
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23408757705707578,
      "learning_rate": 3.940088259210488e-06,
      "loss": 0.031,
      "step": 7311
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2880421104068135,
      "learning_rate": 3.935849370387104e-06,
      "loss": 0.0418,
      "step": 7312
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1979171644171094,
      "learning_rate": 3.931612514105061e-06,
      "loss": 0.028,
      "step": 7313
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1961261870619989,
      "learning_rate": 3.927377690900436e-06,
      "loss": 0.0289,
      "step": 7314
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20006336422846477,
      "learning_rate": 3.923144901309048e-06,
      "loss": 0.024,
      "step": 7315
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19950155679494494,
      "learning_rate": 3.91891414586645e-06,
      "loss": 0.041,
      "step": 7316
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.22234907209577182,
      "learning_rate": 3.914685425107949e-06,
      "loss": 0.0212,
      "step": 7317
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3115815494112536,
      "learning_rate": 3.91045873956859e-06,
      "loss": 0.0509,
      "step": 7318
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23922800929452773,
      "learning_rate": 3.906234089783165e-06,
      "loss": 0.0358,
      "step": 7319
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.30222051755695734,
      "learning_rate": 3.902011476286196e-06,
      "loss": 0.0419,
      "step": 7320
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23292987668024775,
      "learning_rate": 3.897790899611955e-06,
      "loss": 0.0367,
      "step": 7321
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19807583072420265,
      "learning_rate": 3.893572360294471e-06,
      "loss": 0.0311,
      "step": 7322
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.31309892957054214,
      "learning_rate": 3.889355858867487e-06,
      "loss": 0.0381,
      "step": 7323
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20455641917320452,
      "learning_rate": 3.885141395864509e-06,
      "loss": 0.015,
      "step": 7324
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.29357354363079663,
      "learning_rate": 3.880928971818776e-06,
      "loss": 0.043,
      "step": 7325
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21896055850765442,
      "learning_rate": 3.876718587263278e-06,
      "loss": 0.0324,
      "step": 7326
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.22726155004326407,
      "learning_rate": 3.872510242730727e-06,
      "loss": 0.0389,
      "step": 7327
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23160600516101507,
      "learning_rate": 3.868303938753599e-06,
      "loss": 0.0319,
      "step": 7328
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19124101785996905,
      "learning_rate": 3.864099675864104e-06,
      "loss": 0.0309,
      "step": 7329
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1465854042066699,
      "learning_rate": 3.859897454594192e-06,
      "loss": 0.0241,
      "step": 7330
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.25508352393706396,
      "learning_rate": 3.855697275475552e-06,
      "loss": 0.0393,
      "step": 7331
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21110589758356837,
      "learning_rate": 3.851499139039618e-06,
      "loss": 0.0306,
      "step": 7332
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.22957523547728229,
      "learning_rate": 3.84730304581757e-06,
      "loss": 0.0232,
      "step": 7333
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23355907036037143,
      "learning_rate": 3.843108996340323e-06,
      "loss": 0.0335,
      "step": 7334
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1494888168525127,
      "learning_rate": 3.838916991138542e-06,
      "loss": 0.0133,
      "step": 7335
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19305459479871245,
      "learning_rate": 3.834727030742613e-06,
      "loss": 0.0268,
      "step": 7336
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2578191467353125,
      "learning_rate": 3.830539115682685e-06,
      "loss": 0.0439,
      "step": 7337
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2616383585656577,
      "learning_rate": 3.826353246488641e-06,
      "loss": 0.0339,
      "step": 7338
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21180586096128393,
      "learning_rate": 3.822169423690103e-06,
      "loss": 0.0452,
      "step": 7339
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.24691313618041288,
      "learning_rate": 3.817987647816437e-06,
      "loss": 0.0415,
      "step": 7340
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3407690015058231,
      "learning_rate": 3.813807919396748e-06,
      "loss": 0.0497,
      "step": 7341
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2154006156776665,
      "learning_rate": 3.809630238959887e-06,
      "loss": 0.0342,
      "step": 7342
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26624987535070804,
      "learning_rate": 3.805454607034433e-06,
      "loss": 0.0364,
      "step": 7343
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2391253669205526,
      "learning_rate": 3.8012810241487175e-06,
      "loss": 0.0227,
      "step": 7344
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.17285098598405632,
      "learning_rate": 3.7971094908308105e-06,
      "loss": 0.0231,
      "step": 7345
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23947382926109162,
      "learning_rate": 3.7929400076085255e-06,
      "loss": 0.0488,
      "step": 7346
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3161221865630236,
      "learning_rate": 3.7887725750094027e-06,
      "loss": 0.0417,
      "step": 7347
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21156787282299797,
      "learning_rate": 3.7846071935607408e-06,
      "loss": 0.038,
      "step": 7348
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2264279426360328,
      "learning_rate": 3.780443863789569e-06,
      "loss": 0.0222,
      "step": 7349
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2283804159294777,
      "learning_rate": 3.7762825862226637e-06,
      "loss": 0.046,
      "step": 7350
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2900027371145845,
      "learning_rate": 3.772123361386526e-06,
      "loss": 0.0401,
      "step": 7351
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2998607595488389,
      "learning_rate": 3.767966189807415e-06,
      "loss": 0.045,
      "step": 7352
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.25326245190509966,
      "learning_rate": 3.763811072011325e-06,
      "loss": 0.0363,
      "step": 7353
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.22838346621131062,
      "learning_rate": 3.7596580085239897e-06,
      "loss": 0.033,
      "step": 7354
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2895025435049362,
      "learning_rate": 3.7555069998708726e-06,
      "loss": 0.046,
      "step": 7355
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20909757311382712,
      "learning_rate": 3.7513580465771893e-06,
      "loss": 0.0274,
      "step": 7356
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21307020749949826,
      "learning_rate": 3.7472111491679044e-06,
      "loss": 0.0374,
      "step": 7357
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.27735599974323055,
      "learning_rate": 3.7430663081676977e-06,
      "loss": 0.075,
      "step": 7358
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26025603903635497,
      "learning_rate": 3.7389235241010037e-06,
      "loss": 0.033,
      "step": 7359
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3056462316217732,
      "learning_rate": 3.734782797491998e-06,
      "loss": 0.0544,
      "step": 7360
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2198834516116484,
      "learning_rate": 3.7306441288645956e-06,
      "loss": 0.0339,
      "step": 7361
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26190651378520446,
      "learning_rate": 3.7265075187424373e-06,
      "loss": 0.024,
      "step": 7362
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2281713663952483,
      "learning_rate": 3.7223729676489196e-06,
      "loss": 0.0335,
      "step": 7363
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2019241120325882,
      "learning_rate": 3.7182404761071735e-06,
      "loss": 0.0348,
      "step": 7364
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2108367485041814,
      "learning_rate": 3.7141100446400733e-06,
      "loss": 0.0322,
      "step": 7365
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21884807874305998,
      "learning_rate": 3.7099816737702197e-06,
      "loss": 0.0328,
      "step": 7366
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.265129328110118,
      "learning_rate": 3.7058553640199635e-06,
      "loss": 0.0455,
      "step": 7367
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2188236844757285,
      "learning_rate": 3.7017311159113956e-06,
      "loss": 0.0207,
      "step": 7368
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.161858329798298,
      "learning_rate": 3.697608929966343e-06,
      "loss": 0.0232,
      "step": 7369
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.24779953226360146,
      "learning_rate": 3.6934888067063667e-06,
      "loss": 0.0365,
      "step": 7370
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20443720802617485,
      "learning_rate": 3.689370746652774e-06,
      "loss": 0.0382,
      "step": 7371
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1883988294255283,
      "learning_rate": 3.68525475032661e-06,
      "loss": 0.0261,
      "step": 7372
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26848847006348997,
      "learning_rate": 3.681140818248654e-06,
      "loss": 0.0293,
      "step": 7373
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20774241427230264,
      "learning_rate": 3.677028950939434e-06,
      "loss": 0.0243,
      "step": 7374
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.25525829092210983,
      "learning_rate": 3.6729191489192053e-06,
      "loss": 0.0348,
      "step": 7375
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20558128393706007,
      "learning_rate": 3.6688114127079665e-06,
      "loss": 0.0255,
      "step": 7376
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21004945530292993,
      "learning_rate": 3.664705742825463e-06,
      "loss": 0.0326,
      "step": 7377
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19785852372932963,
      "learning_rate": 3.6606021397911605e-06,
      "loss": 0.0304,
      "step": 7378
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2569349119992803,
      "learning_rate": 3.656500604124278e-06,
      "loss": 0.031,
      "step": 7379
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2174463681301021,
      "learning_rate": 3.652401136343768e-06,
      "loss": 0.0389,
      "step": 7380
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.22736589448447986,
      "learning_rate": 3.6483037369683284e-06,
      "loss": 0.0331,
      "step": 7381
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2689381306893739,
      "learning_rate": 3.6442084065163784e-06,
      "loss": 0.0509,
      "step": 7382
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26597423595588005,
      "learning_rate": 3.640115145506091e-06,
      "loss": 0.0264,
      "step": 7383
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21145613807014765,
      "learning_rate": 3.636023954455372e-06,
      "loss": 0.0372,
      "step": 7384
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2349149920663917,
      "learning_rate": 3.6319348338818718e-06,
      "loss": 0.0378,
      "step": 7385
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2400522860057657,
      "learning_rate": 3.6278477843029603e-06,
      "loss": 0.0387,
      "step": 7386
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1988599388286612,
      "learning_rate": 3.6237628062357643e-06,
      "loss": 0.0231,
      "step": 7387
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.25043973396510133,
      "learning_rate": 3.6196799001971416e-06,
      "loss": 0.0357,
      "step": 7388
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.15240011639637518,
      "learning_rate": 3.615599066703692e-06,
      "loss": 0.0271,
      "step": 7389
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.18042056195838546,
      "learning_rate": 3.6115203062717386e-06,
      "loss": 0.0284,
      "step": 7390
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4134745832494071,
      "learning_rate": 3.6074436194173635e-06,
      "loss": 0.064,
      "step": 7391
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2998669836078751,
      "learning_rate": 3.6033690066563765e-06,
      "loss": 0.0364,
      "step": 7392
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.24346497695265298,
      "learning_rate": 3.599296468504314e-06,
      "loss": 0.0383,
      "step": 7393
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21351807549274643,
      "learning_rate": 3.5952260054764663e-06,
      "loss": 0.0369,
      "step": 7394
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23191768390898695,
      "learning_rate": 3.591157618087855e-06,
      "loss": 0.0228,
      "step": 7395
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19185472031759665,
      "learning_rate": 3.5870913068532455e-06,
      "loss": 0.0329,
      "step": 7396
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21688971378862437,
      "learning_rate": 3.58302707228712e-06,
      "loss": 0.032,
      "step": 7397
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3060109908819706,
      "learning_rate": 3.5789649149037197e-06,
      "loss": 0.0374,
      "step": 7398
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2357773549950743,
      "learning_rate": 3.5749048352170146e-06,
      "loss": 0.0382,
      "step": 7399
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23406167544587667,
      "learning_rate": 3.5708468337407177e-06,
      "loss": 0.0286,
      "step": 7400
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16159346076328623,
      "learning_rate": 3.566790910988265e-06,
      "loss": 0.0246,
      "step": 7401
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2121497043887166,
      "learning_rate": 3.562737067472841e-06,
      "loss": 0.019,
      "step": 7402
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2355653970625761,
      "learning_rate": 3.5586853037073674e-06,
      "loss": 0.0347,
      "step": 7403
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1972603655576669,
      "learning_rate": 3.554635620204503e-06,
      "loss": 0.0345,
      "step": 7404
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20577147306361032,
      "learning_rate": 3.5505880174766305e-06,
      "loss": 0.0369,
      "step": 7405
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23267282834241032,
      "learning_rate": 3.546542496035883e-06,
      "loss": 0.042,
      "step": 7406
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2333554343985014,
      "learning_rate": 3.5424990563941265e-06,
      "loss": 0.027,
      "step": 7407
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2684093290030905,
      "learning_rate": 3.5384576990629672e-06,
      "loss": 0.0327,
      "step": 7408
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2215673102998656,
      "learning_rate": 3.5344184245537384e-06,
      "loss": 0.0608,
      "step": 7409
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.24675505959812644,
      "learning_rate": 3.53038123337752e-06,
      "loss": 0.042,
      "step": 7410
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.17952090294139986,
      "learning_rate": 3.5263461260451217e-06,
      "loss": 0.029,
      "step": 7411
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3008566922781658,
      "learning_rate": 3.5223131030670942e-06,
      "loss": 0.0754,
      "step": 7412
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19868479023309843,
      "learning_rate": 3.5182821649537192e-06,
      "loss": 0.045,
      "step": 7413
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19470725695566066,
      "learning_rate": 3.5142533122150147e-06,
      "loss": 0.0293,
      "step": 7414
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.31727193216020894,
      "learning_rate": 3.510226545360742e-06,
      "loss": 0.0347,
      "step": 7415
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23468189809338336,
      "learning_rate": 3.506201864900396e-06,
      "loss": 0.0374,
      "step": 7416
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.24536788703920037,
      "learning_rate": 3.502179271343198e-06,
      "loss": 0.0425,
      "step": 7417
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.17150679074681038,
      "learning_rate": 3.4981587651981185e-06,
      "loss": 0.0251,
      "step": 7418
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21906352565323928,
      "learning_rate": 3.4941403469738555e-06,
      "loss": 0.0352,
      "step": 7419
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2401544081171172,
      "learning_rate": 3.490124017178851e-06,
      "loss": 0.037,
      "step": 7420
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26761707662820367,
      "learning_rate": 3.4861097763212693e-06,
      "loss": 0.0372,
      "step": 7421
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1894305107541527,
      "learning_rate": 3.482097624909022e-06,
      "loss": 0.0409,
      "step": 7422
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2490980597493976,
      "learning_rate": 3.4780875634497526e-06,
      "loss": 0.0382,
      "step": 7423
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.25547108780377437,
      "learning_rate": 3.474079592450845e-06,
      "loss": 0.0293,
      "step": 7424
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22409663347392775,
      "learning_rate": 3.4700737124194015e-06,
      "loss": 0.0305,
      "step": 7425
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19892686142251298,
      "learning_rate": 3.466069923862283e-06,
      "loss": 0.0258,
      "step": 7426
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24378036621651403,
      "learning_rate": 3.4620682272860772e-06,
      "loss": 0.0448,
      "step": 7427
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.21739847036010437,
      "learning_rate": 3.458068623197097e-06,
      "loss": 0.0289,
      "step": 7428
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23974004143497032,
      "learning_rate": 3.4540711121013982e-06,
      "loss": 0.027,
      "step": 7429
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1887085825592145,
      "learning_rate": 3.4500756945047774e-06,
      "loss": 0.0174,
      "step": 7430
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.16868693930711878,
      "learning_rate": 3.4460823709127645e-06,
      "loss": 0.0166,
      "step": 7431
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2538717246424224,
      "learning_rate": 3.442091141830608e-06,
      "loss": 0.0419,
      "step": 7432
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.17340388659641845,
      "learning_rate": 3.438102007763313e-06,
      "loss": 0.0328,
      "step": 7433
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.27920512046060325,
      "learning_rate": 3.4341149692156074e-06,
      "loss": 0.035,
      "step": 7434
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19804177700907516,
      "learning_rate": 3.430130026691967e-06,
      "loss": 0.0361,
      "step": 7435
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1959560663227455,
      "learning_rate": 3.426147180696577e-06,
      "loss": 0.0413,
      "step": 7436
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2288795003136039,
      "learning_rate": 3.4221664317333826e-06,
      "loss": 0.0333,
      "step": 7437
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24211084783302425,
      "learning_rate": 3.4181877803060528e-06,
      "loss": 0.0322,
      "step": 7438
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.28184957966927965,
      "learning_rate": 3.4142112269179962e-06,
      "loss": 0.0371,
      "step": 7439
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.270051523144888,
      "learning_rate": 3.4102367720723438e-06,
      "loss": 0.0397,
      "step": 7440
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24397744638289665,
      "learning_rate": 3.406264416271976e-06,
      "loss": 0.04,
      "step": 7441
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23206830579746976,
      "learning_rate": 3.402294160019499e-06,
      "loss": 0.0377,
      "step": 7442
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2256968007005466,
      "learning_rate": 3.398326003817256e-06,
      "loss": 0.0368,
      "step": 7443
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3031928252014159,
      "learning_rate": 3.394359948167325e-06,
      "loss": 0.0544,
      "step": 7444
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.268016525658452,
      "learning_rate": 3.390395993571516e-06,
      "loss": 0.0509,
      "step": 7445
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.18703540223754123,
      "learning_rate": 3.386434140531378e-06,
      "loss": 0.0207,
      "step": 7446
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23181738903786422,
      "learning_rate": 3.382474389548185e-06,
      "loss": 0.0327,
      "step": 7447
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2192405921457413,
      "learning_rate": 3.3785167411229523e-06,
      "loss": 0.0524,
      "step": 7448
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20525408870490355,
      "learning_rate": 3.3745611957564273e-06,
      "loss": 0.0263,
      "step": 7449
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23466835736303263,
      "learning_rate": 3.3706077539490933e-06,
      "loss": 0.0342,
      "step": 7450
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3167582543629114,
      "learning_rate": 3.3666564162011705e-06,
      "loss": 0.0383,
      "step": 7451
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2404412874945336,
      "learning_rate": 3.362707183012597e-06,
      "loss": 0.0336,
      "step": 7452
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2375760822402757,
      "learning_rate": 3.358760054883059e-06,
      "loss": 0.0298,
      "step": 7453
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23063965296005837,
      "learning_rate": 3.354815032311978e-06,
      "loss": 0.0441,
      "step": 7454
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2323894478111432,
      "learning_rate": 3.3508721157985023e-06,
      "loss": 0.0381,
      "step": 7455
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20004185633467292,
      "learning_rate": 3.34693130584151e-06,
      "loss": 0.0184,
      "step": 7456
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22864691720298666,
      "learning_rate": 3.342992602939623e-06,
      "loss": 0.04,
      "step": 7457
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19791542333078518,
      "learning_rate": 3.3390560075911906e-06,
      "loss": 0.0209,
      "step": 7458
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.165287767175405,
      "learning_rate": 3.3351215202942954e-06,
      "loss": 0.0257,
      "step": 7459
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.15579203008774456,
      "learning_rate": 3.331189141546758e-06,
      "loss": 0.0224,
      "step": 7460
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.27487803160247687,
      "learning_rate": 3.327258871846126e-06,
      "loss": 0.0529,
      "step": 7461
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23277987629217148,
      "learning_rate": 3.3233307116896874e-06,
      "loss": 0.0461,
      "step": 7462
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.17555566724108546,
      "learning_rate": 3.3194046615744503e-06,
      "loss": 0.0207,
      "step": 7463
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23009321360827986,
      "learning_rate": 3.3154807219971684e-06,
      "loss": 0.0368,
      "step": 7464
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2378087909925564,
      "learning_rate": 3.311558893454323e-06,
      "loss": 0.0348,
      "step": 7465
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.34445023741174724,
      "learning_rate": 3.307639176442137e-06,
      "loss": 0.0427,
      "step": 7466
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22513745467064952,
      "learning_rate": 3.3037215714565485e-06,
      "loss": 0.0302,
      "step": 7467
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20298962483529867,
      "learning_rate": 3.299806078993242e-06,
      "loss": 0.021,
      "step": 7468
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19793103609290041,
      "learning_rate": 3.295892699547629e-06,
      "loss": 0.0294,
      "step": 7469
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.16142332302467222,
      "learning_rate": 3.2919814336148657e-06,
      "loss": 0.0187,
      "step": 7470
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.193968401376796,
      "learning_rate": 3.288072281689818e-06,
      "loss": 0.0308,
      "step": 7471
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20980708673270493,
      "learning_rate": 3.2841652442671033e-06,
      "loss": 0.0284,
      "step": 7472
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24221748504830057,
      "learning_rate": 3.2802603218410666e-06,
      "loss": 0.0451,
      "step": 7473
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.21741787575413205,
      "learning_rate": 3.276357514905788e-06,
      "loss": 0.0264,
      "step": 7474
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22864696608127394,
      "learning_rate": 3.2724568239550637e-06,
      "loss": 0.0378,
      "step": 7475
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19013485704322677,
      "learning_rate": 3.2685582494824386e-06,
      "loss": 0.0318,
      "step": 7476
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22356170456648536,
      "learning_rate": 3.2646617919812007e-06,
      "loss": 0.0217,
      "step": 7477
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2590321532721641,
      "learning_rate": 3.260767451944338e-06,
      "loss": 0.0565,
      "step": 7478
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23676297873075064,
      "learning_rate": 3.256875229864591e-06,
      "loss": 0.0287,
      "step": 7479
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20357256877969448,
      "learning_rate": 3.252985126234434e-06,
      "loss": 0.0409,
      "step": 7480
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.17218293975807653,
      "learning_rate": 3.2490971415460716e-06,
      "loss": 0.037,
      "step": 7481
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.21439589725270158,
      "learning_rate": 3.245211276291427e-06,
      "loss": 0.0276,
      "step": 7482
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.18591843595465093,
      "learning_rate": 3.2413275309621684e-06,
      "loss": 0.0366,
      "step": 7483
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2716762221641217,
      "learning_rate": 3.237445906049694e-06,
      "loss": 0.0605,
      "step": 7484
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2515395266545232,
      "learning_rate": 3.2335664020451364e-06,
      "loss": 0.0331,
      "step": 7485
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2701404018177225,
      "learning_rate": 3.229689019439348e-06,
      "loss": 0.0402,
      "step": 7486
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1983329490015436,
      "learning_rate": 3.2258137587229242e-06,
      "loss": 0.0225,
      "step": 7487
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2545477905191409,
      "learning_rate": 3.2219406203861903e-06,
      "loss": 0.0509,
      "step": 7488
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1797385350604584,
      "learning_rate": 3.2180696049192027e-06,
      "loss": 0.032,
      "step": 7489
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2884899474637145,
      "learning_rate": 3.2142007128117393e-06,
      "loss": 0.0308,
      "step": 7490
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2373232265185134,
      "learning_rate": 3.2103339445533233e-06,
      "loss": 0.0301,
      "step": 7491
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2685669219958205,
      "learning_rate": 3.2064693006332013e-06,
      "loss": 0.0342,
      "step": 7492
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23162935882497898,
      "learning_rate": 3.202606781540356e-06,
      "loss": 0.0248,
      "step": 7493
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2517438655331266,
      "learning_rate": 3.1987463877634962e-06,
      "loss": 0.0254,
      "step": 7494
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2219837305115061,
      "learning_rate": 3.1948881197910664e-06,
      "loss": 0.0286,
      "step": 7495
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2147771211824228,
      "learning_rate": 3.1910319781112364e-06,
      "loss": 0.0263,
      "step": 7496
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2103318951691591,
      "learning_rate": 3.1871779632119204e-06,
      "loss": 0.0402,
      "step": 7497
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.21288991630506235,
      "learning_rate": 3.1833260755807392e-06,
      "loss": 0.0247,
      "step": 7498
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22674921168954712,
      "learning_rate": 3.179476315705068e-06,
      "loss": 0.0359,
      "step": 7499
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1830833319952086,
      "learning_rate": 3.1756286840719987e-06,
      "loss": 0.0338,
      "step": 7500
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2321710836298324,
      "learning_rate": 3.171783181168366e-06,
      "loss": 0.0411,
      "step": 7501
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24267933495604038,
      "learning_rate": 3.16793980748072e-06,
      "loss": 0.0473,
      "step": 7502
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2721682791815918,
      "learning_rate": 3.1640985634953524e-06,
      "loss": 0.0358,
      "step": 7503
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2524557496945113,
      "learning_rate": 3.160259449698282e-06,
      "loss": 0.0373,
      "step": 7504
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2315410464306356,
      "learning_rate": 3.156422466575264e-06,
      "loss": 0.0325,
      "step": 7505
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20002625933754795,
      "learning_rate": 3.1525876146117707e-06,
      "loss": 0.041,
      "step": 7506
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19547227046732799,
      "learning_rate": 3.1487548942930136e-06,
      "loss": 0.0304,
      "step": 7507
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.216989517538845,
      "learning_rate": 3.144924306103938e-06,
      "loss": 0.0257,
      "step": 7508
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22981897179073466,
      "learning_rate": 3.1410958505292167e-06,
      "loss": 0.0424,
      "step": 7509
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.16953086809036202,
      "learning_rate": 3.1372695280532415e-06,
      "loss": 0.0304,
      "step": 7510
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2382266967110526,
      "learning_rate": 3.133445339160146e-06,
      "loss": 0.0428,
      "step": 7511
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19943456553718433,
      "learning_rate": 3.129623284333805e-06,
      "loss": 0.0421,
      "step": 7512
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20832149154069063,
      "learning_rate": 3.125803364057793e-06,
      "loss": 0.0325,
      "step": 7513
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22556063687682615,
      "learning_rate": 3.1219855788154385e-06,
      "loss": 0.0356,
      "step": 7514
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2795119624469679,
      "learning_rate": 3.1181699290897916e-06,
      "loss": 0.0313,
      "step": 7515
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1930770932810805,
      "learning_rate": 3.1143564153636395e-06,
      "loss": 0.0155,
      "step": 7516
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18146117380249757,
      "learning_rate": 3.1105450381194835e-06,
      "loss": 0.0312,
      "step": 7517
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20320885101311342,
      "learning_rate": 3.1067357978395663e-06,
      "loss": 0.0302,
      "step": 7518
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.15815770709985794,
      "learning_rate": 3.102928695005858e-06,
      "loss": 0.019,
      "step": 7519
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18898936662078236,
      "learning_rate": 3.0991237301000664e-06,
      "loss": 0.0284,
      "step": 7520
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2842840809156435,
      "learning_rate": 3.095320903603607e-06,
      "loss": 0.0312,
      "step": 7521
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2289755830147388,
      "learning_rate": 3.0915202159976453e-06,
      "loss": 0.0381,
      "step": 7522
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22242539132718261,
      "learning_rate": 3.0877216677630683e-06,
      "loss": 0.0316,
      "step": 7523
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1967054098331257,
      "learning_rate": 3.083925259380498e-06,
      "loss": 0.0209,
      "step": 7524
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23675961157513783,
      "learning_rate": 3.0801309913302724e-06,
      "loss": 0.0334,
      "step": 7525
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.24298225254919875,
      "learning_rate": 3.0763388640924698e-06,
      "loss": 0.0406,
      "step": 7526
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2297374875305951,
      "learning_rate": 3.072548878146897e-06,
      "loss": 0.042,
      "step": 7527
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2996989010739473,
      "learning_rate": 3.068761033973087e-06,
      "loss": 0.0437,
      "step": 7528
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2539570097436333,
      "learning_rate": 3.0649753320503017e-06,
      "loss": 0.0362,
      "step": 7529
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.24119700650283885,
      "learning_rate": 3.0611917728575347e-06,
      "loss": 0.0382,
      "step": 7530
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22393550128028836,
      "learning_rate": 3.057410356873505e-06,
      "loss": 0.0316,
      "step": 7531
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20312731081308394,
      "learning_rate": 3.053631084576667e-06,
      "loss": 0.0313,
      "step": 7532
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.26764594562067223,
      "learning_rate": 3.04985395644519e-06,
      "loss": 0.0439,
      "step": 7533
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23299971643981132,
      "learning_rate": 3.046078972956985e-06,
      "loss": 0.0368,
      "step": 7534
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.15867413353451265,
      "learning_rate": 3.0423061345896875e-06,
      "loss": 0.0188,
      "step": 7535
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2258630017174461,
      "learning_rate": 3.038535441820669e-06,
      "loss": 0.0274,
      "step": 7536
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18832984708693212,
      "learning_rate": 3.0347668951270084e-06,
      "loss": 0.0273,
      "step": 7537
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18064938047394155,
      "learning_rate": 3.0310004949855366e-06,
      "loss": 0.0318,
      "step": 7538
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2710197919501598,
      "learning_rate": 3.0272362418727972e-06,
      "loss": 0.0473,
      "step": 7539
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.17966238136531762,
      "learning_rate": 3.0234741362650787e-06,
      "loss": 0.0228,
      "step": 7540
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22604164729286733,
      "learning_rate": 3.019714178638373e-06,
      "loss": 0.0649,
      "step": 7541
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2817679378034041,
      "learning_rate": 3.0159563694684245e-06,
      "loss": 0.0449,
      "step": 7542
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.24425577118614047,
      "learning_rate": 3.0122007092306903e-06,
      "loss": 0.0358,
      "step": 7543
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18778027011985785,
      "learning_rate": 3.008447198400368e-06,
      "loss": 0.0158,
      "step": 7544
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1737640455627145,
      "learning_rate": 3.004695837452365e-06,
      "loss": 0.0308,
      "step": 7545
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1736311814839811,
      "learning_rate": 3.0009466268613384e-06,
      "loss": 0.0303,
      "step": 7546
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23909056131063683,
      "learning_rate": 2.997199567101663e-06,
      "loss": 0.0349,
      "step": 7547
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2511854378700324,
      "learning_rate": 2.9934546586474346e-06,
      "loss": 0.032,
      "step": 7548
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.16832324257631084,
      "learning_rate": 2.989711901972485e-06,
      "loss": 0.0216,
      "step": 7549
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2061441236453667,
      "learning_rate": 2.985971297550374e-06,
      "loss": 0.0233,
      "step": 7550
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18699323199259152,
      "learning_rate": 2.9822328458543916e-06,
      "loss": 0.0264,
      "step": 7551
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.15171133112619387,
      "learning_rate": 2.9784965473575434e-06,
      "loss": 0.0258,
      "step": 7552
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1764306681841002,
      "learning_rate": 2.9747624025325716e-06,
      "loss": 0.0218,
      "step": 7553
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1893227118517692,
      "learning_rate": 2.9710304118519473e-06,
      "loss": 0.0287,
      "step": 7554
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.29197093159039084,
      "learning_rate": 2.9673005757878703e-06,
      "loss": 0.0198,
      "step": 7555
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23489263117201273,
      "learning_rate": 2.9635728948122542e-06,
      "loss": 0.0364,
      "step": 7556
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.16448805884237272,
      "learning_rate": 2.9598473693967555e-06,
      "loss": 0.0287,
      "step": 7557
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.28402562897767586,
      "learning_rate": 2.95612400001275e-06,
      "loss": 0.0464,
      "step": 7558
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.21629984474141284,
      "learning_rate": 2.9524027871313476e-06,
      "loss": 0.0318,
      "step": 7559
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19412633240740532,
      "learning_rate": 2.9486837312233742e-06,
      "loss": 0.0224,
      "step": 7560
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.25963675378412404,
      "learning_rate": 2.9449668327593885e-06,
      "loss": 0.0431,
      "step": 7561
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20437587209976266,
      "learning_rate": 2.9412520922096834e-06,
      "loss": 0.0308,
      "step": 7562
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22413144056583822,
      "learning_rate": 2.937539510044265e-06,
      "loss": 0.0415,
      "step": 7563
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.24400913512315092,
      "learning_rate": 2.93382908673288e-06,
      "loss": 0.0297,
      "step": 7564
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2342546551408778,
      "learning_rate": 2.9301208227449905e-06,
      "loss": 0.0329,
      "step": 7565
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36450597986959676,
      "learning_rate": 2.926414718549797e-06,
      "loss": 0.0545,
      "step": 7566
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1706969790367068,
      "learning_rate": 2.922710774616211e-06,
      "loss": 0.0314,
      "step": 7567
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3037674412802316,
      "learning_rate": 2.9190089914128837e-06,
      "loss": 0.0437,
      "step": 7568
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3183951211531267,
      "learning_rate": 2.91530936940819e-06,
      "loss": 0.0392,
      "step": 7569
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19752672120297818,
      "learning_rate": 2.911611909070229e-06,
      "loss": 0.0246,
      "step": 7570
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2366123696929768,
      "learning_rate": 2.90791661086683e-06,
      "loss": 0.036,
      "step": 7571
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1861685727523322,
      "learning_rate": 2.9042234752655417e-06,
      "loss": 0.0169,
      "step": 7572
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1817812426591504,
      "learning_rate": 2.9005325027336482e-06,
      "loss": 0.0217,
      "step": 7573
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2282377983166106,
      "learning_rate": 2.8968436937381515e-06,
      "loss": 0.0377,
      "step": 7574
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.21570972699179414,
      "learning_rate": 2.8931570487457894e-06,
      "loss": 0.0331,
      "step": 7575
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22507620686511837,
      "learning_rate": 2.889472568223015e-06,
      "loss": 0.0343,
      "step": 7576
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2541962801118243,
      "learning_rate": 2.8857902526360114e-06,
      "loss": 0.0328,
      "step": 7577
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4486091125897802,
      "learning_rate": 2.8821101024506947e-06,
      "loss": 0.0504,
      "step": 7578
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22870920410281215,
      "learning_rate": 2.8784321181327035e-06,
      "loss": 0.0283,
      "step": 7579
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.25703863115930436,
      "learning_rate": 2.874756300147388e-06,
      "loss": 0.032,
      "step": 7580
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3016914862667133,
      "learning_rate": 2.8710826489598485e-06,
      "loss": 0.0499,
      "step": 7581
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2809703018931384,
      "learning_rate": 2.867411165034901e-06,
      "loss": 0.0406,
      "step": 7582
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.28629385676763286,
      "learning_rate": 2.863741848837074e-06,
      "loss": 0.0274,
      "step": 7583
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.24941785655649665,
      "learning_rate": 2.8600747008306417e-06,
      "loss": 0.0382,
      "step": 7584
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23546679829457529,
      "learning_rate": 2.8564097214795937e-06,
      "loss": 0.0349,
      "step": 7585
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.21802023515438493,
      "learning_rate": 2.8527469112476524e-06,
      "loss": 0.0291,
      "step": 7586
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2033274631910137,
      "learning_rate": 2.849086270598249e-06,
      "loss": 0.0207,
      "step": 7587
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18674789025750416,
      "learning_rate": 2.8454277999945603e-06,
      "loss": 0.023,
      "step": 7588
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3043528944757433,
      "learning_rate": 2.841771499899475e-06,
      "loss": 0.0432,
      "step": 7589
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20592386151362962,
      "learning_rate": 2.8381173707756214e-06,
      "loss": 0.0307,
      "step": 7590
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.27418265308907297,
      "learning_rate": 2.834465413085332e-06,
      "loss": 0.0398,
      "step": 7591
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1363660076596249,
      "learning_rate": 2.8308156272906794e-06,
      "loss": 0.0171,
      "step": 7592
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2132512895968844,
      "learning_rate": 2.827168013853463e-06,
      "loss": 0.0492,
      "step": 7593
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1797707717172014,
      "learning_rate": 2.8235225732352043e-06,
      "loss": 0.0261,
      "step": 7594
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20851552367967222,
      "learning_rate": 2.8198793058971397e-06,
      "loss": 0.014,
      "step": 7595
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23987911286736752,
      "learning_rate": 2.8162382123002418e-06,
      "loss": 0.0346,
      "step": 7596
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.29634870759911625,
      "learning_rate": 2.8125992929052092e-06,
      "loss": 0.0336,
      "step": 7597
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1956021830929876,
      "learning_rate": 2.8089625481724604e-06,
      "loss": 0.0274,
      "step": 7598
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18066717609093813,
      "learning_rate": 2.8053279785621378e-06,
      "loss": 0.0262,
      "step": 7599
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2350615618097985,
      "learning_rate": 2.8016955845341143e-06,
      "loss": 0.0277,
      "step": 7600
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19820320046693707,
      "learning_rate": 2.798065366547986e-06,
      "loss": 0.0256,
      "step": 7601
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2951730833004121,
      "learning_rate": 2.794437325063064e-06,
      "loss": 0.0366,
      "step": 7602
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1948995611839091,
      "learning_rate": 2.7908114605383963e-06,
      "loss": 0.0238,
      "step": 7603
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2213159004495298,
      "learning_rate": 2.7871877734327514e-06,
      "loss": 0.0408,
      "step": 7604
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2121654460974383,
      "learning_rate": 2.7835662642046245e-06,
      "loss": 0.0269,
      "step": 7605
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24670389015909733,
      "learning_rate": 2.7799469333122275e-06,
      "loss": 0.0357,
      "step": 7606
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21498507273219974,
      "learning_rate": 2.7763297812135026e-06,
      "loss": 0.046,
      "step": 7607
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.23506955706422913,
      "learning_rate": 2.772714808366115e-06,
      "loss": 0.0329,
      "step": 7608
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21770459955989604,
      "learning_rate": 2.7691020152274585e-06,
      "loss": 0.0331,
      "step": 7609
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.20548492230201848,
      "learning_rate": 2.76549140225465e-06,
      "loss": 0.041,
      "step": 7610
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2511533715535377,
      "learning_rate": 2.761882969904517e-06,
      "loss": 0.0328,
      "step": 7611
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.27096677054455504,
      "learning_rate": 2.758276718633628e-06,
      "loss": 0.0363,
      "step": 7612
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19968950525808227,
      "learning_rate": 2.754672648898271e-06,
      "loss": 0.0291,
      "step": 7613
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.20565001357331475,
      "learning_rate": 2.751070761154453e-06,
      "loss": 0.0224,
      "step": 7614
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2205976217156827,
      "learning_rate": 2.747471055857913e-06,
      "loss": 0.0258,
      "step": 7615
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2736638630819423,
      "learning_rate": 2.743873533464105e-06,
      "loss": 0.0389,
      "step": 7616
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2977077824997038,
      "learning_rate": 2.7402781944282164e-06,
      "loss": 0.0353,
      "step": 7617
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19000655873485214,
      "learning_rate": 2.7366850392051468e-06,
      "loss": 0.0229,
      "step": 7618
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.34686368975371545,
      "learning_rate": 2.733094068249527e-06,
      "loss": 0.0428,
      "step": 7619
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.25636586442495835,
      "learning_rate": 2.7295052820157097e-06,
      "loss": 0.0302,
      "step": 7620
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2837643029057331,
      "learning_rate": 2.72591868095778e-06,
      "loss": 0.0397,
      "step": 7621
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.25689548628855835,
      "learning_rate": 2.722334265529527e-06,
      "loss": 0.0335,
      "step": 7622
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2098684861437258,
      "learning_rate": 2.7187520361844776e-06,
      "loss": 0.028,
      "step": 7623
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18787319078672018,
      "learning_rate": 2.715171993375878e-06,
      "loss": 0.0262,
      "step": 7624
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21804441177961417,
      "learning_rate": 2.711594137556708e-06,
      "loss": 0.0323,
      "step": 7625
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.32105543460241515,
      "learning_rate": 2.7080184691796474e-06,
      "loss": 0.0345,
      "step": 7626
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22500249212262394,
      "learning_rate": 2.7044449886971213e-06,
      "loss": 0.03,
      "step": 7627
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17858965985010777,
      "learning_rate": 2.7008736965612658e-06,
      "loss": 0.0115,
      "step": 7628
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17528832517121057,
      "learning_rate": 2.6973045932239526e-06,
      "loss": 0.0166,
      "step": 7629
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3125683471324996,
      "learning_rate": 2.6937376791367566e-06,
      "loss": 0.0476,
      "step": 7630
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2124116917449492,
      "learning_rate": 2.6901729547509934e-06,
      "loss": 0.0366,
      "step": 7631
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22591288105336194,
      "learning_rate": 2.6866104205176925e-06,
      "loss": 0.0379,
      "step": 7632
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.23916193539700292,
      "learning_rate": 2.683050076887612e-06,
      "loss": 0.0328,
      "step": 7633
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18496191272256948,
      "learning_rate": 2.679491924311226e-06,
      "loss": 0.0222,
      "step": 7634
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2786003042916692,
      "learning_rate": 2.6759359632387407e-06,
      "loss": 0.0396,
      "step": 7635
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.20074704134852306,
      "learning_rate": 2.67238219412008e-06,
      "loss": 0.0271,
      "step": 7636
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3189146897331374,
      "learning_rate": 2.6688306174048807e-06,
      "loss": 0.059,
      "step": 7637
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2278434371218837,
      "learning_rate": 2.6652812335425184e-06,
      "loss": 0.0238,
      "step": 7638
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2388356591141597,
      "learning_rate": 2.6617340429820825e-06,
      "loss": 0.05,
      "step": 7639
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.20810065190054117,
      "learning_rate": 2.6581890461723925e-06,
      "loss": 0.0329,
      "step": 7640
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1884146573881054,
      "learning_rate": 2.6546462435619755e-06,
      "loss": 0.0381,
      "step": 7641
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.23376778782262872,
      "learning_rate": 2.651105635599094e-06,
      "loss": 0.0278,
      "step": 7642
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2767728113647652,
      "learning_rate": 2.6475672227317282e-06,
      "loss": 0.047,
      "step": 7643
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2088890214662041,
      "learning_rate": 2.6440310054075877e-06,
      "loss": 0.0387,
      "step": 7644
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22184546401655003,
      "learning_rate": 2.6404969840740858e-06,
      "loss": 0.0385,
      "step": 7645
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18417381406704897,
      "learning_rate": 2.6369651591783774e-06,
      "loss": 0.0252,
      "step": 7646
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21899323904474913,
      "learning_rate": 2.633435531167332e-06,
      "loss": 0.0319,
      "step": 7647
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.28716727730382985,
      "learning_rate": 2.629908100487544e-06,
      "loss": 0.0588,
      "step": 7648
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24797952293257808,
      "learning_rate": 2.626382867585313e-06,
      "loss": 0.0387,
      "step": 7649
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1657360471621508,
      "learning_rate": 2.6228598329066902e-06,
      "loss": 0.0319,
      "step": 7650
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.27341377972851105,
      "learning_rate": 2.619338996897427e-06,
      "loss": 0.0242,
      "step": 7651
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21904022684477306,
      "learning_rate": 2.6158203600030076e-06,
      "loss": 0.0263,
      "step": 7652
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.20140843232646816,
      "learning_rate": 2.612303922668624e-06,
      "loss": 0.0178,
      "step": 7653
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22433960511910805,
      "learning_rate": 2.6087896853392037e-06,
      "loss": 0.0364,
      "step": 7654
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2466762853585662,
      "learning_rate": 2.60527764845939e-06,
      "loss": 0.0291,
      "step": 7655
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22148049433569833,
      "learning_rate": 2.6017678124735545e-06,
      "loss": 0.0281,
      "step": 7656
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22832268941276734,
      "learning_rate": 2.5982601778257733e-06,
      "loss": 0.0394,
      "step": 7657
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24904132853546684,
      "learning_rate": 2.594754744959862e-06,
      "loss": 0.0331,
      "step": 7658
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.31715142838897503,
      "learning_rate": 2.5912515143193506e-06,
      "loss": 0.0603,
      "step": 7659
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3282299101100885,
      "learning_rate": 2.5877504863474933e-06,
      "loss": 0.049,
      "step": 7660
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.23427066069941918,
      "learning_rate": 2.584251661487258e-06,
      "loss": 0.0378,
      "step": 7661
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17040282181747796,
      "learning_rate": 2.58075504018134e-06,
      "loss": 0.0202,
      "step": 7662
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3024061577234958,
      "learning_rate": 2.5772606228721555e-06,
      "loss": 0.057,
      "step": 7663
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24001482879256286,
      "learning_rate": 2.5737684100018446e-06,
      "loss": 0.0492,
      "step": 7664
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24036648072099676,
      "learning_rate": 2.5702784020122582e-06,
      "loss": 0.0441,
      "step": 7665
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1883154956650253,
      "learning_rate": 2.566790599344973e-06,
      "loss": 0.023,
      "step": 7666
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3097774522599185,
      "learning_rate": 2.563305002441303e-06,
      "loss": 0.0357,
      "step": 7667
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24300508012877306,
      "learning_rate": 2.5598216117422547e-06,
      "loss": 0.035,
      "step": 7668
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2094741945721532,
      "learning_rate": 2.556340427688575e-06,
      "loss": 0.027,
      "step": 7669
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2227773169926037,
      "learning_rate": 2.552861450720725e-06,
      "loss": 0.0341,
      "step": 7670
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.13619264467246087,
      "learning_rate": 2.5493846812788923e-06,
      "loss": 0.0108,
      "step": 7671
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.26831786377417394,
      "learning_rate": 2.5459101198029724e-06,
      "loss": 0.0313,
      "step": 7672
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.15257217326277978,
      "learning_rate": 2.542437766732593e-06,
      "loss": 0.0205,
      "step": 7673
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21841776686397005,
      "learning_rate": 2.538967622507098e-06,
      "loss": 0.0305,
      "step": 7674
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2131413025088449,
      "learning_rate": 2.535499687565559e-06,
      "loss": 0.0283,
      "step": 7675
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.23409968736477768,
      "learning_rate": 2.532033962346754e-06,
      "loss": 0.0336,
      "step": 7676
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21867116290838887,
      "learning_rate": 2.528570447289189e-06,
      "loss": 0.0279,
      "step": 7677
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.265822505451381,
      "learning_rate": 2.525109142831095e-06,
      "loss": 0.0374,
      "step": 7678
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19859364553849468,
      "learning_rate": 2.52165004941042e-06,
      "loss": 0.0274,
      "step": 7679
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22097252127108136,
      "learning_rate": 2.5181931674648265e-06,
      "loss": 0.023,
      "step": 7680
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2090036971527243,
      "learning_rate": 2.5147384974317014e-06,
      "loss": 0.0246,
      "step": 7681
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17914124052438576,
      "learning_rate": 2.5112860397481553e-06,
      "loss": 0.0385,
      "step": 7682
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36674578456107254,
      "learning_rate": 2.507835794851017e-06,
      "loss": 0.0741,
      "step": 7683
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22445193038447256,
      "learning_rate": 2.50438776317683e-06,
      "loss": 0.0263,
      "step": 7684
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21830122758653003,
      "learning_rate": 2.5009419451618634e-06,
      "loss": 0.0351,
      "step": 7685
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21726627910780288,
      "learning_rate": 2.497498341242104e-06,
      "loss": 0.0373,
      "step": 7686
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2314400330275358,
      "learning_rate": 2.4940569518532677e-06,
      "loss": 0.0371,
      "step": 7687
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24373962276328054,
      "learning_rate": 2.490617777430766e-06,
      "loss": 0.029,
      "step": 7688
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17332051666399056,
      "learning_rate": 2.4871808184097558e-06,
      "loss": 0.0195,
      "step": 7689
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2083561626482775,
      "learning_rate": 2.4837460752251e-06,
      "loss": 0.036,
      "step": 7690
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.23903304460858993,
      "learning_rate": 2.4803135483113903e-06,
      "loss": 0.0328,
      "step": 7691
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19917820536601982,
      "learning_rate": 2.476883238102925e-06,
      "loss": 0.0344,
      "step": 7692
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.40944573759486974,
      "learning_rate": 2.473455145033734e-06,
      "loss": 0.064,
      "step": 7693
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.23786641578700735,
      "learning_rate": 2.4700292695375596e-06,
      "loss": 0.0293,
      "step": 7694
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23913851508051295,
      "learning_rate": 2.46660561204787e-06,
      "loss": 0.0426,
      "step": 7695
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.26505084254061295,
      "learning_rate": 2.4631841729978435e-06,
      "loss": 0.0356,
      "step": 7696
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23045340988055701,
      "learning_rate": 2.459764952820385e-06,
      "loss": 0.0439,
      "step": 7697
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1685558968828825,
      "learning_rate": 2.456347951948115e-06,
      "loss": 0.0213,
      "step": 7698
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2698645811097252,
      "learning_rate": 2.452933170813383e-06,
      "loss": 0.0522,
      "step": 7699
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2933298809803834,
      "learning_rate": 2.449520609848237e-06,
      "loss": 0.0268,
      "step": 7700
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2824123787239722,
      "learning_rate": 2.4461102694844605e-06,
      "loss": 0.0286,
      "step": 7701
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2424009828112054,
      "learning_rate": 2.442702150153562e-06,
      "loss": 0.038,
      "step": 7702
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2536387868812389,
      "learning_rate": 2.439296252286747e-06,
      "loss": 0.039,
      "step": 7703
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18415975574074012,
      "learning_rate": 2.4358925763149557e-06,
      "loss": 0.0211,
      "step": 7704
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1973684439533622,
      "learning_rate": 2.432491122668843e-06,
      "loss": 0.0302,
      "step": 7705
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18034695813384868,
      "learning_rate": 2.4290918917787876e-06,
      "loss": 0.0276,
      "step": 7706
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21660120614025083,
      "learning_rate": 2.425694884074876e-06,
      "loss": 0.0351,
      "step": 7707
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.29989806469988955,
      "learning_rate": 2.4223000999869227e-06,
      "loss": 0.0524,
      "step": 7708
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1984272454833669,
      "learning_rate": 2.418907539944457e-06,
      "loss": 0.0411,
      "step": 7709
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.16936174787781888,
      "learning_rate": 2.4155172043767337e-06,
      "loss": 0.0202,
      "step": 7710
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1894116405411674,
      "learning_rate": 2.412129093712712e-06,
      "loss": 0.0266,
      "step": 7711
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.24967479415668112,
      "learning_rate": 2.4087432083810792e-06,
      "loss": 0.0302,
      "step": 7712
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.24881023929577925,
      "learning_rate": 2.40535954881024e-06,
      "loss": 0.0237,
      "step": 7713
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.25225018576604774,
      "learning_rate": 2.401978115428325e-06,
      "loss": 0.0315,
      "step": 7714
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.29597386572783446,
      "learning_rate": 2.3985989086631633e-06,
      "loss": 0.0382,
      "step": 7715
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23358201345092822,
      "learning_rate": 2.395221928942322e-06,
      "loss": 0.0343,
      "step": 7716
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22153920499101093,
      "learning_rate": 2.391847176693074e-06,
      "loss": 0.0414,
      "step": 7717
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2062726156522282,
      "learning_rate": 2.388474652342416e-06,
      "loss": 0.0318,
      "step": 7718
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17318005786339366,
      "learning_rate": 2.385104356317065e-06,
      "loss": 0.0381,
      "step": 7719
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18725059730666252,
      "learning_rate": 2.3817362890434526e-06,
      "loss": 0.049,
      "step": 7720
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2296196441586464,
      "learning_rate": 2.378370450947729e-06,
      "loss": 0.0385,
      "step": 7721
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.25302715706385887,
      "learning_rate": 2.375006842455756e-06,
      "loss": 0.0398,
      "step": 7722
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2929474377827566,
      "learning_rate": 2.3716454639931243e-06,
      "loss": 0.0497,
      "step": 7723
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2366943988927843,
      "learning_rate": 2.3682863159851377e-06,
      "loss": 0.0252,
      "step": 7724
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.24347108967905065,
      "learning_rate": 2.364929398856819e-06,
      "loss": 0.059,
      "step": 7725
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2064907467884103,
      "learning_rate": 2.3615747130329013e-06,
      "loss": 0.0377,
      "step": 7726
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20658114826669727,
      "learning_rate": 2.3582222589378457e-06,
      "loss": 0.0331,
      "step": 7727
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2014120020611004,
      "learning_rate": 2.3548720369958256e-06,
      "loss": 0.0363,
      "step": 7728
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19892929590909753,
      "learning_rate": 2.351524047630731e-06,
      "loss": 0.0216,
      "step": 7729
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19089089321140426,
      "learning_rate": 2.3481782912661788e-06,
      "loss": 0.03,
      "step": 7730
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4642982487768126,
      "learning_rate": 2.344834768325488e-06,
      "loss": 0.0587,
      "step": 7731
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18184357247062863,
      "learning_rate": 2.3414934792317047e-06,
      "loss": 0.0353,
      "step": 7732
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.26270967749136565,
      "learning_rate": 2.338154424407593e-06,
      "loss": 0.0414,
      "step": 7733
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21577644765428147,
      "learning_rate": 2.334817604275632e-06,
      "loss": 0.0308,
      "step": 7734
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22764025226797893,
      "learning_rate": 2.3314830192580117e-06,
      "loss": 0.0159,
      "step": 7735
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20446738171895779,
      "learning_rate": 2.3281506697766522e-06,
      "loss": 0.0448,
      "step": 7736
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2745801619813049,
      "learning_rate": 2.324820556253187e-06,
      "loss": 0.0346,
      "step": 7737
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20171543869149186,
      "learning_rate": 2.3214926791089563e-06,
      "loss": 0.0289,
      "step": 7738
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22112635314812032,
      "learning_rate": 2.3181670387650268e-06,
      "loss": 0.0404,
      "step": 7739
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3684026011147866,
      "learning_rate": 2.3148436356421813e-06,
      "loss": 0.0612,
      "step": 7740
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17693680189015856,
      "learning_rate": 2.311522470160923e-06,
      "loss": 0.0213,
      "step": 7741
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18262174672030046,
      "learning_rate": 2.3082035427414585e-06,
      "loss": 0.033,
      "step": 7742
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.26274660028151026,
      "learning_rate": 2.3048868538037227e-06,
      "loss": 0.0378,
      "step": 7743
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18303939644502495,
      "learning_rate": 2.301572403767369e-06,
      "loss": 0.0252,
      "step": 7744
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23453589321066498,
      "learning_rate": 2.298260193051767e-06,
      "loss": 0.0358,
      "step": 7745
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18955922842488446,
      "learning_rate": 2.2949502220759866e-06,
      "loss": 0.0195,
      "step": 7746
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22538228771584054,
      "learning_rate": 2.291642491258832e-06,
      "loss": 0.0322,
      "step": 7747
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23442264708348803,
      "learning_rate": 2.2883370010188232e-06,
      "loss": 0.0497,
      "step": 7748
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21797750516966352,
      "learning_rate": 2.2850337517741926e-06,
      "loss": 0.0435,
      "step": 7749
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.177253575863299,
      "learning_rate": 2.2817327439428836e-06,
      "loss": 0.0234,
      "step": 7750
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.24503952863087308,
      "learning_rate": 2.2784339779425626e-06,
      "loss": 0.0398,
      "step": 7751
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20090878514316018,
      "learning_rate": 2.2751374541906122e-06,
      "loss": 0.0387,
      "step": 7752
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1962073996040948,
      "learning_rate": 2.271843173104129e-06,
      "loss": 0.0342,
      "step": 7753
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21662818944534437,
      "learning_rate": 2.26855113509993e-06,
      "loss": 0.0285,
      "step": 7754
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1891642757630882,
      "learning_rate": 2.2652613405945423e-06,
      "loss": 0.034,
      "step": 7755
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2561127667351713,
      "learning_rate": 2.261973790004217e-06,
      "loss": 0.0313,
      "step": 7756
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22680476817168074,
      "learning_rate": 2.2586884837449108e-06,
      "loss": 0.0218,
      "step": 7757
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2514915970579732,
      "learning_rate": 2.2554054222323018e-06,
      "loss": 0.0403,
      "step": 7758
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2125804373725085,
      "learning_rate": 2.252124605881789e-06,
      "loss": 0.0268,
      "step": 7759
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19432001671680188,
      "learning_rate": 2.2488460351084827e-06,
      "loss": 0.0419,
      "step": 7760
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21257585475933272,
      "learning_rate": 2.2455697103272022e-06,
      "loss": 0.0361,
      "step": 7761
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19085599669106665,
      "learning_rate": 2.242295631952496e-06,
      "loss": 0.017,
      "step": 7762
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23255052932669992,
      "learning_rate": 2.239023800398621e-06,
      "loss": 0.0307,
      "step": 7763
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21448856818251472,
      "learning_rate": 2.235754216079551e-06,
      "loss": 0.0368,
      "step": 7764
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.26113981296054933,
      "learning_rate": 2.2324868794089727e-06,
      "loss": 0.0277,
      "step": 7765
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.276401099830377,
      "learning_rate": 2.229221790800291e-06,
      "loss": 0.0447,
      "step": 7766
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22557549225376047,
      "learning_rate": 2.2259589506666267e-06,
      "loss": 0.0416,
      "step": 7767
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21602742368128502,
      "learning_rate": 2.2226983594208187e-06,
      "loss": 0.0323,
      "step": 7768
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17996091358672361,
      "learning_rate": 2.219440017475418e-06,
      "loss": 0.0296,
      "step": 7769
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3579282456805789,
      "learning_rate": 2.216183925242681e-06,
      "loss": 0.0396,
      "step": 7770
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2148267392446688,
      "learning_rate": 2.2129300831346033e-06,
      "loss": 0.0276,
      "step": 7771
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.27491615220076,
      "learning_rate": 2.209678491562881e-06,
      "loss": 0.0316,
      "step": 7772
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18473558242146437,
      "learning_rate": 2.206429150938918e-06,
      "loss": 0.0224,
      "step": 7773
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20002848489391037,
      "learning_rate": 2.2031820616738477e-06,
      "loss": 0.0238,
      "step": 7774
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19476562230691247,
      "learning_rate": 2.199937224178512e-06,
      "loss": 0.0367,
      "step": 7775
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18066858853006224,
      "learning_rate": 2.1966946388634746e-06,
      "loss": 0.0288,
      "step": 7776
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17057858644531207,
      "learning_rate": 2.193454306138998e-06,
      "loss": 0.0201,
      "step": 7777
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1674343690736097,
      "learning_rate": 2.190216226415074e-06,
      "loss": 0.0265,
      "step": 7778
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2614546197397004,
      "learning_rate": 2.1869804001014085e-06,
      "loss": 0.0451,
      "step": 7779
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22725299321460085,
      "learning_rate": 2.1837468276074227e-06,
      "loss": 0.0466,
      "step": 7780
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18497122763153595,
      "learning_rate": 2.1805155093422402e-06,
      "loss": 0.022,
      "step": 7781
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21714030377354016,
      "learning_rate": 2.1772864457147126e-06,
      "loss": 0.0359,
      "step": 7782
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2257479131566246,
      "learning_rate": 2.174059637133403e-06,
      "loss": 0.0326,
      "step": 7783
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.24847196963640764,
      "learning_rate": 2.1708350840065927e-06,
      "loss": 0.0444,
      "step": 7784
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22189190659113614,
      "learning_rate": 2.1676127867422637e-06,
      "loss": 0.035,
      "step": 7785
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22376283255865045,
      "learning_rate": 2.164392745748125e-06,
      "loss": 0.0285,
      "step": 7786
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18958655309213665,
      "learning_rate": 2.1611749614316004e-06,
      "loss": 0.0294,
      "step": 7787
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22695593544931894,
      "learning_rate": 2.1579594341998235e-06,
      "loss": 0.0459,
      "step": 7788
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19187613634462636,
      "learning_rate": 2.1547461644596446e-06,
      "loss": 0.0235,
      "step": 7789
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21911279362215938,
      "learning_rate": 2.151535152617625e-06,
      "loss": 0.0344,
      "step": 7790
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20046058385425244,
      "learning_rate": 2.148326399080052e-06,
      "loss": 0.0272,
      "step": 7791
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19551664174328526,
      "learning_rate": 2.1451199042529035e-06,
      "loss": 0.0241,
      "step": 7792
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.16344634794351515,
      "learning_rate": 2.1419156685418964e-06,
      "loss": 0.0151,
      "step": 7793
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20937803216418222,
      "learning_rate": 2.1387136923524475e-06,
      "loss": 0.0305,
      "step": 7794
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.25172641815875657,
      "learning_rate": 2.1355139760896957e-06,
      "loss": 0.0356,
      "step": 7795
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2562171094999848,
      "learning_rate": 2.1323165201584863e-06,
      "loss": 0.0313,
      "step": 7796
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17292346644149642,
      "learning_rate": 2.1291213249633813e-06,
      "loss": 0.0308,
      "step": 7797
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2359648742332226,
      "learning_rate": 2.125928390908658e-06,
      "loss": 0.0274,
      "step": 7798
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18758848208583861,
      "learning_rate": 2.1227377183983154e-06,
      "loss": 0.0217,
      "step": 7799
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2193224008515313,
      "learning_rate": 2.1195493078360486e-06,
      "loss": 0.023,
      "step": 7800
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20483069252397137,
      "learning_rate": 2.1163631596252785e-06,
      "loss": 0.0278,
      "step": 7801
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22196979277509635,
      "learning_rate": 2.113179274169137e-06,
      "loss": 0.0264,
      "step": 7802
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24198387414970693,
      "learning_rate": 2.109997651870477e-06,
      "loss": 0.0334,
      "step": 7803
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22899970109734746,
      "learning_rate": 2.1068182931318424e-06,
      "loss": 0.0223,
      "step": 7804
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19346933417215884,
      "learning_rate": 2.1036411983555237e-06,
      "loss": 0.0257,
      "step": 7805
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19151442254277118,
      "learning_rate": 2.1004663679434987e-06,
      "loss": 0.0165,
      "step": 7806
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.23501312488062776,
      "learning_rate": 2.0972938022974733e-06,
      "loss": 0.0457,
      "step": 7807
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21188176695670952,
      "learning_rate": 2.0941235018188543e-06,
      "loss": 0.033,
      "step": 7808
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22720612164963516,
      "learning_rate": 2.0909554669087706e-06,
      "loss": 0.028,
      "step": 7809
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22607865957425907,
      "learning_rate": 2.0877896979680654e-06,
      "loss": 0.0284,
      "step": 7810
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22876386119714157,
      "learning_rate": 2.0846261953972967e-06,
      "loss": 0.0373,
      "step": 7811
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24453470849220896,
      "learning_rate": 2.0814649595967194e-06,
      "loss": 0.0383,
      "step": 7812
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19415344602744983,
      "learning_rate": 2.0783059909663206e-06,
      "loss": 0.0224,
      "step": 7813
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19452367752534103,
      "learning_rate": 2.0751492899057957e-06,
      "loss": 0.0277,
      "step": 7814
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1519874048669296,
      "learning_rate": 2.07199485681455e-06,
      "loss": 0.0143,
      "step": 7815
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21751336960972478,
      "learning_rate": 2.0688426920916992e-06,
      "loss": 0.0402,
      "step": 7816
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2679816217917729,
      "learning_rate": 2.0656927961360785e-06,
      "loss": 0.0261,
      "step": 7817
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19735513676935454,
      "learning_rate": 2.062545169346235e-06,
      "loss": 0.0191,
      "step": 7818
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2534276412852567,
      "learning_rate": 2.0593998121204264e-06,
      "loss": 0.0479,
      "step": 7819
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2764394281549932,
      "learning_rate": 2.05625672485662e-06,
      "loss": 0.0311,
      "step": 7820
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18219273243975537,
      "learning_rate": 2.053115907952499e-06,
      "loss": 0.0218,
      "step": 7821
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19931312270499343,
      "learning_rate": 2.049977361805471e-06,
      "loss": 0.0303,
      "step": 7822
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2082707082480229,
      "learning_rate": 2.0468410868126343e-06,
      "loss": 0.0319,
      "step": 7823
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3296984684281185,
      "learning_rate": 2.043707083370814e-06,
      "loss": 0.0368,
      "step": 7824
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21641233208437777,
      "learning_rate": 2.0405753518765457e-06,
      "loss": 0.0307,
      "step": 7825
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19958385035587894,
      "learning_rate": 2.03744589272608e-06,
      "loss": 0.038,
      "step": 7826
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.26299869645633306,
      "learning_rate": 2.034318706315368e-06,
      "loss": 0.033,
      "step": 7827
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38151802428627263,
      "learning_rate": 2.031193793040087e-06,
      "loss": 0.0454,
      "step": 7828
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21418370379658888,
      "learning_rate": 2.028071153295621e-06,
      "loss": 0.0326,
      "step": 7829
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.13953515106078992,
      "learning_rate": 2.0249507874770714e-06,
      "loss": 0.0248,
      "step": 7830
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18903826487297606,
      "learning_rate": 2.0218326959792357e-06,
      "loss": 0.035,
      "step": 7831
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.23306386102456478,
      "learning_rate": 2.018716879196645e-06,
      "loss": 0.0271,
      "step": 7832
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19999979287375216,
      "learning_rate": 2.0156033375235286e-06,
      "loss": 0.0357,
      "step": 7833
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3290704774749966,
      "learning_rate": 2.0124920713538378e-06,
      "loss": 0.0413,
      "step": 7834
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20094415127141455,
      "learning_rate": 2.009383081081222e-06,
      "loss": 0.0376,
      "step": 7835
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21115602547415008,
      "learning_rate": 2.006276367099054e-06,
      "loss": 0.0299,
      "step": 7836
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17481565174688957,
      "learning_rate": 2.003171929800416e-06,
      "loss": 0.0218,
      "step": 7837
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1567136323270094,
      "learning_rate": 2.000069769578108e-06,
      "loss": 0.0243,
      "step": 7838
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24722588959588923,
      "learning_rate": 1.99696988682462e-06,
      "loss": 0.0254,
      "step": 7839
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2751564220404101,
      "learning_rate": 1.9938722819321854e-06,
      "loss": 0.0364,
      "step": 7840
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1993064500339682,
      "learning_rate": 1.9907769552927304e-06,
      "loss": 0.0363,
      "step": 7841
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20347238143459712,
      "learning_rate": 1.987683907297888e-06,
      "loss": 0.0216,
      "step": 7842
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17353879766985414,
      "learning_rate": 1.984593138339015e-06,
      "loss": 0.0272,
      "step": 7843
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21003687172999896,
      "learning_rate": 1.9815046488071774e-06,
      "loss": 0.0271,
      "step": 7844
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22389184534093173,
      "learning_rate": 1.9784184390931504e-06,
      "loss": 0.0301,
      "step": 7845
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.25424214666328915,
      "learning_rate": 1.9753345095874234e-06,
      "loss": 0.0344,
      "step": 7846
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24692807230697378,
      "learning_rate": 1.9722528606801906e-06,
      "loss": 0.0287,
      "step": 7847
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.277819279544011,
      "learning_rate": 1.9691734927613625e-06,
      "loss": 0.0405,
      "step": 7848
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.23186107103971815,
      "learning_rate": 1.9660964062205657e-06,
      "loss": 0.029,
      "step": 7849
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18243360216237806,
      "learning_rate": 1.9630216014471326e-06,
      "loss": 0.0223,
      "step": 7850
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19514721552489403,
      "learning_rate": 1.9599490788301033e-06,
      "loss": 0.0169,
      "step": 7851
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.23507576924465626,
      "learning_rate": 1.9568788387582338e-06,
      "loss": 0.0485,
      "step": 7852
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21133806025477678,
      "learning_rate": 1.9538108816199953e-06,
      "loss": 0.015,
      "step": 7853
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1865882960493375,
      "learning_rate": 1.950745207803566e-06,
      "loss": 0.0202,
      "step": 7854
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2547915211149743,
      "learning_rate": 1.947681817696827e-06,
      "loss": 0.0315,
      "step": 7855
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22103638090027158,
      "learning_rate": 1.9446207116873815e-06,
      "loss": 0.0428,
      "step": 7856
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.15675748428169678,
      "learning_rate": 1.941561890162551e-06,
      "loss": 0.0254,
      "step": 7857
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.16412827898789895,
      "learning_rate": 1.9385053535093455e-06,
      "loss": 0.0277,
      "step": 7858
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18364920691667633,
      "learning_rate": 1.935451102114503e-06,
      "loss": 0.0254,
      "step": 7859
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3230474755720947,
      "learning_rate": 1.9323991363644645e-06,
      "loss": 0.0375,
      "step": 7860
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24119455073366722,
      "learning_rate": 1.929349456645393e-06,
      "loss": 0.0211,
      "step": 7861
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.252791215851772,
      "learning_rate": 1.9263020633431416e-06,
      "loss": 0.0279,
      "step": 7862
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.30249848558503956,
      "learning_rate": 1.9232569568432933e-06,
      "loss": 0.046,
      "step": 7863
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24358671294417142,
      "learning_rate": 1.9202141375311335e-06,
      "loss": 0.0486,
      "step": 7864
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.23051718833226184,
      "learning_rate": 1.9171736057916623e-06,
      "loss": 0.0396,
      "step": 7865
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2625220198705256,
      "learning_rate": 1.9141353620095835e-06,
      "loss": 0.037,
      "step": 7866
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1665907267079886,
      "learning_rate": 1.9110994065693166e-06,
      "loss": 0.0216,
      "step": 7867
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19124906274777803,
      "learning_rate": 1.9080657398549916e-06,
      "loss": 0.0271,
      "step": 7868
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1824135894854214,
      "learning_rate": 1.905034362250453e-06,
      "loss": 0.0246,
      "step": 7869
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21556965698213304,
      "learning_rate": 1.902005274139238e-06,
      "loss": 0.0262,
      "step": 7870
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20217440157387007,
      "learning_rate": 1.8989784759046158e-06,
      "loss": 0.0197,
      "step": 7871
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2331335567816457,
      "learning_rate": 1.8959539679295536e-06,
      "loss": 0.0381,
      "step": 7872
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1879255512157403,
      "learning_rate": 1.8929317505967338e-06,
      "loss": 0.0359,
      "step": 7873
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1922801174512988,
      "learning_rate": 1.8899118242885462e-06,
      "loss": 0.0378,
      "step": 7874
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1926312560941338,
      "learning_rate": 1.8868941893870918e-06,
      "loss": 0.0325,
      "step": 7875
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.211330242463678,
      "learning_rate": 1.8838788462741852e-06,
      "loss": 0.0262,
      "step": 7876
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23234168039566105,
      "learning_rate": 1.8808657953313391e-06,
      "loss": 0.0299,
      "step": 7877
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1899279513656114,
      "learning_rate": 1.8778550369397886e-06,
      "loss": 0.0263,
      "step": 7878
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.26021272143768537,
      "learning_rate": 1.8748465714804753e-06,
      "loss": 0.0249,
      "step": 7879
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2777372443242748,
      "learning_rate": 1.8718403993340528e-06,
      "loss": 0.0363,
      "step": 7880
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.14700750400405224,
      "learning_rate": 1.8688365208808723e-06,
      "loss": 0.0191,
      "step": 7881
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2446700269055719,
      "learning_rate": 1.865834936501012e-06,
      "loss": 0.0411,
      "step": 7882
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18618530061747474,
      "learning_rate": 1.8628356465742503e-06,
      "loss": 0.0226,
      "step": 7883
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20829674975736465,
      "learning_rate": 1.8598386514800793e-06,
      "loss": 0.0195,
      "step": 7884
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.252207354245023,
      "learning_rate": 1.8568439515976932e-06,
      "loss": 0.0319,
      "step": 7885
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.28411563007128116,
      "learning_rate": 1.8538515473060026e-06,
      "loss": 0.0537,
      "step": 7886
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20247397264355838,
      "learning_rate": 1.8508614389836289e-06,
      "loss": 0.0268,
      "step": 7887
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2333770486775457,
      "learning_rate": 1.847873627008896e-06,
      "loss": 0.0442,
      "step": 7888
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19974457777228918,
      "learning_rate": 1.8448881117598484e-06,
      "loss": 0.039,
      "step": 7889
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21231059581182296,
      "learning_rate": 1.8419048936142191e-06,
      "loss": 0.0347,
      "step": 7890
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19353614779926795,
      "learning_rate": 1.8389239729494778e-06,
      "loss": 0.0323,
      "step": 7891
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18320920930043838,
      "learning_rate": 1.8359453501427916e-06,
      "loss": 0.0268,
      "step": 7892
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2114590096736922,
      "learning_rate": 1.832969025571023e-06,
      "loss": 0.0371,
      "step": 7893
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21235114171044794,
      "learning_rate": 1.8299949996107646e-06,
      "loss": 0.0244,
      "step": 7894
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.22112650477003562,
      "learning_rate": 1.8270232726383064e-06,
      "loss": 0.0246,
      "step": 7895
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18684562099723942,
      "learning_rate": 1.8240538450296563e-06,
      "loss": 0.042,
      "step": 7896
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20909525701606596,
      "learning_rate": 1.8210867171605185e-06,
      "loss": 0.0407,
      "step": 7897
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20344163900697407,
      "learning_rate": 1.8181218894063146e-06,
      "loss": 0.0454,
      "step": 7898
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.15661957068938992,
      "learning_rate": 1.815159362142176e-06,
      "loss": 0.0204,
      "step": 7899
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.25707855658882445,
      "learning_rate": 1.8121991357429425e-06,
      "loss": 0.0208,
      "step": 7900
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23156069844944088,
      "learning_rate": 1.809241210583159e-06,
      "loss": 0.038,
      "step": 7901
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2157503420262372,
      "learning_rate": 1.8062855870370798e-06,
      "loss": 0.0392,
      "step": 7902
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17850745451612188,
      "learning_rate": 1.8033322654786723e-06,
      "loss": 0.0276,
      "step": 7903
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1806825782979088,
      "learning_rate": 1.8003812462816127e-06,
      "loss": 0.0304,
      "step": 7904
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2163544513668072,
      "learning_rate": 1.7974325298192762e-06,
      "loss": 0.0321,
      "step": 7905
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.24308614746457385,
      "learning_rate": 1.7944861164647576e-06,
      "loss": 0.0525,
      "step": 7906
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17603746391298183,
      "learning_rate": 1.7915420065908585e-06,
      "loss": 0.0211,
      "step": 7907
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1757718083707014,
      "learning_rate": 1.788600200570083e-06,
      "loss": 0.0354,
      "step": 7908
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.24477590686036726,
      "learning_rate": 1.7856606987746494e-06,
      "loss": 0.0386,
      "step": 7909
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18863884457035762,
      "learning_rate": 1.782723501576482e-06,
      "loss": 0.0366,
      "step": 7910
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23034008119771987,
      "learning_rate": 1.7797886093472195e-06,
      "loss": 0.0425,
      "step": 7911
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18517581578787393,
      "learning_rate": 1.7768560224581955e-06,
      "loss": 0.0283,
      "step": 7912
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1917498676630332,
      "learning_rate": 1.7739257412804623e-06,
      "loss": 0.0175,
      "step": 7913
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.27134389467377207,
      "learning_rate": 1.770997766184781e-06,
      "loss": 0.0425,
      "step": 7914
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.155096409005671,
      "learning_rate": 1.76807209754162e-06,
      "loss": 0.0127,
      "step": 7915
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1813598534216297,
      "learning_rate": 1.7651487357211472e-06,
      "loss": 0.0159,
      "step": 7916
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1770506061904471,
      "learning_rate": 1.7622276810932492e-06,
      "loss": 0.0225,
      "step": 7917
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3575194387219582,
      "learning_rate": 1.7593089340275149e-06,
      "loss": 0.0457,
      "step": 7918
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21684770589623867,
      "learning_rate": 1.7563924948932487e-06,
      "loss": 0.0296,
      "step": 7919
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3036167697945338,
      "learning_rate": 1.7534783640594533e-06,
      "loss": 0.0238,
      "step": 7920
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18773872357752094,
      "learning_rate": 1.7505665418948404e-06,
      "loss": 0.0275,
      "step": 7921
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.14997211038076155,
      "learning_rate": 1.7476570287678396e-06,
      "loss": 0.0203,
      "step": 7922
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1479350294101062,
      "learning_rate": 1.7447498250465788e-06,
      "loss": 0.0204,
      "step": 7923
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1951493535529808,
      "learning_rate": 1.741844931098895e-06,
      "loss": 0.0235,
      "step": 7924
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1793195440052939,
      "learning_rate": 1.7389423472923318e-06,
      "loss": 0.027,
      "step": 7925
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21369000938495214,
      "learning_rate": 1.7360420739941486e-06,
      "loss": 0.0364,
      "step": 7926
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2002626259379726,
      "learning_rate": 1.73314411157131e-06,
      "loss": 0.0292,
      "step": 7927
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19496304721260208,
      "learning_rate": 1.7302484603904756e-06,
      "loss": 0.0408,
      "step": 7928
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1570431486527773,
      "learning_rate": 1.727355120818026e-06,
      "loss": 0.0215,
      "step": 7929
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21995434292700441,
      "learning_rate": 1.7244640932200484e-06,
      "loss": 0.0341,
      "step": 7930
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2197952078508393,
      "learning_rate": 1.7215753779623346e-06,
      "loss": 0.049,
      "step": 7931
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2148772473664342,
      "learning_rate": 1.7186889754103763e-06,
      "loss": 0.0224,
      "step": 7932
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.22067599847229957,
      "learning_rate": 1.7158048859293863e-06,
      "loss": 0.0275,
      "step": 7933
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20085702720406495,
      "learning_rate": 1.7129231098842791e-06,
      "loss": 0.0391,
      "step": 7934
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.213639107236686,
      "learning_rate": 1.7100436476396742e-06,
      "loss": 0.0412,
      "step": 7935
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.22289524302226313,
      "learning_rate": 1.707166499559898e-06,
      "loss": 0.0424,
      "step": 7936
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2126394767789094,
      "learning_rate": 1.704291666008988e-06,
      "loss": 0.028,
      "step": 7937
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19241805165341572,
      "learning_rate": 1.701419147350687e-06,
      "loss": 0.0291,
      "step": 7938
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18691018836970197,
      "learning_rate": 1.698548943948448e-06,
      "loss": 0.0242,
      "step": 7939
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2730643041720939,
      "learning_rate": 1.6956810561654213e-06,
      "loss": 0.0353,
      "step": 7940
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.16982099673706363,
      "learning_rate": 1.6928154843644717e-06,
      "loss": 0.0179,
      "step": 7941
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.154833933368002,
      "learning_rate": 1.6899522289081737e-06,
      "loss": 0.0234,
      "step": 7942
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.24796870644466257,
      "learning_rate": 1.6870912901588044e-06,
      "loss": 0.0327,
      "step": 7943
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20477279458416758,
      "learning_rate": 1.6842326684783473e-06,
      "loss": 0.0176,
      "step": 7944
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.15502041167982097,
      "learning_rate": 1.6813763642284953e-06,
      "loss": 0.0278,
      "step": 7945
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21465272213631295,
      "learning_rate": 1.6785223777706482e-06,
      "loss": 0.0294,
      "step": 7946
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4015118229278483,
      "learning_rate": 1.6756707094659063e-06,
      "loss": 0.0559,
      "step": 7947
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19306698279519022,
      "learning_rate": 1.6728213596750831e-06,
      "loss": 0.0243,
      "step": 7948
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.22122094502379186,
      "learning_rate": 1.6699743287586967e-06,
      "loss": 0.0334,
      "step": 7949
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.22587818355090103,
      "learning_rate": 1.667129617076977e-06,
      "loss": 0.0463,
      "step": 7950
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.24701797172188855,
      "learning_rate": 1.664287224989849e-06,
      "loss": 0.0347,
      "step": 7951
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2561457974538717,
      "learning_rate": 1.661447152856952e-06,
      "loss": 0.037,
      "step": 7952
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.16796007799340026,
      "learning_rate": 1.658609401037632e-06,
      "loss": 0.0266,
      "step": 7953
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23761036498249274,
      "learning_rate": 1.6557739698909436e-06,
      "loss": 0.0325,
      "step": 7954
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2519971823799712,
      "learning_rate": 1.6529408597756359e-06,
      "loss": 0.0411,
      "step": 7955
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23477529674036418,
      "learning_rate": 1.650110071050175e-06,
      "loss": 0.0226,
      "step": 7956
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.26944641492036814,
      "learning_rate": 1.6472816040727346e-06,
      "loss": 0.0243,
      "step": 7957
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2194209537846694,
      "learning_rate": 1.6444554592011909e-06,
      "loss": 0.023,
      "step": 7958
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18514601924654112,
      "learning_rate": 1.6416316367931173e-06,
      "loss": 0.0245,
      "step": 7959
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17140778523490457,
      "learning_rate": 1.638810137205813e-06,
      "loss": 0.0268,
      "step": 7960
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2409443405184701,
      "learning_rate": 1.6359909607962676e-06,
      "loss": 0.029,
      "step": 7961
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18308959892165264,
      "learning_rate": 1.6331741079211872e-06,
      "loss": 0.0284,
      "step": 7962
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2210233610005486,
      "learning_rate": 1.630359578936971e-06,
      "loss": 0.0332,
      "step": 7963
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.15409226069159063,
      "learning_rate": 1.627547374199734e-06,
      "loss": 0.0222,
      "step": 7964
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2537657862718482,
      "learning_rate": 1.624737494065294e-06,
      "loss": 0.0256,
      "step": 7965
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2758209032507009,
      "learning_rate": 1.6219299388891797e-06,
      "loss": 0.0256,
      "step": 7966
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.22660797583532027,
      "learning_rate": 1.619124709026616e-06,
      "loss": 0.0342,
      "step": 7967
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1772977055000426,
      "learning_rate": 1.6163218048325413e-06,
      "loss": 0.0264,
      "step": 7968
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42842704392910747,
      "learning_rate": 1.6135212266615963e-06,
      "loss": 0.0532,
      "step": 7969
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.265535171129659,
      "learning_rate": 1.610722974868133e-06,
      "loss": 0.0358,
      "step": 7970
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36208694783606293,
      "learning_rate": 1.6079270498061995e-06,
      "loss": 0.0381,
      "step": 7971
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19888617299617603,
      "learning_rate": 1.6051334518295546e-06,
      "loss": 0.0312,
      "step": 7972
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.23326172276177234,
      "learning_rate": 1.6023421812916628e-06,
      "loss": 0.028,
      "step": 7973
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.29779875240237524,
      "learning_rate": 1.5995532385456992e-06,
      "loss": 0.0368,
      "step": 7974
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2704987231769148,
      "learning_rate": 1.59676662394453e-06,
      "loss": 0.0394,
      "step": 7975
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19269490947020068,
      "learning_rate": 1.5939823378407426e-06,
      "loss": 0.0414,
      "step": 7976
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2513750527076912,
      "learning_rate": 1.5912003805866216e-06,
      "loss": 0.0279,
      "step": 7977
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2128318215448173,
      "learning_rate": 1.5884207525341566e-06,
      "loss": 0.0147,
      "step": 7978
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.26181646202624653,
      "learning_rate": 1.5856434540350462e-06,
      "loss": 0.0401,
      "step": 7979
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21412469872165646,
      "learning_rate": 1.5828684854406918e-06,
      "loss": 0.0262,
      "step": 7980
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.25767178018471165,
      "learning_rate": 1.5800958471022032e-06,
      "loss": 0.0421,
      "step": 7981
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19291822606033907,
      "learning_rate": 1.577325539370389e-06,
      "loss": 0.0266,
      "step": 7982
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18335442466863613,
      "learning_rate": 1.5745575625957665e-06,
      "loss": 0.0366,
      "step": 7983
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19633944419961502,
      "learning_rate": 1.57179191712856e-06,
      "loss": 0.0376,
      "step": 7984
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.17817342417852305,
      "learning_rate": 1.5690286033186985e-06,
      "loss": 0.0355,
      "step": 7985
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.30630844492739323,
      "learning_rate": 1.5662676215158112e-06,
      "loss": 0.0446,
      "step": 7986
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22457843235668706,
      "learning_rate": 1.5635089720692366e-06,
      "loss": 0.0276,
      "step": 7987
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.23044246588359513,
      "learning_rate": 1.5607526553280172e-06,
      "loss": 0.0273,
      "step": 7988
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2203293479402333,
      "learning_rate": 1.5579986716409036e-06,
      "loss": 0.0296,
      "step": 7989
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19257002885060853,
      "learning_rate": 1.5552470213563408e-06,
      "loss": 0.032,
      "step": 7990
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2546424316830903,
      "learning_rate": 1.5524977048224888e-06,
      "loss": 0.0315,
      "step": 7991
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1907247165269804,
      "learning_rate": 1.549750722387211e-06,
      "loss": 0.037,
      "step": 7992
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.483785917097084,
      "learning_rate": 1.5470060743980765e-06,
      "loss": 0.0817,
      "step": 7993
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18970231254472153,
      "learning_rate": 1.5442637612023425e-06,
      "loss": 0.0295,
      "step": 7994
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20054782442195834,
      "learning_rate": 1.5415237831470008e-06,
      "loss": 0.0219,
      "step": 7995
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18206602959464085,
      "learning_rate": 1.5387861405787252e-06,
      "loss": 0.0295,
      "step": 7996
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20744311552615868,
      "learning_rate": 1.5360508338438962e-06,
      "loss": 0.0236,
      "step": 7997
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.24357862255503948,
      "learning_rate": 1.5333178632886058e-06,
      "loss": 0.0562,
      "step": 7998
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19426827714549116,
      "learning_rate": 1.5305872292586489e-06,
      "loss": 0.0287,
      "step": 7999
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.14979020935281698,
      "learning_rate": 1.5278589320995218e-06,
      "loss": 0.0165,
      "step": 8000
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22219107371486266,
      "learning_rate": 1.5251329721564245e-06,
      "loss": 0.0269,
      "step": 8001
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20202302036797745,
      "learning_rate": 1.5224093497742654e-06,
      "loss": 0.0344,
      "step": 8002
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.17795456880865826,
      "learning_rate": 1.5196880652976553e-06,
      "loss": 0.026,
      "step": 8003
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21498140780047229,
      "learning_rate": 1.5169691190709057e-06,
      "loss": 0.0522,
      "step": 8004
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19789309842340624,
      "learning_rate": 1.5142525114380436e-06,
      "loss": 0.0397,
      "step": 8005
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.28882247393040084,
      "learning_rate": 1.5115382427427827e-06,
      "loss": 0.0394,
      "step": 8006
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1445606498199923,
      "learning_rate": 1.508826313328553e-06,
      "loss": 0.0189,
      "step": 8007
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18835116944067726,
      "learning_rate": 1.5061167235384867e-06,
      "loss": 0.043,
      "step": 8008
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.17654901891909514,
      "learning_rate": 1.5034094737154204e-06,
      "loss": 0.034,
      "step": 8009
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2584321637969228,
      "learning_rate": 1.5007045642018868e-06,
      "loss": 0.029,
      "step": 8010
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.17510615092584242,
      "learning_rate": 1.4980019953401327e-06,
      "loss": 0.0268,
      "step": 8011
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19896464878730663,
      "learning_rate": 1.4953017674721083e-06,
      "loss": 0.027,
      "step": 8012
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.23013502948979184,
      "learning_rate": 1.4926038809394606e-06,
      "loss": 0.0337,
      "step": 8013
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.23607700183724187,
      "learning_rate": 1.4899083360835408e-06,
      "loss": 0.0458,
      "step": 8014
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2883132942605547,
      "learning_rate": 1.4872151332454122e-06,
      "loss": 0.0408,
      "step": 8015
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2031435041068808,
      "learning_rate": 1.484524272765837e-06,
      "loss": 0.0263,
      "step": 8016
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2153970356437948,
      "learning_rate": 1.4818357549852747e-06,
      "loss": 0.0318,
      "step": 8017
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18207555403629702,
      "learning_rate": 1.479149580243895e-06,
      "loss": 0.0288,
      "step": 8018
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.16530708129237068,
      "learning_rate": 1.4764657488815748e-06,
      "loss": 0.0193,
      "step": 8019
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21682695231705063,
      "learning_rate": 1.4737842612378894e-06,
      "loss": 0.0305,
      "step": 8020
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18897665218550008,
      "learning_rate": 1.471105117652114e-06,
      "loss": 0.0233,
      "step": 8021
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1509863831754559,
      "learning_rate": 1.468428318463233e-06,
      "loss": 0.0213,
      "step": 8022
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1847319828399,
      "learning_rate": 1.4657538640099333e-06,
      "loss": 0.0267,
      "step": 8023
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.15537337919480862,
      "learning_rate": 1.4630817546306087e-06,
      "loss": 0.026,
      "step": 8024
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21107534919916818,
      "learning_rate": 1.460411990663344e-06,
      "loss": 0.0478,
      "step": 8025
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21762419406373532,
      "learning_rate": 1.4577445724459382e-06,
      "loss": 0.0387,
      "step": 8026
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19120625343079603,
      "learning_rate": 1.4550795003158902e-06,
      "loss": 0.0294,
      "step": 8027
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.17440145218426678,
      "learning_rate": 1.4524167746104034e-06,
      "loss": 0.0396,
      "step": 8028
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1819761417943983,
      "learning_rate": 1.449756395666384e-06,
      "loss": 0.0328,
      "step": 8029
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21349745191889433,
      "learning_rate": 1.4470983638204384e-06,
      "loss": 0.0422,
      "step": 8030
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19151451980148682,
      "learning_rate": 1.444442679408884e-06,
      "loss": 0.0294,
      "step": 8031
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.203144329325111,
      "learning_rate": 1.4417893427677276e-06,
      "loss": 0.034,
      "step": 8032
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.30827546908434406,
      "learning_rate": 1.4391383542326875e-06,
      "loss": 0.0361,
      "step": 8033
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2233949420653495,
      "learning_rate": 1.4364897141391888e-06,
      "loss": 0.039,
      "step": 8034
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.26912777874190214,
      "learning_rate": 1.4338434228223541e-06,
      "loss": 0.0667,
      "step": 8035
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20267366738865317,
      "learning_rate": 1.4311994806170048e-06,
      "loss": 0.0285,
      "step": 8036
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2503723709189332,
      "learning_rate": 1.4285578878576734e-06,
      "loss": 0.0389,
      "step": 8037
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.272093850327352,
      "learning_rate": 1.42591864487859e-06,
      "loss": 0.0203,
      "step": 8038
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19977832258536013,
      "learning_rate": 1.4232817520136922e-06,
      "loss": 0.0337,
      "step": 8039
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1756499329784818,
      "learning_rate": 1.4206472095966107e-06,
      "loss": 0.0306,
      "step": 8040
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1650859608124183,
      "learning_rate": 1.41801501796069e-06,
      "loss": 0.0266,
      "step": 8041
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1699920049182769,
      "learning_rate": 1.4153851774389703e-06,
      "loss": 0.019,
      "step": 8042
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21420148736805092,
      "learning_rate": 1.4127576883642014e-06,
      "loss": 0.0281,
      "step": 8043
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18611832008902707,
      "learning_rate": 1.4101325510688192e-06,
      "loss": 0.0186,
      "step": 8044
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2832860364893558,
      "learning_rate": 1.4075097658849824e-06,
      "loss": 0.0357,
      "step": 8045
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21569811265614477,
      "learning_rate": 1.4048893331445367e-06,
      "loss": 0.02,
      "step": 8046
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22187091125159475,
      "learning_rate": 1.4022712531790462e-06,
      "loss": 0.0269,
      "step": 8047
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2826525235708043,
      "learning_rate": 1.3996555263197587e-06,
      "loss": 0.0293,
      "step": 8048
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20756747421258975,
      "learning_rate": 1.3970421528976342e-06,
      "loss": 0.0413,
      "step": 8049
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21787853014543335,
      "learning_rate": 1.3944311332433368e-06,
      "loss": 0.0261,
      "step": 8050
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2566588919306328,
      "learning_rate": 1.3918224676872294e-06,
      "loss": 0.0436,
      "step": 8051
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1961498620957768,
      "learning_rate": 1.3892161565593743e-06,
      "loss": 0.0288,
      "step": 8052
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1820423953692304,
      "learning_rate": 1.386612200189541e-06,
      "loss": 0.0375,
      "step": 8053
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.25915954270082875,
      "learning_rate": 1.3840105989071995e-06,
      "loss": 0.0373,
      "step": 8054
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35149368036403383,
      "learning_rate": 1.3814113530415218e-06,
      "loss": 0.0476,
      "step": 8055
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18617289492847594,
      "learning_rate": 1.3788144629213785e-06,
      "loss": 0.0298,
      "step": 8056
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.319603903795184,
      "learning_rate": 1.3762199288753485e-06,
      "loss": 0.0443,
      "step": 8057
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19687867690997315,
      "learning_rate": 1.3736277512317076e-06,
      "loss": 0.0427,
      "step": 8058
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19770394153225865,
      "learning_rate": 1.3710379303184374e-06,
      "loss": 0.025,
      "step": 8059
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18163010351410475,
      "learning_rate": 1.3684504664632137e-06,
      "loss": 0.019,
      "step": 8060
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19604836124809333,
      "learning_rate": 1.3658653599934235e-06,
      "loss": 0.0466,
      "step": 8061
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20865078621361277,
      "learning_rate": 1.3632826112361497e-06,
      "loss": 0.0335,
      "step": 8062
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19071613189272868,
      "learning_rate": 1.3607022205181798e-06,
      "loss": 0.0224,
      "step": 8063
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.17919461402901632,
      "learning_rate": 1.3581241881660011e-06,
      "loss": 0.0253,
      "step": 8064
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20057624303165597,
      "learning_rate": 1.3555485145058044e-06,
      "loss": 0.0384,
      "step": 8065
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20989389458100902,
      "learning_rate": 1.352975199863482e-06,
      "loss": 0.0243,
      "step": 8066
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3578338749926915,
      "learning_rate": 1.3504042445646205e-06,
      "loss": 0.0494,
      "step": 8067
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23241561592714144,
      "learning_rate": 1.3478356489345168e-06,
      "loss": 0.0174,
      "step": 8068
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1957165163373474,
      "learning_rate": 1.345269413298167e-06,
      "loss": 0.0263,
      "step": 8069
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2510463305919896,
      "learning_rate": 1.3427055379802733e-06,
      "loss": 0.0476,
      "step": 8070
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1708102461917083,
      "learning_rate": 1.3401440233052233e-06,
      "loss": 0.0208,
      "step": 8071
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.17977720592186733,
      "learning_rate": 1.3375848695971239e-06,
      "loss": 0.0254,
      "step": 8072
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20638241851938854,
      "learning_rate": 1.335028077179774e-06,
      "loss": 0.0284,
      "step": 8073
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1685974641449145,
      "learning_rate": 1.3324736463766775e-06,
      "loss": 0.0263,
      "step": 8074
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21196650370392212,
      "learning_rate": 1.329921577511033e-06,
      "loss": 0.0388,
      "step": 8075
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1863458397079767,
      "learning_rate": 1.3273718709057493e-06,
      "loss": 0.0245,
      "step": 8076
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20820650273315364,
      "learning_rate": 1.3248245268834304e-06,
      "loss": 0.0215,
      "step": 8077
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19686201559875893,
      "learning_rate": 1.322279545766385e-06,
      "loss": 0.0248,
      "step": 8078
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1644193084452681,
      "learning_rate": 1.3197369278766182e-06,
      "loss": 0.0229,
      "step": 8079
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20544142559526438,
      "learning_rate": 1.3171966735358343e-06,
      "loss": 0.0309,
      "step": 8080
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.17404322228099126,
      "learning_rate": 1.31465878306545e-06,
      "loss": 0.0214,
      "step": 8081
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21334523474351408,
      "learning_rate": 1.3121232567865793e-06,
      "loss": 0.03,
      "step": 8082
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2888267689993707,
      "learning_rate": 1.3095900950200235e-06,
      "loss": 0.0341,
      "step": 8083
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21944974628480754,
      "learning_rate": 1.3070592980862994e-06,
      "loss": 0.0291,
      "step": 8084
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19668181114224614,
      "learning_rate": 1.3045308663056179e-06,
      "loss": 0.0196,
      "step": 8085
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21699094248362136,
      "learning_rate": 1.3020047999979002e-06,
      "loss": 0.0417,
      "step": 8086
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18290534718347867,
      "learning_rate": 1.2994810994827511e-06,
      "loss": 0.0293,
      "step": 8087
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2280492016528875,
      "learning_rate": 1.2969597650794907e-06,
      "loss": 0.0276,
      "step": 8088
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20649149548669107,
      "learning_rate": 1.2944407971071326e-06,
      "loss": 0.0219,
      "step": 8089
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19503651193073768,
      "learning_rate": 1.2919241958843975e-06,
      "loss": 0.0282,
      "step": 8090
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22498393133853425,
      "learning_rate": 1.289409961729695e-06,
      "loss": 0.0326,
      "step": 8091
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1995833930557682,
      "learning_rate": 1.2868980949611486e-06,
      "loss": 0.0335,
      "step": 8092
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22862793534481746,
      "learning_rate": 1.2843885958965708e-06,
      "loss": 0.0369,
      "step": 8093
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2395499489509689,
      "learning_rate": 1.2818814648534895e-06,
      "loss": 0.0404,
      "step": 8094
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20635799484964087,
      "learning_rate": 1.279376702149111e-06,
      "loss": 0.0209,
      "step": 8095
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2387144341116745,
      "learning_rate": 1.2768743081003598e-06,
      "loss": 0.0399,
      "step": 8096
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.24765848555894632,
      "learning_rate": 1.2743742830238582e-06,
      "loss": 0.0439,
      "step": 8097
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.14794857667084915,
      "learning_rate": 1.2718766272359195e-06,
      "loss": 0.0186,
      "step": 8098
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.28252392583146135,
      "learning_rate": 1.269381341052569e-06,
      "loss": 0.0388,
      "step": 8099
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.16836228954509633,
      "learning_rate": 1.266888424789523e-06,
      "loss": 0.0242,
      "step": 8100
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.35681785032370694,
      "learning_rate": 1.2643978787622069e-06,
      "loss": 0.0457,
      "step": 8101
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19081107852267626,
      "learning_rate": 1.261909703285733e-06,
      "loss": 0.0302,
      "step": 8102
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22440415834462196,
      "learning_rate": 1.2594238986749273e-06,
      "loss": 0.047,
      "step": 8103
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.15405061239724732,
      "learning_rate": 1.2569404652443073e-06,
      "loss": 0.012,
      "step": 8104
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.16322141719218639,
      "learning_rate": 1.2544594033080947e-06,
      "loss": 0.0246,
      "step": 8105
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18345814627697715,
      "learning_rate": 1.2519807131802097e-06,
      "loss": 0.0366,
      "step": 8106
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22879683480226257,
      "learning_rate": 1.2495043951742703e-06,
      "loss": 0.0463,
      "step": 8107
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2875033160723305,
      "learning_rate": 1.2470304496035968e-06,
      "loss": 0.0374,
      "step": 8108
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18928059853146756,
      "learning_rate": 1.2445588767812145e-06,
      "loss": 0.0219,
      "step": 8109
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.24059148932862864,
      "learning_rate": 1.2420896770198355e-06,
      "loss": 0.0378,
      "step": 8110
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21259041713853652,
      "learning_rate": 1.239622850631881e-06,
      "loss": 0.0215,
      "step": 8111
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.16772253265649945,
      "learning_rate": 1.237158397929472e-06,
      "loss": 0.0139,
      "step": 8112
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2511345629468641,
      "learning_rate": 1.2346963192244287e-06,
      "loss": 0.0382,
      "step": 8113
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.26662164130422566,
      "learning_rate": 1.2322366148282638e-06,
      "loss": 0.0232,
      "step": 8114
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18192631803697984,
      "learning_rate": 1.2297792850521973e-06,
      "loss": 0.0137,
      "step": 8115
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22075382426784543,
      "learning_rate": 1.2273243302071513e-06,
      "loss": 0.0266,
      "step": 8116
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.17563211684553523,
      "learning_rate": 1.2248717506037377e-06,
      "loss": 0.0273,
      "step": 8117
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21438683560566324,
      "learning_rate": 1.2224215465522726e-06,
      "loss": 0.0377,
      "step": 8118
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1878684615706186,
      "learning_rate": 1.2199737183627746e-06,
      "loss": 0.0292,
      "step": 8119
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18451518290031707,
      "learning_rate": 1.2175282663449584e-06,
      "loss": 0.0377,
      "step": 8120
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18551300141175717,
      "learning_rate": 1.2150851908082406e-06,
      "loss": 0.0264,
      "step": 8121
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20999712077789623,
      "learning_rate": 1.2126444920617297e-06,
      "loss": 0.0331,
      "step": 8122
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20624544254238392,
      "learning_rate": 1.2102061704142408e-06,
      "loss": 0.0297,
      "step": 8123
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18398038248318357,
      "learning_rate": 1.2077702261742875e-06,
      "loss": 0.0365,
      "step": 8124
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21233545762937733,
      "learning_rate": 1.2053366596500849e-06,
      "loss": 0.0257,
      "step": 8125
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1739435402115767,
      "learning_rate": 1.2029054711495358e-06,
      "loss": 0.0298,
      "step": 8126
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18857947193574698,
      "learning_rate": 1.2004766609802543e-06,
      "loss": 0.0283,
      "step": 8127
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22744293802023172,
      "learning_rate": 1.19805022944955e-06,
      "loss": 0.0183,
      "step": 8128
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18339907693802224,
      "learning_rate": 1.1956261768644328e-06,
      "loss": 0.0322,
      "step": 8129
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.256323996769043,
      "learning_rate": 1.193204503531602e-06,
      "loss": 0.0281,
      "step": 8130
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21182444211174353,
      "learning_rate": 1.1907852097574723e-06,
      "loss": 0.0247,
      "step": 8131
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.15462338631295972,
      "learning_rate": 1.1883682958481413e-06,
      "loss": 0.0219,
      "step": 8132
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19340168846385952,
      "learning_rate": 1.1859537621094175e-06,
      "loss": 0.018,
      "step": 8133
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.16932275538190913,
      "learning_rate": 1.1835416088468033e-06,
      "loss": 0.0284,
      "step": 8134
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19175409317683567,
      "learning_rate": 1.1811318363654967e-06,
      "loss": 0.0241,
      "step": 8135
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2571991937048482,
      "learning_rate": 1.178724444970405e-06,
      "loss": 0.0273,
      "step": 8136
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2910050383824771,
      "learning_rate": 1.17631943496612e-06,
      "loss": 0.0371,
      "step": 8137
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.14879430376567432,
      "learning_rate": 1.1739168066569406e-06,
      "loss": 0.0186,
      "step": 8138
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1794712070992341,
      "learning_rate": 1.1715165603468637e-06,
      "loss": 0.0235,
      "step": 8139
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23144780733075693,
      "learning_rate": 1.1691186963395861e-06,
      "loss": 0.0279,
      "step": 8140
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18887376191966082,
      "learning_rate": 1.1667232149384989e-06,
      "loss": 0.025,
      "step": 8141
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19330697361051746,
      "learning_rate": 1.1643301164466926e-06,
      "loss": 0.0123,
      "step": 8142
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2536770593168057,
      "learning_rate": 1.1619394011669605e-06,
      "loss": 0.0323,
      "step": 8143
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18273740269312416,
      "learning_rate": 1.1595510694017943e-06,
      "loss": 0.0207,
      "step": 8144
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19219626662566153,
      "learning_rate": 1.1571651214533764e-06,
      "loss": 0.0237,
      "step": 8145
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2027404795262321,
      "learning_rate": 1.154781557623592e-06,
      "loss": 0.0376,
      "step": 8146
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.15081378309081103,
      "learning_rate": 1.1524003782140269e-06,
      "loss": 0.0162,
      "step": 8147
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1695457823629119,
      "learning_rate": 1.1500215835259664e-06,
      "loss": 0.0237,
      "step": 8148
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18068020722705433,
      "learning_rate": 1.1476451738603855e-06,
      "loss": 0.0391,
      "step": 8149
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.26530709597023083,
      "learning_rate": 1.1452711495179659e-06,
      "loss": 0.0447,
      "step": 8150
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18722748821243787,
      "learning_rate": 1.1428995107990892e-06,
      "loss": 0.0266,
      "step": 8151
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.22255358085661164,
      "learning_rate": 1.1405302580038224e-06,
      "loss": 0.0308,
      "step": 8152
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21457256194428165,
      "learning_rate": 1.1381633914319434e-06,
      "loss": 0.0347,
      "step": 8153
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.20973767675122582,
      "learning_rate": 1.1357989113829237e-06,
      "loss": 0.0484,
      "step": 8154
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2583803512229484,
      "learning_rate": 1.1334368181559351e-06,
      "loss": 0.0406,
      "step": 8155
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.25708935205554145,
      "learning_rate": 1.1310771120498386e-06,
      "loss": 0.0374,
      "step": 8156
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.28250638823755325,
      "learning_rate": 1.1287197933632022e-06,
      "loss": 0.0535,
      "step": 8157
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2538457358213464,
      "learning_rate": 1.1263648623942912e-06,
      "loss": 0.0276,
      "step": 8158
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.24326919672807157,
      "learning_rate": 1.124012319441068e-06,
      "loss": 0.0442,
      "step": 8159
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2387901016454865,
      "learning_rate": 1.1216621648011873e-06,
      "loss": 0.0316,
      "step": 8160
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.23196353116820476,
      "learning_rate": 1.1193143987720067e-06,
      "loss": 0.0175,
      "step": 8161
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18492958383161237,
      "learning_rate": 1.1169690216505846e-06,
      "loss": 0.0284,
      "step": 8162
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2742841283997708,
      "learning_rate": 1.11462603373367e-06,
      "loss": 0.0291,
      "step": 8163
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.26405287166399877,
      "learning_rate": 1.1122854353177171e-06,
      "loss": 0.0395,
      "step": 8164
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.23485040926803213,
      "learning_rate": 1.1099472266988686e-06,
      "loss": 0.0286,
      "step": 8165
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.20421760399693153,
      "learning_rate": 1.1076114081729682e-06,
      "loss": 0.0288,
      "step": 8166
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.20372889687362652,
      "learning_rate": 1.1052779800355707e-06,
      "loss": 0.0279,
      "step": 8167
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.22648018953180518,
      "learning_rate": 1.1029469425819039e-06,
      "loss": 0.0381,
      "step": 8168
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1658447283595015,
      "learning_rate": 1.1006182961069123e-06,
      "loss": 0.018,
      "step": 8169
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18974680603490496,
      "learning_rate": 1.0982920409052312e-06,
      "loss": 0.0295,
      "step": 8170
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19420371997154975,
      "learning_rate": 1.095968177271194e-06,
      "loss": 0.0351,
      "step": 8171
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19263376047461428,
      "learning_rate": 1.0936467054988276e-06,
      "loss": 0.0316,
      "step": 8172
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18404661288130614,
      "learning_rate": 1.0913276258818617e-06,
      "loss": 0.0232,
      "step": 8173
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2304224032336941,
      "learning_rate": 1.0890109387137216e-06,
      "loss": 0.0368,
      "step": 8174
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.209428653674787,
      "learning_rate": 1.0866966442875304e-06,
      "loss": 0.0349,
      "step": 8175
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18893556589820637,
      "learning_rate": 1.0843847428961074e-06,
      "loss": 0.0259,
      "step": 8176
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21481023006777086,
      "learning_rate": 1.0820752348319673e-06,
      "loss": 0.0525,
      "step": 8177
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19360318851323052,
      "learning_rate": 1.0797681203873255e-06,
      "loss": 0.0239,
      "step": 8178
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.24797456543418325,
      "learning_rate": 1.0774633998540974e-06,
      "loss": 0.0532,
      "step": 8179
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18944382398917967,
      "learning_rate": 1.0751610735238848e-06,
      "loss": 0.0417,
      "step": 8180
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2727139573115948,
      "learning_rate": 1.0728611416879975e-06,
      "loss": 0.0505,
      "step": 8181
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19017606621006206,
      "learning_rate": 1.0705636046374334e-06,
      "loss": 0.0157,
      "step": 8182
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2707596030500379,
      "learning_rate": 1.0682684626629002e-06,
      "loss": 0.0323,
      "step": 8183
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.195041926840636,
      "learning_rate": 1.0659757160547813e-06,
      "loss": 0.0266,
      "step": 8184
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2631091293840357,
      "learning_rate": 1.0636853651031797e-06,
      "loss": 0.0362,
      "step": 8185
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2532716597024558,
      "learning_rate": 1.0613974100978885e-06,
      "loss": 0.036,
      "step": 8186
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.22369384765625,
      "learning_rate": 1.059111851328385e-06,
      "loss": 0.0226,
      "step": 8187
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.23997596861310688,
      "learning_rate": 1.0568286890838575e-06,
      "loss": 0.0256,
      "step": 8188
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18300212741355235,
      "learning_rate": 1.0545479236531885e-06,
      "loss": 0.0205,
      "step": 8189
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40957907984294883,
      "learning_rate": 1.0522695553249562e-06,
      "loss": 0.0525,
      "step": 8190
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.32759796686577936,
      "learning_rate": 1.0499935843874275e-06,
      "loss": 0.0509,
      "step": 8191
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2185881390509298,
      "learning_rate": 1.047720011128579e-06,
      "loss": 0.0341,
      "step": 8192
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19067662196731774,
      "learning_rate": 1.045448835836076e-06,
      "loss": 0.0282,
      "step": 8193
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19990667609592316,
      "learning_rate": 1.0431800587972862e-06,
      "loss": 0.0164,
      "step": 8194
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2125035362790888,
      "learning_rate": 1.0409136802992647e-06,
      "loss": 0.0379,
      "step": 8195
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21321329110056897,
      "learning_rate": 1.038649700628771e-06,
      "loss": 0.0351,
      "step": 8196
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.25643445679813426,
      "learning_rate": 1.036388120072258e-06,
      "loss": 0.0364,
      "step": 8197
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.20116294915109967,
      "learning_rate": 1.0341289389158793e-06,
      "loss": 0.0245,
      "step": 8198
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19956297213781976,
      "learning_rate": 1.031872157445477e-06,
      "loss": 0.0245,
      "step": 8199
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21598542055733674,
      "learning_rate": 1.029617775946592e-06,
      "loss": 0.0294,
      "step": 8200
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2232396530314543,
      "learning_rate": 1.0273657947044647e-06,
      "loss": 0.0465,
      "step": 8201
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21620458164539047,
      "learning_rate": 1.0251162140040383e-06,
      "loss": 0.0153,
      "step": 8202
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1647575731693121,
      "learning_rate": 1.022869034129934e-06,
      "loss": 0.0153,
      "step": 8203
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2184068935192134,
      "learning_rate": 1.0206242553664868e-06,
      "loss": 0.0301,
      "step": 8204
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3173511782799954,
      "learning_rate": 1.0183818779977162e-06,
      "loss": 0.0494,
      "step": 8205
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1920471652103783,
      "learning_rate": 1.016141902307346e-06,
      "loss": 0.0299,
      "step": 8206
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19413349975159716,
      "learning_rate": 1.0139043285787897e-06,
      "loss": 0.0383,
      "step": 8207
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19461334977901326,
      "learning_rate": 1.011669157095161e-06,
      "loss": 0.0353,
      "step": 8208
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18873129317758788,
      "learning_rate": 1.0094363881392665e-06,
      "loss": 0.0206,
      "step": 8209
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.22791468841442727,
      "learning_rate": 1.0072060219936164e-06,
      "loss": 0.0225,
      "step": 8210
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.24479566056801333,
      "learning_rate": 1.0049780589404045e-06,
      "loss": 0.034,
      "step": 8211
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21795623531442687,
      "learning_rate": 1.002752499261528e-06,
      "loss": 0.0377,
      "step": 8212
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19785990758621003,
      "learning_rate": 1.0005293432385832e-06,
      "loss": 0.0366,
      "step": 8213
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2252275170940452,
      "learning_rate": 9.98308591152859e-07,
      "loss": 0.0393,
      "step": 8214
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2640023033103023,
      "learning_rate": 9.960902432853349e-07,
      "loss": 0.0233,
      "step": 8215
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.300959497125171,
      "learning_rate": 9.93874299916693e-07,
      "loss": 0.0412,
      "step": 8216
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21366304731527114,
      "learning_rate": 9.916607613273067e-07,
      "loss": 0.0374,
      "step": 8217
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19288711481311144,
      "learning_rate": 9.894496277972498e-07,
      "loss": 0.0333,
      "step": 8218
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19007976366646723,
      "learning_rate": 9.8724089960629e-07,
      "loss": 0.0305,
      "step": 8219
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19160212012160233,
      "learning_rate": 9.850345770338875e-07,
      "loss": 0.0343,
      "step": 8220
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.24360752649115458,
      "learning_rate": 9.828306603592063e-07,
      "loss": 0.044,
      "step": 8221
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19712945751890557,
      "learning_rate": 9.80629149861092e-07,
      "loss": 0.0278,
      "step": 8222
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19936790679403657,
      "learning_rate": 9.784300458180994e-07,
      "loss": 0.0357,
      "step": 8223
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21959241374770835,
      "learning_rate": 9.76233348508473e-07,
      "loss": 0.0282,
      "step": 8224
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.20111109917353348,
      "learning_rate": 9.740390582101545e-07,
      "loss": 0.0254,
      "step": 8225
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.16654274786693715,
      "learning_rate": 9.718471752007753e-07,
      "loss": 0.02,
      "step": 8226
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18185214575819714,
      "learning_rate": 9.69657699757669e-07,
      "loss": 0.0313,
      "step": 8227
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.17936204378342663,
      "learning_rate": 9.67470632157863e-07,
      "loss": 0.015,
      "step": 8228
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19407313994941713,
      "learning_rate": 9.652859726780827e-07,
      "loss": 0.0285,
      "step": 8229
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1796495874463076,
      "learning_rate": 9.63103721594738e-07,
      "loss": 0.03,
      "step": 8230
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3431152856212495,
      "learning_rate": 9.60923879183946e-07,
      "loss": 0.0417,
      "step": 8231
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.217390373557757,
      "learning_rate": 9.587464457215146e-07,
      "loss": 0.0273,
      "step": 8232
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1697552949665261,
      "learning_rate": 9.565714214829503e-07,
      "loss": 0.0224,
      "step": 8233
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.24912640555837498,
      "learning_rate": 9.54398806743444e-07,
      "loss": 0.0255,
      "step": 8234
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.243218396715547,
      "learning_rate": 9.522286017778892e-07,
      "loss": 0.0252,
      "step": 8235
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2819960685829629,
      "learning_rate": 9.500608068608841e-07,
      "loss": 0.0345,
      "step": 8236
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.22887833656179069,
      "learning_rate": 9.478954222667025e-07,
      "loss": 0.0307,
      "step": 8237
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21275178073895368,
      "learning_rate": 9.457324482693275e-07,
      "loss": 0.0264,
      "step": 8238
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.26779351301779925,
      "learning_rate": 9.435718851424336e-07,
      "loss": 0.0458,
      "step": 8239
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21481218973181224,
      "learning_rate": 9.414137331593842e-07,
      "loss": 0.0438,
      "step": 8240
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.30718018508537476,
      "learning_rate": 9.392579925932499e-07,
      "loss": 0.02,
      "step": 8241
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1616745367748136,
      "learning_rate": 9.371046637167835e-07,
      "loss": 0.0213,
      "step": 8242
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1943421866020821,
      "learning_rate": 9.349537468024406e-07,
      "loss": 0.0319,
      "step": 8243
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1801895611876995,
      "learning_rate": 9.328052421223676e-07,
      "loss": 0.0197,
      "step": 8244
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23448535387515215,
      "learning_rate": 9.306591499484119e-07,
      "loss": 0.034,
      "step": 8245
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19631257556888324,
      "learning_rate": 9.285154705521048e-07,
      "loss": 0.033,
      "step": 8246
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.16547725561072543,
      "learning_rate": 9.26374204204683e-07,
      "loss": 0.0198,
      "step": 8247
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.22447749702971492,
      "learning_rate": 9.242353511770697e-07,
      "loss": 0.0315,
      "step": 8248
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.25576799522642574,
      "learning_rate": 9.220989117398926e-07,
      "loss": 0.0225,
      "step": 8249
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19572030408919244,
      "learning_rate": 9.199648861634625e-07,
      "loss": 0.032,
      "step": 8250
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19454685822648343,
      "learning_rate": 9.17833274717792e-07,
      "loss": 0.0164,
      "step": 8251
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18337977906560887,
      "learning_rate": 9.157040776725856e-07,
      "loss": 0.0303,
      "step": 8252
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20508025032032087,
      "learning_rate": 9.135772952972455e-07,
      "loss": 0.0241,
      "step": 8253
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2032347895947305,
      "learning_rate": 9.11452927860863e-07,
      "loss": 0.034,
      "step": 8254
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23185767286395545,
      "learning_rate": 9.0933097563223e-07,
      "loss": 0.0352,
      "step": 8255
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21107869367743845,
      "learning_rate": 9.072114388798314e-07,
      "loss": 0.0164,
      "step": 8256
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1812799420782213,
      "learning_rate": 9.050943178718396e-07,
      "loss": 0.0213,
      "step": 8257
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23501533614245435,
      "learning_rate": 9.029796128761292e-07,
      "loss": 0.0462,
      "step": 8258
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17987446801860468,
      "learning_rate": 9.008673241602639e-07,
      "loss": 0.0134,
      "step": 8259
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21500580484571444,
      "learning_rate": 8.987574519915121e-07,
      "loss": 0.0221,
      "step": 8260
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2875447507942639,
      "learning_rate": 8.966499966368202e-07,
      "loss": 0.0324,
      "step": 8261
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17285456358464918,
      "learning_rate": 8.945449583628396e-07,
      "loss": 0.0219,
      "step": 8262
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2186172627584308,
      "learning_rate": 8.924423374359148e-07,
      "loss": 0.0399,
      "step": 8263
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.16168439265186016,
      "learning_rate": 8.903421341220842e-07,
      "loss": 0.0228,
      "step": 8264
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21596027168629722,
      "learning_rate": 8.882443486870751e-07,
      "loss": 0.0226,
      "step": 8265
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19266367515121915,
      "learning_rate": 8.861489813963154e-07,
      "loss": 0.0274,
      "step": 8266
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17407398303282848,
      "learning_rate": 8.840560325149261e-07,
      "loss": 0.0195,
      "step": 8267
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.25294432726582544,
      "learning_rate": 8.819655023077201e-07,
      "loss": 0.0463,
      "step": 8268
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2548247377143722,
      "learning_rate": 8.798773910392033e-07,
      "loss": 0.0439,
      "step": 8269
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18935253973235466,
      "learning_rate": 8.777916989735736e-07,
      "loss": 0.0262,
      "step": 8270
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1994803898394527,
      "learning_rate": 8.757084263747373e-07,
      "loss": 0.032,
      "step": 8271
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18605138552705197,
      "learning_rate": 8.736275735062749e-07,
      "loss": 0.0152,
      "step": 8272
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3722836340984241,
      "learning_rate": 8.715491406314713e-07,
      "loss": 0.0647,
      "step": 8273
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2057271949255089,
      "learning_rate": 8.694731280133051e-07,
      "loss": 0.0281,
      "step": 8274
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1838929092734693,
      "learning_rate": 8.673995359144483e-07,
      "loss": 0.0205,
      "step": 8275
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21421806084920603,
      "learning_rate": 8.653283645972598e-07,
      "loss": 0.0257,
      "step": 8276
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1926045471233725,
      "learning_rate": 8.632596143237992e-07,
      "loss": 0.0352,
      "step": 8277
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23870126259245716,
      "learning_rate": 8.611932853558236e-07,
      "loss": 0.0531,
      "step": 8278
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2004222235203365,
      "learning_rate": 8.591293779547727e-07,
      "loss": 0.0179,
      "step": 8279
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18264973192357153,
      "learning_rate": 8.570678923817888e-07,
      "loss": 0.0225,
      "step": 8280
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.26262541908256143,
      "learning_rate": 8.550088288977032e-07,
      "loss": 0.0472,
      "step": 8281
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.25773529861992367,
      "learning_rate": 8.529521877630409e-07,
      "loss": 0.0334,
      "step": 8282
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1868243162451725,
      "learning_rate": 8.508979692380249e-07,
      "loss": 0.0273,
      "step": 8283
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1743671760034363,
      "learning_rate": 8.48846173582567e-07,
      "loss": 0.0215,
      "step": 8284
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.22221440259682573,
      "learning_rate": 8.467968010562711e-07,
      "loss": 0.036,
      "step": 8285
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2816999862270061,
      "learning_rate": 8.447498519184405e-07,
      "loss": 0.0304,
      "step": 8286
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.15049164339402962,
      "learning_rate": 8.42705326428066e-07,
      "loss": 0.0201,
      "step": 8287
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.198307806345388,
      "learning_rate": 8.406632248438362e-07,
      "loss": 0.0191,
      "step": 8288
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20663963923372394,
      "learning_rate": 8.386235474241311e-07,
      "loss": 0.0309,
      "step": 8289
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.26835280719921417,
      "learning_rate": 8.365862944270243e-07,
      "loss": 0.0494,
      "step": 8290
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.25674922915420034,
      "learning_rate": 8.345514661102827e-07,
      "loss": 0.0389,
      "step": 8291
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18573245756093892,
      "learning_rate": 8.325190627313628e-07,
      "loss": 0.0307,
      "step": 8292
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17700843536082836,
      "learning_rate": 8.304890845474189e-07,
      "loss": 0.0328,
      "step": 8293
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20095938969712215,
      "learning_rate": 8.284615318152988e-07,
      "loss": 0.0246,
      "step": 8294
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23885577931769858,
      "learning_rate": 8.264364047915441e-07,
      "loss": 0.0273,
      "step": 8295
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2234165695364045,
      "learning_rate": 8.244137037323807e-07,
      "loss": 0.0341,
      "step": 8296
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23234503942566662,
      "learning_rate": 8.223934288937374e-07,
      "loss": 0.0344,
      "step": 8297
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21781043521006416,
      "learning_rate": 8.203755805312319e-07,
      "loss": 0.032,
      "step": 8298
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2640930767526961,
      "learning_rate": 8.1836015890018e-07,
      "loss": 0.0475,
      "step": 8299
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2025630127477468,
      "learning_rate": 8.163471642555798e-07,
      "loss": 0.0199,
      "step": 8300
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2611568168843046,
      "learning_rate": 8.143365968521277e-07,
      "loss": 0.0302,
      "step": 8301
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.22005412798652046,
      "learning_rate": 8.123284569442203e-07,
      "loss": 0.035,
      "step": 8302
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21233357160200297,
      "learning_rate": 8.103227447859385e-07,
      "loss": 0.0329,
      "step": 8303
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.22018050033830233,
      "learning_rate": 8.083194606310507e-07,
      "loss": 0.0305,
      "step": 8304
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2089986439762432,
      "learning_rate": 8.063186047330362e-07,
      "loss": 0.0469,
      "step": 8305
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18398102030358274,
      "learning_rate": 8.043201773450526e-07,
      "loss": 0.0301,
      "step": 8306
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1939473124206015,
      "learning_rate": 8.02324178719951e-07,
      "loss": 0.0376,
      "step": 8307
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2534852131616495,
      "learning_rate": 8.003306091102803e-07,
      "loss": 0.0491,
      "step": 8308
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.14857831349302236,
      "learning_rate": 7.983394687682811e-07,
      "loss": 0.0228,
      "step": 8309
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.24847488571870194,
      "learning_rate": 7.963507579458851e-07,
      "loss": 0.0276,
      "step": 8310
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21501468447505942,
      "learning_rate": 7.943644768947157e-07,
      "loss": 0.03,
      "step": 8311
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18173029952275782,
      "learning_rate": 7.923806258660893e-07,
      "loss": 0.0157,
      "step": 8312
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.22132643731655707,
      "learning_rate": 7.903992051110188e-07,
      "loss": 0.025,
      "step": 8313
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2245354572239109,
      "learning_rate": 7.884202148802056e-07,
      "loss": 0.028,
      "step": 8314
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19612606359865667,
      "learning_rate": 7.864436554240429e-07,
      "loss": 0.0263,
      "step": 8315
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.22572838220399857,
      "learning_rate": 7.844695269926194e-07,
      "loss": 0.0206,
      "step": 8316
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.165413949886668,
      "learning_rate": 7.824978298357111e-07,
      "loss": 0.028,
      "step": 8317
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1912501730325951,
      "learning_rate": 7.805285642027983e-07,
      "loss": 0.0259,
      "step": 8318
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23225523480509927,
      "learning_rate": 7.78561730343037e-07,
      "loss": 0.0487,
      "step": 8319
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21158025969900446,
      "learning_rate": 7.765973285052863e-07,
      "loss": 0.0262,
      "step": 8320
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20679104563672152,
      "learning_rate": 7.746353589380962e-07,
      "loss": 0.0252,
      "step": 8321
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2098543916987807,
      "learning_rate": 7.726758218897079e-07,
      "loss": 0.0331,
      "step": 8322
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17279907550371298,
      "learning_rate": 7.707187176080544e-07,
      "loss": 0.0212,
      "step": 8323
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.25404107842219026,
      "learning_rate": 7.687640463407597e-07,
      "loss": 0.0337,
      "step": 8324
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19785788357457332,
      "learning_rate": 7.668118083351461e-07,
      "loss": 0.0396,
      "step": 8325
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17696038119699498,
      "learning_rate": 7.648620038382204e-07,
      "loss": 0.0213,
      "step": 8326
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2119630150573829,
      "learning_rate": 7.629146330966853e-07,
      "loss": 0.0223,
      "step": 8327
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21826744498778183,
      "learning_rate": 7.609696963569325e-07,
      "loss": 0.0316,
      "step": 8328
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23212229230202275,
      "learning_rate": 7.590271938650518e-07,
      "loss": 0.0397,
      "step": 8329
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.24391809621680238,
      "learning_rate": 7.5708712586682e-07,
      "loss": 0.0232,
      "step": 8330
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.16508441505056867,
      "learning_rate": 7.551494926077052e-07,
      "loss": 0.0211,
      "step": 8331
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19428187246040374,
      "learning_rate": 7.532142943328713e-07,
      "loss": 0.0266,
      "step": 8332
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2518836968153419,
      "learning_rate": 7.512815312871735e-07,
      "loss": 0.0548,
      "step": 8333
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.25576458696677545,
      "learning_rate": 7.493512037151563e-07,
      "loss": 0.028,
      "step": 8334
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.23377185939891926,
      "learning_rate": 7.474233118610553e-07,
      "loss": 0.0296,
      "step": 8335
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.16079145165600636,
      "learning_rate": 7.454978559688019e-07,
      "loss": 0.0185,
      "step": 8336
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20114282750162718,
      "learning_rate": 7.435748362820172e-07,
      "loss": 0.0367,
      "step": 8337
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2099352177680145,
      "learning_rate": 7.416542530440174e-07,
      "loss": 0.0267,
      "step": 8338
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.17064276507575793,
      "learning_rate": 7.39736106497797e-07,
      "loss": 0.0263,
      "step": 8339
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2907342894984491,
      "learning_rate": 7.378203968860643e-07,
      "loss": 0.0428,
      "step": 8340
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2429870129434039,
      "learning_rate": 7.359071244512051e-07,
      "loss": 0.037,
      "step": 8341
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19689027560478325,
      "learning_rate": 7.339962894352925e-07,
      "loss": 0.0417,
      "step": 8342
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.22965578781540286,
      "learning_rate": 7.320878920801044e-07,
      "loss": 0.0275,
      "step": 8343
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20803986781456973,
      "learning_rate": 7.30181932627101e-07,
      "loss": 0.0405,
      "step": 8344
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18372696243919345,
      "learning_rate": 7.282784113174379e-07,
      "loss": 0.0279,
      "step": 8345
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2088753959793547,
      "learning_rate": 7.263773283919584e-07,
      "loss": 0.0337,
      "step": 8346
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2161109055407055,
      "learning_rate": 7.244786840912033e-07,
      "loss": 0.0248,
      "step": 8347
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2292522880966251,
      "learning_rate": 7.225824786553981e-07,
      "loss": 0.0314,
      "step": 8348
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20722740705283857,
      "learning_rate": 7.206887123244687e-07,
      "loss": 0.0269,
      "step": 8349
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21685167427663435,
      "learning_rate": 7.187973853380215e-07,
      "loss": 0.0273,
      "step": 8350
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21911285312810636,
      "learning_rate": 7.169084979353603e-07,
      "loss": 0.035,
      "step": 8351
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.16650670292538797,
      "learning_rate": 7.150220503554783e-07,
      "loss": 0.0168,
      "step": 8352
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.23844254615863555,
      "learning_rate": 7.131380428370671e-07,
      "loss": 0.0413,
      "step": 8353
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19736905738347355,
      "learning_rate": 7.112564756184981e-07,
      "loss": 0.022,
      "step": 8354
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20810555682276666,
      "learning_rate": 7.093773489378408e-07,
      "loss": 0.016,
      "step": 8355
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2405570507096395,
      "learning_rate": 7.075006630328518e-07,
      "loss": 0.0267,
      "step": 8356
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.22532994331106057,
      "learning_rate": 7.056264181409922e-07,
      "loss": 0.0459,
      "step": 8357
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.17066836537886965,
      "learning_rate": 7.037546144993901e-07,
      "loss": 0.0169,
      "step": 8358
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.25722549542529416,
      "learning_rate": 7.018852523448871e-07,
      "loss": 0.0261,
      "step": 8359
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.23987468681493912,
      "learning_rate": 7.000183319140053e-07,
      "loss": 0.032,
      "step": 8360
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.26530156359019463,
      "learning_rate": 6.981538534429599e-07,
      "loss": 0.0469,
      "step": 8361
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2248861514915893,
      "learning_rate": 6.962918171676536e-07,
      "loss": 0.0313,
      "step": 8362
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2499288144687979,
      "learning_rate": 6.944322233236844e-07,
      "loss": 0.0292,
      "step": 8363
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2569748390466309,
      "learning_rate": 6.925750721463443e-07,
      "loss": 0.0369,
      "step": 8364
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2111929300631266,
      "learning_rate": 6.907203638706094e-07,
      "loss": 0.0318,
      "step": 8365
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.25302257820450147,
      "learning_rate": 6.88868098731148e-07,
      "loss": 0.0531,
      "step": 8366
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2178474522873738,
      "learning_rate": 6.870182769623235e-07,
      "loss": 0.0211,
      "step": 8367
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3198811254760812,
      "learning_rate": 6.851708987981865e-07,
      "loss": 0.0396,
      "step": 8368
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.22959975298708743,
      "learning_rate": 6.833259644724811e-07,
      "loss": 0.019,
      "step": 8369
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.23707362098666884,
      "learning_rate": 6.814834742186361e-07,
      "loss": 0.0251,
      "step": 8370
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21209947763187123,
      "learning_rate": 6.796434282697783e-07,
      "loss": 0.0453,
      "step": 8371
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2294151176741445,
      "learning_rate": 6.778058268587217e-07,
      "loss": 0.0443,
      "step": 8372
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.220088863496754,
      "learning_rate": 6.759706702179713e-07,
      "loss": 0.0297,
      "step": 8373
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20209323663795287,
      "learning_rate": 6.741379585797236e-07,
      "loss": 0.0245,
      "step": 8374
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.26108945088525415,
      "learning_rate": 6.723076921758664e-07,
      "loss": 0.0408,
      "step": 8375
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21960898750458263,
      "learning_rate": 6.704798712379768e-07,
      "loss": 0.0341,
      "step": 8376
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1982818244432766,
      "learning_rate": 6.686544959973207e-07,
      "loss": 0.023,
      "step": 8377
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21411860941877806,
      "learning_rate": 6.66831566684858e-07,
      "loss": 0.0387,
      "step": 8378
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20386571810167772,
      "learning_rate": 6.650110835312351e-07,
      "loss": 0.0232,
      "step": 8379
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.15969013795148154,
      "learning_rate": 6.631930467667991e-07,
      "loss": 0.0291,
      "step": 8380
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19799292918786857,
      "learning_rate": 6.61377456621568e-07,
      "loss": 0.0307,
      "step": 8381
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2592476578149597,
      "learning_rate": 6.595643133252716e-07,
      "loss": 0.026,
      "step": 8382
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.16889081748593573,
      "learning_rate": 6.577536171073173e-07,
      "loss": 0.0282,
      "step": 8383
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1862055181183217,
      "learning_rate": 6.559453681968064e-07,
      "loss": 0.0296,
      "step": 8384
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1676830756723088,
      "learning_rate": 6.541395668225314e-07,
      "loss": 0.0212,
      "step": 8385
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2024904113577425,
      "learning_rate": 6.523362132129718e-07,
      "loss": 0.029,
      "step": 8386
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1573365934159292,
      "learning_rate": 6.505353075963005e-07,
      "loss": 0.0237,
      "step": 8387
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.32704181356998224,
      "learning_rate": 6.487368502003821e-07,
      "loss": 0.0248,
      "step": 8388
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1808261144551506,
      "learning_rate": 6.469408412527656e-07,
      "loss": 0.0333,
      "step": 8389
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2132033492269207,
      "learning_rate": 6.451472809806958e-07,
      "loss": 0.0214,
      "step": 8390
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21403256645583105,
      "learning_rate": 6.433561696111046e-07,
      "loss": 0.0216,
      "step": 8391
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.198098821429767,
      "learning_rate": 6.415675073706174e-07,
      "loss": 0.0249,
      "step": 8392
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.22633907383806465,
      "learning_rate": 6.397812944855464e-07,
      "loss": 0.0411,
      "step": 8393
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.22249488671995937,
      "learning_rate": 6.379975311818931e-07,
      "loss": 0.0456,
      "step": 8394
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.16418217653706288,
      "learning_rate": 6.362162176853548e-07,
      "loss": 0.0227,
      "step": 8395
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21776450784619336,
      "learning_rate": 6.344373542213112e-07,
      "loss": 0.0319,
      "step": 8396
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21109391523759927,
      "learning_rate": 6.326609410148354e-07,
      "loss": 0.0262,
      "step": 8397
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1787576041238118,
      "learning_rate": 6.308869782906946e-07,
      "loss": 0.0171,
      "step": 8398
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20020439313437643,
      "learning_rate": 6.291154662733379e-07,
      "loss": 0.0226,
      "step": 8399
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.24270355710390398,
      "learning_rate": 6.27346405186915e-07,
      "loss": 0.0339,
      "step": 8400
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1993100387251412,
      "learning_rate": 6.255797952552511e-07,
      "loss": 0.0214,
      "step": 8401
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.25237342612733243,
      "learning_rate": 6.238156367018744e-07,
      "loss": 0.0428,
      "step": 8402
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20536148839700594,
      "learning_rate": 6.22053929749995e-07,
      "loss": 0.031,
      "step": 8403
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.228785152105541,
      "learning_rate": 6.202946746225191e-07,
      "loss": 0.0433,
      "step": 8404
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20532077784231703,
      "learning_rate": 6.185378715420331e-07,
      "loss": 0.0273,
      "step": 8405
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1828695277577136,
      "learning_rate": 6.16783520730826e-07,
      "loss": 0.0203,
      "step": 8406
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19333701534265987,
      "learning_rate": 6.150316224108643e-07,
      "loss": 0.0418,
      "step": 8407
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20727168822379813,
      "learning_rate": 6.132821768038133e-07,
      "loss": 0.0558,
      "step": 8408
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.17836172870113923,
      "learning_rate": 6.115351841310224e-07,
      "loss": 0.0252,
      "step": 8409
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1579723146170363,
      "learning_rate": 6.097906446135349e-07,
      "loss": 0.0266,
      "step": 8410
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.24195380626184704,
      "learning_rate": 6.080485584720808e-07,
      "loss": 0.0278,
      "step": 8411
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.24683755156020554,
      "learning_rate": 6.063089259270749e-07,
      "loss": 0.0331,
      "step": 8412
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20206476405616014,
      "learning_rate": 6.045717471986345e-07,
      "loss": 0.0304,
      "step": 8413
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1552739053395259,
      "learning_rate": 6.028370225065527e-07,
      "loss": 0.025,
      "step": 8414
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20924453941065727,
      "learning_rate": 6.011047520703228e-07,
      "loss": 0.0242,
      "step": 8415
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1827767842608636,
      "learning_rate": 5.993749361091206e-07,
      "loss": 0.0279,
      "step": 8416
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19183940920817746,
      "learning_rate": 5.976475748418131e-07,
      "loss": 0.0295,
      "step": 8417
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.16988337288854205,
      "learning_rate": 5.95922668486959e-07,
      "loss": 0.0214,
      "step": 8418
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.17770318396897788,
      "learning_rate": 5.942002172628058e-07,
      "loss": 0.0284,
      "step": 8419
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20335425680770988,
      "learning_rate": 5.92480221387286e-07,
      "loss": 0.0319,
      "step": 8420
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20325146003083785,
      "learning_rate": 5.907626810780276e-07,
      "loss": 0.0237,
      "step": 8421
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20049094724671024,
      "learning_rate": 5.890475965523412e-07,
      "loss": 0.0222,
      "step": 8422
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.24296646055755877,
      "learning_rate": 5.873349680272356e-07,
      "loss": 0.0301,
      "step": 8423
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19479930211638352,
      "learning_rate": 5.856247957193995e-07,
      "loss": 0.0263,
      "step": 8424
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2058500656976333,
      "learning_rate": 5.839170798452154e-07,
      "loss": 0.0475,
      "step": 8425
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1690838637048341,
      "learning_rate": 5.822118206207594e-07,
      "loss": 0.0207,
      "step": 8426
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2052925623614891,
      "learning_rate": 5.805090182617878e-07,
      "loss": 0.0147,
      "step": 8427
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.26213314215839717,
      "learning_rate": 5.788086729837505e-07,
      "loss": 0.0251,
      "step": 8428
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22363364540178207,
      "learning_rate": 5.771107850017865e-07,
      "loss": 0.0355,
      "step": 8429
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.25182848072447,
      "learning_rate": 5.754153545307262e-07,
      "loss": 0.0399,
      "step": 8430
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19509452100115968,
      "learning_rate": 5.737223817850845e-07,
      "loss": 0.031,
      "step": 8431
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22119743557895566,
      "learning_rate": 5.720318669790636e-07,
      "loss": 0.0248,
      "step": 8432
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18563328788699343,
      "learning_rate": 5.703438103265657e-07,
      "loss": 0.0345,
      "step": 8433
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21678475182806214,
      "learning_rate": 5.68658212041171e-07,
      "loss": 0.0338,
      "step": 8434
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2545525760978105,
      "learning_rate": 5.669750723361533e-07,
      "loss": 0.0231,
      "step": 8435
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1840065717974786,
      "learning_rate": 5.652943914244713e-07,
      "loss": 0.018,
      "step": 8436
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1961086639872842,
      "learning_rate": 5.636161695187792e-07,
      "loss": 0.0271,
      "step": 8437
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20151802947479283,
      "learning_rate": 5.61940406831416e-07,
      "loss": 0.0312,
      "step": 8438
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21139075008842315,
      "learning_rate": 5.602671035744123e-07,
      "loss": 0.0322,
      "step": 8439
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3327968522096054,
      "learning_rate": 5.585962599594807e-07,
      "loss": 0.0342,
      "step": 8440
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.17186633543369212,
      "learning_rate": 5.569278761980301e-07,
      "loss": 0.0172,
      "step": 8441
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.24979978232977448,
      "learning_rate": 5.552619525011538e-07,
      "loss": 0.0253,
      "step": 8442
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.23240423536280208,
      "learning_rate": 5.535984890796365e-07,
      "loss": 0.0196,
      "step": 8443
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20138871441391443,
      "learning_rate": 5.519374861439497e-07,
      "loss": 0.0474,
      "step": 8444
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3246919671486572,
      "learning_rate": 5.502789439042566e-07,
      "loss": 0.0424,
      "step": 8445
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20651024814003943,
      "learning_rate": 5.486228625704049e-07,
      "loss": 0.0429,
      "step": 8446
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.24252174754304018,
      "learning_rate": 5.469692423519335e-07,
      "loss": 0.0347,
      "step": 8447
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20464762961848362,
      "learning_rate": 5.453180834580663e-07,
      "loss": 0.0232,
      "step": 8448
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.27053622155139706,
      "learning_rate": 5.436693860977227e-07,
      "loss": 0.0483,
      "step": 8449
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18661866079168807,
      "learning_rate": 5.42023150479507e-07,
      "loss": 0.0254,
      "step": 8450
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22425238393602995,
      "learning_rate": 5.40379376811706e-07,
      "loss": 0.0322,
      "step": 8451
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.25532887330426185,
      "learning_rate": 5.387380653023066e-07,
      "loss": 0.0364,
      "step": 8452
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.14500897301665297,
      "learning_rate": 5.37099216158976e-07,
      "loss": 0.019,
      "step": 8453
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.183656620867129,
      "learning_rate": 5.354628295890729e-07,
      "loss": 0.0283,
      "step": 8454
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18354898272195813,
      "learning_rate": 5.338289057996426e-07,
      "loss": 0.0306,
      "step": 8455
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19892808803293913,
      "learning_rate": 5.321974449974198e-07,
      "loss": 0.0208,
      "step": 8456
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20859031424389596,
      "learning_rate": 5.305684473888284e-07,
      "loss": 0.0304,
      "step": 8457
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.17360973569537094,
      "learning_rate": 5.289419131799811e-07,
      "loss": 0.0235,
      "step": 8458
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2081914816171872,
      "learning_rate": 5.273178425766734e-07,
      "loss": 0.0158,
      "step": 8459
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18857801009821112,
      "learning_rate": 5.256962357843942e-07,
      "loss": 0.0275,
      "step": 8460
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1888262514014326,
      "learning_rate": 5.240770930083261e-07,
      "loss": 0.0263,
      "step": 8461
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21525795629131933,
      "learning_rate": 5.224604144533274e-07,
      "loss": 0.033,
      "step": 8462
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2398874522600459,
      "learning_rate": 5.208462003239522e-07,
      "loss": 0.0323,
      "step": 8463
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18715514815753465,
      "learning_rate": 5.192344508244418e-07,
      "loss": 0.0257,
      "step": 8464
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18205498020882127,
      "learning_rate": 5.176251661587262e-07,
      "loss": 0.0334,
      "step": 8465
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.25190873457328483,
      "learning_rate": 5.160183465304203e-07,
      "loss": 0.0454,
      "step": 8466
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2646369931607562,
      "learning_rate": 5.144139921428303e-07,
      "loss": 0.0227,
      "step": 8467
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1727333095384827,
      "learning_rate": 5.128121031989497e-07,
      "loss": 0.0177,
      "step": 8468
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.15503818756802384,
      "learning_rate": 5.112126799014628e-07,
      "loss": 0.0204,
      "step": 8469
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22728875927299597,
      "learning_rate": 5.096157224527343e-07,
      "loss": 0.0333,
      "step": 8470
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.38961350133778766,
      "learning_rate": 5.080212310548249e-07,
      "loss": 0.0477,
      "step": 8471
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.15363352667019123,
      "learning_rate": 5.0642920590948e-07,
      "loss": 0.0167,
      "step": 8472
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2225463495573057,
      "learning_rate": 5.048396472181316e-07,
      "loss": 0.0289,
      "step": 8473
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18319043032132049,
      "learning_rate": 5.032525551819012e-07,
      "loss": 0.0254,
      "step": 8474
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.17581090676948313,
      "learning_rate": 5.01667930001597e-07,
      "loss": 0.0171,
      "step": 8475
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.26942419607455437,
      "learning_rate": 5.000857718777186e-07,
      "loss": 0.0497,
      "step": 8476
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2394588248864191,
      "learning_rate": 4.985060810104503e-07,
      "loss": 0.0389,
      "step": 8477
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.24034996663318114,
      "learning_rate": 4.969288575996656e-07,
      "loss": 0.0446,
      "step": 8478
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.24040554120608754,
      "learning_rate": 4.953541018449226e-07,
      "loss": 0.0284,
      "step": 8479
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.27821203434472375,
      "learning_rate": 4.937818139454709e-07,
      "loss": 0.0318,
      "step": 8480
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40424619347803703,
      "learning_rate": 4.922119941002512e-07,
      "loss": 0.0592,
      "step": 8481
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.26562739819958725,
      "learning_rate": 4.906446425078782e-07,
      "loss": 0.0239,
      "step": 8482
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.186545605365763,
      "learning_rate": 4.890797593666708e-07,
      "loss": 0.0273,
      "step": 8483
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.24292673839417397,
      "learning_rate": 4.87517344874624e-07,
      "loss": 0.037,
      "step": 8484
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.17916370840365747,
      "learning_rate": 4.859573992294309e-07,
      "loss": 0.0211,
      "step": 8485
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20205597905386824,
      "learning_rate": 4.843999226284579e-07,
      "loss": 0.0248,
      "step": 8486
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.23498005650487097,
      "learning_rate": 4.82844915268772e-07,
      "loss": 0.0442,
      "step": 8487
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18656126108189497,
      "learning_rate": 4.812923773471201e-07,
      "loss": 0.0353,
      "step": 8488
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1672313940058219,
      "learning_rate": 4.797423090599451e-07,
      "loss": 0.0177,
      "step": 8489
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19812775113575354,
      "learning_rate": 4.781947106033635e-07,
      "loss": 0.0171,
      "step": 8490
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22391539631522134,
      "learning_rate": 4.7664958217319425e-07,
      "loss": 0.032,
      "step": 8491
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.155685418074848,
      "learning_rate": 4.7510692396493197e-07,
      "loss": 0.0249,
      "step": 8492
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22012540434670774,
      "learning_rate": 4.735667361737717e-07,
      "loss": 0.0342,
      "step": 8493
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21301856265144178,
      "learning_rate": 4.720290189945775e-07,
      "loss": 0.0229,
      "step": 8494
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21050044097990114,
      "learning_rate": 4.704937726219183e-07,
      "loss": 0.0269,
      "step": 8495
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18882153619106684,
      "learning_rate": 4.689609972500453e-07,
      "loss": 0.0259,
      "step": 8496
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1876996983582365,
      "learning_rate": 4.6743069307289e-07,
      "loss": 0.0215,
      "step": 8497
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.28610841302052187,
      "learning_rate": 4.659028602840776e-07,
      "loss": 0.0275,
      "step": 8498
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21757398974668332,
      "learning_rate": 4.643774990769201e-07,
      "loss": 0.0392,
      "step": 8499
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22618732959607027,
      "learning_rate": 4.628546096444186e-07,
      "loss": 0.0445,
      "step": 8500
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2897951146987235,
      "learning_rate": 4.613341921792569e-07,
      "loss": 0.033,
      "step": 8501
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.16572747748964892,
      "learning_rate": 4.5981624687380764e-07,
      "loss": 0.0185,
      "step": 8502
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21338993978016757,
      "learning_rate": 4.5830077392013064e-07,
      "loss": 0.0451,
      "step": 8503
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.24637319956372747,
      "learning_rate": 4.567877735099768e-07,
      "loss": 0.0319,
      "step": 8504
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18817690019328517,
      "learning_rate": 4.5527724583477537e-07,
      "loss": 0.0316,
      "step": 8505
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20283618898864303,
      "learning_rate": 4.5376919108565345e-07,
      "loss": 0.0427,
      "step": 8506
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1921731900881123,
      "learning_rate": 4.5226360945341607e-07,
      "loss": 0.0194,
      "step": 8507
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2673352245842023,
      "learning_rate": 4.507605011285643e-07,
      "loss": 0.0419,
      "step": 8508
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18281969276363283,
      "learning_rate": 4.4925986630127705e-07,
      "loss": 0.0346,
      "step": 8509
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2469516966988744,
      "learning_rate": 4.477617051614225e-07,
      "loss": 0.0225,
      "step": 8510
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2507970293969189,
      "learning_rate": 4.462660178985623e-07,
      "loss": 0.0474,
      "step": 8511
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.23158095203916151,
      "learning_rate": 4.4477280470194064e-07,
      "loss": 0.0295,
      "step": 8512
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20232670981041637,
      "learning_rate": 4.4328206576048414e-07,
      "loss": 0.0372,
      "step": 8513
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.16015584294337204,
      "learning_rate": 4.4179380126281533e-07,
      "loss": 0.0122,
      "step": 8514
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.23909338147034398,
      "learning_rate": 4.403080113972369e-07,
      "loss": 0.0466,
      "step": 8515
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21879980678292657,
      "learning_rate": 4.3882469635174287e-07,
      "loss": 0.037,
      "step": 8516
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18356716694027975,
      "learning_rate": 4.3734385631400976e-07,
      "loss": 0.0258,
      "step": 8517
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1950312114765048,
      "learning_rate": 4.358654914714033e-07,
      "loss": 0.0435,
      "step": 8518
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.22106834171267492,
      "learning_rate": 4.343896020109739e-07,
      "loss": 0.0489,
      "step": 8519
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2582378058291895,
      "learning_rate": 4.329161881194677e-07,
      "loss": 0.0229,
      "step": 8520
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24726344464683137,
      "learning_rate": 4.314452499833044e-07,
      "loss": 0.0359,
      "step": 8521
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17501671149297438,
      "learning_rate": 4.299767877885974e-07,
      "loss": 0.0254,
      "step": 8522
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19869288058583753,
      "learning_rate": 4.2851080172114703e-07,
      "loss": 0.0402,
      "step": 8523
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.15341339808696702,
      "learning_rate": 4.270472919664426e-07,
      "loss": 0.0237,
      "step": 8524
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20991070168576212,
      "learning_rate": 4.255862587096493e-07,
      "loss": 0.0424,
      "step": 8525
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19770893480378912,
      "learning_rate": 4.241277021356327e-07,
      "loss": 0.0196,
      "step": 8526
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2590692694576773,
      "learning_rate": 4.2267162242893846e-07,
      "loss": 0.0291,
      "step": 8527
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.22832070702224205,
      "learning_rate": 4.212180197737992e-07,
      "loss": 0.0292,
      "step": 8528
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.22958031443522428,
      "learning_rate": 4.1976689435413e-07,
      "loss": 0.0307,
      "step": 8529
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17204829712753653,
      "learning_rate": 4.183182463535418e-07,
      "loss": 0.0242,
      "step": 8530
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1848489081694108,
      "learning_rate": 4.1687207595532575e-07,
      "loss": 0.0214,
      "step": 8531
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19685404870404544,
      "learning_rate": 4.1542838334245994e-07,
      "loss": 0.0302,
      "step": 8532
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.22808550433970154,
      "learning_rate": 4.1398716869760936e-07,
      "loss": 0.0261,
      "step": 8533
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21967224297376345,
      "learning_rate": 4.1254843220312814e-07,
      "loss": 0.0285,
      "step": 8534
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18656843950730673,
      "learning_rate": 4.1111217404105286e-07,
      "loss": 0.0321,
      "step": 8535
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20425766811194307,
      "learning_rate": 4.096783943931093e-07,
      "loss": 0.035,
      "step": 8536
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2378283010313845,
      "learning_rate": 4.082470934407057e-07,
      "loss": 0.021,
      "step": 8537
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24951705628301746,
      "learning_rate": 4.0681827136494157e-07,
      "loss": 0.04,
      "step": 8538
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.27596928289252065,
      "learning_rate": 4.0539192834660346e-07,
      "loss": 0.0257,
      "step": 8539
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.16925648604013438,
      "learning_rate": 4.039680645661581e-07,
      "loss": 0.0287,
      "step": 8540
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19812483673209313,
      "learning_rate": 4.0254668020376366e-07,
      "loss": 0.0308,
      "step": 8541
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2352735540499613,
      "learning_rate": 4.011277754392606e-07,
      "loss": 0.0418,
      "step": 8542
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18347693846292293,
      "learning_rate": 3.9971135045218324e-07,
      "loss": 0.03,
      "step": 8543
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.25769029957936523,
      "learning_rate": 3.9829740542174143e-07,
      "loss": 0.0385,
      "step": 8544
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17769172701737182,
      "learning_rate": 3.968859405268388e-07,
      "loss": 0.022,
      "step": 8545
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.22493435113880472,
      "learning_rate": 3.954769559460614e-07,
      "loss": 0.0245,
      "step": 8546
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2125566031998893,
      "learning_rate": 3.9407045185768653e-07,
      "loss": 0.0253,
      "step": 8547
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2063302732801414,
      "learning_rate": 3.9266642843967415e-07,
      "loss": 0.0318,
      "step": 8548
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1527114002089824,
      "learning_rate": 3.912648858696666e-07,
      "loss": 0.0164,
      "step": 8549
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.16440860254344405,
      "learning_rate": 3.8986582432500196e-07,
      "loss": 0.0128,
      "step": 8550
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24528603988132788,
      "learning_rate": 3.8846924398269426e-07,
      "loss": 0.032,
      "step": 8551
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.15986126389858904,
      "learning_rate": 3.8707514501944657e-07,
      "loss": 0.0219,
      "step": 8552
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.27993088899621643,
      "learning_rate": 3.8568352761165327e-07,
      "loss": 0.0399,
      "step": 8553
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20659471769889556,
      "learning_rate": 3.842943919353914e-07,
      "loss": 0.0279,
      "step": 8554
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.16415103725905345,
      "learning_rate": 3.8290773816642035e-07,
      "loss": 0.0238,
      "step": 8555
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21029844436893214,
      "learning_rate": 3.815235664801908e-07,
      "loss": 0.0324,
      "step": 8556
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19070711712251118,
      "learning_rate": 3.801418770518339e-07,
      "loss": 0.0334,
      "step": 8557
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1516780123052012,
      "learning_rate": 3.787626700561742e-07,
      "loss": 0.0232,
      "step": 8558
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2401636376172256,
      "learning_rate": 3.7738594566771647e-07,
      "loss": 0.023,
      "step": 8559
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2239835894929799,
      "learning_rate": 3.7601170406065034e-07,
      "loss": 0.0263,
      "step": 8560
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24939471224266604,
      "learning_rate": 3.74639945408859e-07,
      "loss": 0.019,
      "step": 8561
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1822073412179125,
      "learning_rate": 3.732706698859012e-07,
      "loss": 0.0352,
      "step": 8562
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.15737887521734142,
      "learning_rate": 3.719038776650297e-07,
      "loss": 0.0187,
      "step": 8563
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2297675246361551,
      "learning_rate": 3.705395689191771e-07,
      "loss": 0.0395,
      "step": 8564
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19974198536794188,
      "learning_rate": 3.6917774382096984e-07,
      "loss": 0.0325,
      "step": 8565
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.22086383234848803,
      "learning_rate": 3.6781840254271227e-07,
      "loss": 0.0256,
      "step": 8566
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19444837145207994,
      "learning_rate": 3.6646154525639354e-07,
      "loss": 0.0344,
      "step": 8567
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19867574327237914,
      "learning_rate": 3.651071721336963e-07,
      "loss": 0.0269,
      "step": 8568
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2885035962536822,
      "learning_rate": 3.6375528334598343e-07,
      "loss": 0.041,
      "step": 8569
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.14988696612780852,
      "learning_rate": 3.62405879064307e-07,
      "loss": 0.0123,
      "step": 8570
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19290906315094386,
      "learning_rate": 3.610589594593994e-07,
      "loss": 0.0266,
      "step": 8571
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20667600743433642,
      "learning_rate": 3.59714524701682e-07,
      "loss": 0.0242,
      "step": 8572
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3196135314730812,
      "learning_rate": 3.5837257496126097e-07,
      "loss": 0.041,
      "step": 8573
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2103664651477876,
      "learning_rate": 3.5703311040793167e-07,
      "loss": 0.0246,
      "step": 8574
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.27069232858593484,
      "learning_rate": 3.5569613121116953e-07,
      "loss": 0.0299,
      "step": 8575
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18641511657808196,
      "learning_rate": 3.543616375401393e-07,
      "loss": 0.0277,
      "step": 8576
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.16748263744140146,
      "learning_rate": 3.5302962956368593e-07,
      "loss": 0.0205,
      "step": 8577
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21971879733457722,
      "learning_rate": 3.517001074503501e-07,
      "loss": 0.0397,
      "step": 8578
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21462903130288402,
      "learning_rate": 3.503730713683462e-07,
      "loss": 0.0295,
      "step": 8579
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24341620735315356,
      "learning_rate": 3.490485214855799e-07,
      "loss": 0.0431,
      "step": 8580
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17002344932182756,
      "learning_rate": 3.4772645796964824e-07,
      "loss": 0.0352,
      "step": 8581
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.23617304292794486,
      "learning_rate": 3.464068809878196e-07,
      "loss": 0.0203,
      "step": 8582
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.33190738387076246,
      "learning_rate": 3.4508979070705827e-07,
      "loss": 0.0404,
      "step": 8583
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24128678708441745,
      "learning_rate": 3.4377518729401317e-07,
      "loss": 0.0384,
      "step": 8584
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17993066753812728,
      "learning_rate": 3.4246307091501563e-07,
      "loss": 0.0229,
      "step": 8585
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20509025901137862,
      "learning_rate": 3.4115344173607957e-07,
      "loss": 0.0239,
      "step": 8586
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17688234694760688,
      "learning_rate": 3.398462999229124e-07,
      "loss": 0.0321,
      "step": 8587
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.23480918721310948,
      "learning_rate": 3.3854164564089964e-07,
      "loss": 0.0341,
      "step": 8588
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18343105615153737,
      "learning_rate": 3.372394790551159e-07,
      "loss": 0.0259,
      "step": 8589
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36601158310847753,
      "learning_rate": 3.359398003303183e-07,
      "loss": 0.0327,
      "step": 8590
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20515207138222125,
      "learning_rate": 3.3464260963095296e-07,
      "loss": 0.0294,
      "step": 8591
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1989280973963484,
      "learning_rate": 3.333479071211465e-07,
      "loss": 0.0334,
      "step": 8592
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21324569943833158,
      "learning_rate": 3.3205569296471675e-07,
      "loss": 0.0415,
      "step": 8593
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.31504375362037146,
      "learning_rate": 3.307659673251595e-07,
      "loss": 0.0346,
      "step": 8594
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21849175945800248,
      "learning_rate": 3.294787303656599e-07,
      "loss": 0.0453,
      "step": 8595
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2272270584822395,
      "learning_rate": 3.281939822490876e-07,
      "loss": 0.0326,
      "step": 8596
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21703630405800234,
      "learning_rate": 3.2691172313799925e-07,
      "loss": 0.0224,
      "step": 8597
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.16253392800690297,
      "learning_rate": 3.256319531946317e-07,
      "loss": 0.0143,
      "step": 8598
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.23217377925035143,
      "learning_rate": 3.243546725809132e-07,
      "loss": 0.02,
      "step": 8599
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2077514967617739,
      "learning_rate": 3.230798814584502e-07,
      "loss": 0.0455,
      "step": 8600
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20955170960353578,
      "learning_rate": 3.218075799885423e-07,
      "loss": 0.0313,
      "step": 8601
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2041855142045526,
      "learning_rate": 3.2053776833216533e-07,
      "loss": 0.0286,
      "step": 8602
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2700238494803682,
      "learning_rate": 3.192704466499841e-07,
      "loss": 0.0167,
      "step": 8603
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2102352568155349,
      "learning_rate": 3.1800561510234805e-07,
      "loss": 0.0269,
      "step": 8604
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20601155521161596,
      "learning_rate": 3.1674327384929593e-07,
      "loss": 0.0362,
      "step": 8605
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17974835899357347,
      "learning_rate": 3.1548342305054435e-07,
      "loss": 0.0256,
      "step": 8606
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18722919935808596,
      "learning_rate": 3.1422606286549915e-07,
      "loss": 0.0336,
      "step": 8607
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20983967495197475,
      "learning_rate": 3.1297119345324645e-07,
      "loss": 0.0246,
      "step": 8608
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3352326383881315,
      "learning_rate": 3.1171881497256585e-07,
      "loss": 0.0526,
      "step": 8609
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18821854793810713,
      "learning_rate": 3.104689275819128e-07,
      "loss": 0.0172,
      "step": 8610
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21629474673071145,
      "learning_rate": 3.09221531439432e-07,
      "loss": 0.0196,
      "step": 8611
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17038020441862262,
      "learning_rate": 3.079766267029527e-07,
      "loss": 0.024,
      "step": 8612
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18133856155359315,
      "learning_rate": 3.067342135299867e-07,
      "loss": 0.0275,
      "step": 8613
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1653726918693068,
      "learning_rate": 3.0549429207773483e-07,
      "loss": 0.0167,
      "step": 8614
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21832684904187177,
      "learning_rate": 3.0425686250307616e-07,
      "loss": 0.0261,
      "step": 8615
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19437140711379983,
      "learning_rate": 3.030219249625854e-07,
      "loss": 0.0198,
      "step": 8616
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.25258798973484187,
      "learning_rate": 3.017894796125065e-07,
      "loss": 0.0347,
      "step": 8617
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22867072783802445,
      "learning_rate": 3.005595266087835e-07,
      "loss": 0.038,
      "step": 8618
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18944378466046882,
      "learning_rate": 2.9933206610703203e-07,
      "loss": 0.0235,
      "step": 8619
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20009014809980036,
      "learning_rate": 2.9810709826256557e-07,
      "loss": 0.0219,
      "step": 8620
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2026882795435912,
      "learning_rate": 2.9688462323036685e-07,
      "loss": 0.0316,
      "step": 8621
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.277354052177682,
      "learning_rate": 2.956646411651165e-07,
      "loss": 0.0468,
      "step": 8622
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21367655928531423,
      "learning_rate": 2.944471522211756e-07,
      "loss": 0.0354,
      "step": 8623
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21471886906782237,
      "learning_rate": 2.932321565525853e-07,
      "loss": 0.0318,
      "step": 8624
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19926723189488613,
      "learning_rate": 2.920196543130782e-07,
      "loss": 0.0407,
      "step": 8625
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19702486888526335,
      "learning_rate": 2.9080964565606273e-07,
      "loss": 0.0305,
      "step": 8626
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.24807929780772595,
      "learning_rate": 2.8960213073464305e-07,
      "loss": 0.0334,
      "step": 8627
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20870450254983694,
      "learning_rate": 2.883971097015992e-07,
      "loss": 0.0342,
      "step": 8628
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.24114173792306193,
      "learning_rate": 2.8719458270939805e-07,
      "loss": 0.0391,
      "step": 8629
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18148611531796444,
      "learning_rate": 2.859945499101913e-07,
      "loss": 0.0223,
      "step": 8630
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20827598477839246,
      "learning_rate": 2.8479701145581296e-07,
      "loss": 0.0343,
      "step": 8631
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2023911516348676,
      "learning_rate": 2.8360196749778857e-07,
      "loss": 0.0317,
      "step": 8632
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19551306916600292,
      "learning_rate": 2.824094181873194e-07,
      "loss": 0.0379,
      "step": 8633
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2606711558767825,
      "learning_rate": 2.8121936367529357e-07,
      "loss": 0.0316,
      "step": 8634
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2046358880645793,
      "learning_rate": 2.800318041122885e-07,
      "loss": 0.0288,
      "step": 8635
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17812063354025826,
      "learning_rate": 2.788467396485595e-07,
      "loss": 0.0218,
      "step": 8636
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.23556589521095353,
      "learning_rate": 2.776641704340466e-07,
      "loss": 0.0422,
      "step": 8637
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.32835729641452743,
      "learning_rate": 2.7648409661837903e-07,
      "loss": 0.0512,
      "step": 8638
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2304422799766463,
      "learning_rate": 2.7530651835086854e-07,
      "loss": 0.0345,
      "step": 8639
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.24438107665914258,
      "learning_rate": 2.7413143578050915e-07,
      "loss": 0.0198,
      "step": 8640
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.25114189075764987,
      "learning_rate": 2.7295884905597536e-07,
      "loss": 0.0285,
      "step": 8641
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21279865868267142,
      "learning_rate": 2.7178875832563734e-07,
      "loss": 0.027,
      "step": 8642
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.213098870364833,
      "learning_rate": 2.706211637375367e-07,
      "loss": 0.0276,
      "step": 8643
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20320541367347428,
      "learning_rate": 2.6945606543941073e-07,
      "loss": 0.0394,
      "step": 8644
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1389426019901279,
      "learning_rate": 2.682934635786727e-07,
      "loss": 0.0216,
      "step": 8645
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.16812550083780672,
      "learning_rate": 2.671333583024205e-07,
      "loss": 0.0221,
      "step": 8646
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21070911264592893,
      "learning_rate": 2.659757497574411e-07,
      "loss": 0.038,
      "step": 8647
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1850226072887055,
      "learning_rate": 2.6482063809020186e-07,
      "loss": 0.03,
      "step": 8648
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17965256309873284,
      "learning_rate": 2.636680234468525e-07,
      "loss": 0.0183,
      "step": 8649
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18191220889335893,
      "learning_rate": 2.625179059732341e-07,
      "loss": 0.0264,
      "step": 8650
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18788445317844266,
      "learning_rate": 2.613702858148659e-07,
      "loss": 0.0307,
      "step": 8651
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22395485590667272,
      "learning_rate": 2.6022516311695166e-07,
      "loss": 0.0286,
      "step": 8652
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21229528601078365,
      "learning_rate": 2.5908253802437555e-07,
      "loss": 0.0295,
      "step": 8653
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3119257777240814,
      "learning_rate": 2.579424106817174e-07,
      "loss": 0.0517,
      "step": 8654
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22517244002819475,
      "learning_rate": 2.568047812332286e-07,
      "loss": 0.0373,
      "step": 8655
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.24037222280103096,
      "learning_rate": 2.556696498228495e-07,
      "loss": 0.029,
      "step": 8656
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2562675278181715,
      "learning_rate": 2.5453701659420735e-07,
      "loss": 0.0327,
      "step": 8657
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2208832368755895,
      "learning_rate": 2.5340688169060767e-07,
      "loss": 0.0277,
      "step": 8658
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2539777801699454,
      "learning_rate": 2.522792452550449e-07,
      "loss": 0.048,
      "step": 8659
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21972674051913363,
      "learning_rate": 2.511541074301915e-07,
      "loss": 0.0177,
      "step": 8660
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4600370145760082,
      "learning_rate": 2.500314683584093e-07,
      "loss": 0.0516,
      "step": 8661
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1968085025322031,
      "learning_rate": 2.489113281817424e-07,
      "loss": 0.0348,
      "step": 8662
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19551116376475933,
      "learning_rate": 2.477936870419195e-07,
      "loss": 0.0358,
      "step": 8663
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17816695296492321,
      "learning_rate": 2.4667854508034774e-07,
      "loss": 0.0314,
      "step": 8664
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22978288622641532,
      "learning_rate": 2.4556590243812515e-07,
      "loss": 0.0308,
      "step": 8665
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21939392342419994,
      "learning_rate": 2.444557592560304e-07,
      "loss": 0.045,
      "step": 8666
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1737265611609018,
      "learning_rate": 2.433481156745243e-07,
      "loss": 0.0279,
      "step": 8667
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2204489890501506,
      "learning_rate": 2.4224297183375487e-07,
      "loss": 0.0262,
      "step": 8668
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2513867451396066,
      "learning_rate": 2.4114032787355246e-07,
      "loss": 0.03,
      "step": 8669
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.23942818317294873,
      "learning_rate": 2.400401839334299e-07,
      "loss": 0.0339,
      "step": 8670
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1708324741035656,
      "learning_rate": 2.3894254015258467e-07,
      "loss": 0.0382,
      "step": 8671
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17016075049509563,
      "learning_rate": 2.378473966698991e-07,
      "loss": 0.0238,
      "step": 8672
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21863160164889495,
      "learning_rate": 2.3675475362393562e-07,
      "loss": 0.0197,
      "step": 8673
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17165372975588103,
      "learning_rate": 2.356646111529415e-07,
      "loss": 0.0202,
      "step": 8674
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18773788025020802,
      "learning_rate": 2.345769693948552e-07,
      "loss": 0.0235,
      "step": 8675
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22634589594585167,
      "learning_rate": 2.3349182848728447e-07,
      "loss": 0.0267,
      "step": 8676
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20798678582884628,
      "learning_rate": 2.324091885675328e-07,
      "loss": 0.0227,
      "step": 8677
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.32563660152510937,
      "learning_rate": 2.3132904977258175e-07,
      "loss": 0.0439,
      "step": 8678
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2342346565947724,
      "learning_rate": 2.3025141223909975e-07,
      "loss": 0.0442,
      "step": 8679
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2857989554033083,
      "learning_rate": 2.291762761034333e-07,
      "loss": 0.0412,
      "step": 8680
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.26594532557268624,
      "learning_rate": 2.2810364150161578e-07,
      "loss": 0.0381,
      "step": 8681
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22798662005815787,
      "learning_rate": 2.2703350856936534e-07,
      "loss": 0.0419,
      "step": 8682
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1707937355834813,
      "learning_rate": 2.2596587744208254e-07,
      "loss": 0.0179,
      "step": 8683
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20069499111064418,
      "learning_rate": 2.249007482548482e-07,
      "loss": 0.0395,
      "step": 8684
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1956326150173752,
      "learning_rate": 2.2383812114243453e-07,
      "loss": 0.0355,
      "step": 8685
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22086562866705764,
      "learning_rate": 2.2277799623928953e-07,
      "loss": 0.0353,
      "step": 8686
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2313041336903829,
      "learning_rate": 2.2172037367954368e-07,
      "loss": 0.0371,
      "step": 8687
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22586839505488646,
      "learning_rate": 2.2066525359701885e-07,
      "loss": 0.0445,
      "step": 8688
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.16569625765622503,
      "learning_rate": 2.196126361252149e-07,
      "loss": 0.0277,
      "step": 8689
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22487159018505065,
      "learning_rate": 2.1856252139731637e-07,
      "loss": 0.0389,
      "step": 8690
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2215080605142478,
      "learning_rate": 2.1751490954618814e-07,
      "loss": 0.0398,
      "step": 8691
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.23735139332230346,
      "learning_rate": 2.1646980070437973e-07,
      "loss": 0.0516,
      "step": 8692
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22935181240677988,
      "learning_rate": 2.154271950041298e-07,
      "loss": 0.0314,
      "step": 8693
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.283157450182638,
      "learning_rate": 2.14387092577355e-07,
      "loss": 0.0368,
      "step": 8694
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19664583947198228,
      "learning_rate": 2.1334949355565237e-07,
      "loss": 0.0172,
      "step": 8695
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.25732791036603486,
      "learning_rate": 2.1231439807031019e-07,
      "loss": 0.0384,
      "step": 8696
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1962740024084322,
      "learning_rate": 2.1128180625229033e-07,
      "loss": 0.0176,
      "step": 8697
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.25764050670988675,
      "learning_rate": 2.102517182322483e-07,
      "loss": 0.0368,
      "step": 8698
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.16427001998480692,
      "learning_rate": 2.0922413414051324e-07,
      "loss": 0.0132,
      "step": 8699
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2661667376381603,
      "learning_rate": 2.0819905410710327e-07,
      "loss": 0.0325,
      "step": 8700
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.16565415917433382,
      "learning_rate": 2.0717647826171917e-07,
      "loss": 0.025,
      "step": 8701
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18639798963843018,
      "learning_rate": 2.0615640673374181e-07,
      "loss": 0.0202,
      "step": 8702
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.17820527501462835,
      "learning_rate": 2.0513883965224136e-07,
      "loss": 0.0185,
      "step": 8703
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1558295015756042,
      "learning_rate": 2.0412377714596365e-07,
      "loss": 0.0206,
      "step": 8704
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22946891666683195,
      "learning_rate": 2.0311121934334155e-07,
      "loss": 0.026,
      "step": 8705
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22361841113766576,
      "learning_rate": 2.0210116637249032e-07,
      "loss": 0.0351,
      "step": 8706
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1860691550118275,
      "learning_rate": 2.0109361836121e-07,
      "loss": 0.0203,
      "step": 8707
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24734992387672888,
      "learning_rate": 2.0008857543698078e-07,
      "loss": 0.0283,
      "step": 8708
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.16472088316185876,
      "learning_rate": 1.9908603772696988e-07,
      "loss": 0.0205,
      "step": 8709
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19457749349488115,
      "learning_rate": 1.9808600535802024e-07,
      "loss": 0.0249,
      "step": 8710
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24792848600268036,
      "learning_rate": 1.97088478456664e-07,
      "loss": 0.0307,
      "step": 8711
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.16460276385384964,
      "learning_rate": 1.9609345714911575e-07,
      "loss": 0.0304,
      "step": 8712
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2817525215450733,
      "learning_rate": 1.9510094156127258e-07,
      "loss": 0.0319,
      "step": 8713
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21592069681929513,
      "learning_rate": 1.941109318187162e-07,
      "loss": 0.0336,
      "step": 8714
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2503272537283905,
      "learning_rate": 1.9312342804670425e-07,
      "loss": 0.0271,
      "step": 8715
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18382826497485616,
      "learning_rate": 1.9213843037018344e-07,
      "loss": 0.0265,
      "step": 8716
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.28449225523535515,
      "learning_rate": 1.9115593891378294e-07,
      "loss": 0.0333,
      "step": 8717
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20514299184256218,
      "learning_rate": 1.9017595380181442e-07,
      "loss": 0.0435,
      "step": 8718
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.25111864573052306,
      "learning_rate": 1.8919847515827205e-07,
      "loss": 0.0323,
      "step": 8719
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2204088595533447,
      "learning_rate": 1.8822350310683246e-07,
      "loss": 0.0259,
      "step": 8720
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19115970264152365,
      "learning_rate": 1.8725103777085696e-07,
      "loss": 0.0289,
      "step": 8721
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23806926021106145,
      "learning_rate": 1.862810792733849e-07,
      "loss": 0.0323,
      "step": 8722
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22445882644608597,
      "learning_rate": 1.8531362773714478e-07,
      "loss": 0.0295,
      "step": 8723
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20504282692992654,
      "learning_rate": 1.843486832845409e-07,
      "loss": 0.0294,
      "step": 8724
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2785837231990959,
      "learning_rate": 1.833862460376712e-07,
      "loss": 0.0377,
      "step": 8725
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18639706030206207,
      "learning_rate": 1.8242631611830263e-07,
      "loss": 0.0303,
      "step": 8726
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22244493596080567,
      "learning_rate": 1.8146889364789365e-07,
      "loss": 0.0459,
      "step": 8727
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.16856025076621864,
      "learning_rate": 1.8051397874758736e-07,
      "loss": 0.0232,
      "step": 8728
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.35854492519980763,
      "learning_rate": 1.7956157153820263e-07,
      "loss": 0.0242,
      "step": 8729
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18458326990572813,
      "learning_rate": 1.786116721402431e-07,
      "loss": 0.0257,
      "step": 8730
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21693626424832563,
      "learning_rate": 1.7766428067389929e-07,
      "loss": 0.0342,
      "step": 8731
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.17757064488067797,
      "learning_rate": 1.7671939725903752e-07,
      "loss": 0.0256,
      "step": 8732
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23044289427729753,
      "learning_rate": 1.7577702201521552e-07,
      "loss": 0.0353,
      "step": 8733
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2281756847721565,
      "learning_rate": 1.7483715506166455e-07,
      "loss": 0.0339,
      "step": 8734
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18809631017998235,
      "learning_rate": 1.7389979651730503e-07,
      "loss": 0.0369,
      "step": 8735
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20987863923791927,
      "learning_rate": 1.729649465007377e-07,
      "loss": 0.0243,
      "step": 8736
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18856081290180998,
      "learning_rate": 1.7203260513024345e-07,
      "loss": 0.0245,
      "step": 8737
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23034486835618645,
      "learning_rate": 1.7110277252379238e-07,
      "loss": 0.0431,
      "step": 8738
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23875800887001444,
      "learning_rate": 1.7017544879902813e-07,
      "loss": 0.03,
      "step": 8739
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2574691509082116,
      "learning_rate": 1.692506340732858e-07,
      "loss": 0.0292,
      "step": 8740
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23162589291690217,
      "learning_rate": 1.6832832846357837e-07,
      "loss": 0.0198,
      "step": 8741
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1538498111780723,
      "learning_rate": 1.6740853208659923e-07,
      "loss": 0.0204,
      "step": 8742
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2979419508566156,
      "learning_rate": 1.664912450587286e-07,
      "loss": 0.0476,
      "step": 8743
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23228707932673953,
      "learning_rate": 1.655764674960292e-07,
      "loss": 0.0303,
      "step": 8744
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24256632003685585,
      "learning_rate": 1.6466419951424395e-07,
      "loss": 0.035,
      "step": 8745
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1886892156747234,
      "learning_rate": 1.6375444122879613e-07,
      "loss": 0.0255,
      "step": 8746
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24529041385238903,
      "learning_rate": 1.628471927547981e-07,
      "loss": 0.0278,
      "step": 8747
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18393948649255015,
      "learning_rate": 1.6194245420704025e-07,
      "loss": 0.0285,
      "step": 8748
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2326993488078889,
      "learning_rate": 1.6104022569999323e-07,
      "loss": 0.0245,
      "step": 8749
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19768047146660006,
      "learning_rate": 1.6014050734781461e-07,
      "loss": 0.0165,
      "step": 8750
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20214817950098649,
      "learning_rate": 1.5924329926434225e-07,
      "loss": 0.0248,
      "step": 8751
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20138647614235908,
      "learning_rate": 1.583486015630986e-07,
      "loss": 0.0396,
      "step": 8752
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.17790708429706126,
      "learning_rate": 1.5745641435728432e-07,
      "loss": 0.0247,
      "step": 8753
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2147278212839516,
      "learning_rate": 1.565667377597868e-07,
      "loss": 0.0289,
      "step": 8754
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24563264032910329,
      "learning_rate": 1.5567957188317162e-07,
      "loss": 0.0499,
      "step": 8755
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1959162345553162,
      "learning_rate": 1.5479491683969117e-07,
      "loss": 0.019,
      "step": 8756
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.15475789741661294,
      "learning_rate": 1.539127727412759e-07,
      "loss": 0.0124,
      "step": 8757
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2337968211531824,
      "learning_rate": 1.5303313969954103e-07,
      "loss": 0.0349,
      "step": 8758
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.14595679325108346,
      "learning_rate": 1.5215601782578416e-07,
      "loss": 0.0241,
      "step": 8759
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24746499182459178,
      "learning_rate": 1.5128140723098317e-07,
      "loss": 0.025,
      "step": 8760
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19813106034212843,
      "learning_rate": 1.5040930802580066e-07,
      "loss": 0.023,
      "step": 8761
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2703189259244915,
      "learning_rate": 1.4953972032057952e-07,
      "loss": 0.0287,
      "step": 8762
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3282242011930816,
      "learning_rate": 1.48672644225345e-07,
      "loss": 0.0241,
      "step": 8763
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20593735667384344,
      "learning_rate": 1.4780807984980716e-07,
      "loss": 0.0389,
      "step": 8764
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21362045722481685,
      "learning_rate": 1.4694602730335626e-07,
      "loss": 0.0439,
      "step": 8765
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19380484427727424,
      "learning_rate": 1.4608648669506287e-07,
      "loss": 0.0278,
      "step": 8766
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20894393351341048,
      "learning_rate": 1.4522945813368216e-07,
      "loss": 0.0371,
      "step": 8767
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.17357704159443765,
      "learning_rate": 1.443749417276541e-07,
      "loss": 0.032,
      "step": 8768
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21573464605082726,
      "learning_rate": 1.4352293758509218e-07,
      "loss": 0.0351,
      "step": 8769
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24357158721561883,
      "learning_rate": 1.4267344581380127e-07,
      "loss": 0.0377,
      "step": 8770
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2319143427840215,
      "learning_rate": 1.4182646652126652e-07,
      "loss": 0.0398,
      "step": 8771
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18050443453373913,
      "learning_rate": 1.4098199981464887e-07,
      "loss": 0.0317,
      "step": 8772
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3103363475106859,
      "learning_rate": 1.401400458007962e-07,
      "loss": 0.0466,
      "step": 8773
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18228047995075222,
      "learning_rate": 1.3930060458624106e-07,
      "loss": 0.0253,
      "step": 8774
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1929269347657742,
      "learning_rate": 1.3846367627719402e-07,
      "loss": 0.0208,
      "step": 8775
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20458900615365908,
      "learning_rate": 1.37629260979546e-07,
      "loss": 0.0284,
      "step": 8776
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23223062869427047,
      "learning_rate": 1.3679735879887468e-07,
      "loss": 0.0335,
      "step": 8777
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24302910119610532,
      "learning_rate": 1.3596796984044037e-07,
      "loss": 0.0248,
      "step": 8778
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21359427122039937,
      "learning_rate": 1.351410942091791e-07,
      "loss": 0.0328,
      "step": 8779
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18608433030996446,
      "learning_rate": 1.3431673200971386e-07,
      "loss": 0.0132,
      "step": 8780
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39687609935217755,
      "learning_rate": 1.3349488334634565e-07,
      "loss": 0.0582,
      "step": 8781
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23109737430797175,
      "learning_rate": 1.3267554832306463e-07,
      "loss": 0.0367,
      "step": 8782
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.15203252816324558,
      "learning_rate": 1.3185872704353898e-07,
      "loss": 0.0232,
      "step": 8783
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23426329811928717,
      "learning_rate": 1.310444196111127e-07,
      "loss": 0.0533,
      "step": 8784
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20209973435030545,
      "learning_rate": 1.3023262612882116e-07,
      "loss": 0.0221,
      "step": 8785
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.25584715894117294,
      "learning_rate": 1.2942334669937773e-07,
      "loss": 0.0403,
      "step": 8786
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2064357867249212,
      "learning_rate": 1.2861658142517608e-07,
      "loss": 0.0251,
      "step": 8787
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20337371088321607,
      "learning_rate": 1.2781233040829234e-07,
      "loss": 0.0181,
      "step": 8788
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.27046586134859407,
      "learning_rate": 1.2701059375049174e-07,
      "loss": 0.034,
      "step": 8789
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21523235894728446,
      "learning_rate": 1.2621137155320872e-07,
      "loss": 0.0398,
      "step": 8790
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21681457309825938,
      "learning_rate": 1.2541466391756907e-07,
      "loss": 0.0187,
      "step": 8791
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18662048730996217,
      "learning_rate": 1.2462047094437657e-07,
      "loss": 0.0258,
      "step": 8792
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2708380986063448,
      "learning_rate": 1.2382879273411753e-07,
      "loss": 0.0291,
      "step": 8793
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19884157940922187,
      "learning_rate": 1.2303962938696068e-07,
      "loss": 0.0309,
      "step": 8794
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2374785928113264,
      "learning_rate": 1.222529810027573e-07,
      "loss": 0.0344,
      "step": 8795
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.210061328751313,
      "learning_rate": 1.2146884768103883e-07,
      "loss": 0.0281,
      "step": 8796
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.25380423036811794,
      "learning_rate": 1.2068722952101707e-07,
      "loss": 0.0342,
      "step": 8797
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17138275728782582,
      "learning_rate": 1.1990812662158846e-07,
      "loss": 0.0287,
      "step": 8798
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.22685520403645804,
      "learning_rate": 1.191315390813319e-07,
      "loss": 0.0287,
      "step": 8799
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38724759095084776,
      "learning_rate": 1.1835746699850215e-07,
      "loss": 0.0375,
      "step": 8800
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20292787890688466,
      "learning_rate": 1.1758591047104306e-07,
      "loss": 0.0326,
      "step": 8801
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21147031953662404,
      "learning_rate": 1.1681686959657879e-07,
      "loss": 0.0213,
      "step": 8802
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.24423717878322407,
      "learning_rate": 1.1605034447240925e-07,
      "loss": 0.0323,
      "step": 8803
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.299142186302002,
      "learning_rate": 1.1528633519552357e-07,
      "loss": 0.038,
      "step": 8804
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17280086485063506,
      "learning_rate": 1.1452484186258439e-07,
      "loss": 0.0216,
      "step": 8805
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19992484453864742,
      "learning_rate": 1.1376586456994798e-07,
      "loss": 0.0322,
      "step": 8806
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1948669788215871,
      "learning_rate": 1.1300940341363752e-07,
      "loss": 0.0326,
      "step": 8807
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.16161638005231468,
      "learning_rate": 1.1225545848937203e-07,
      "loss": 0.0311,
      "step": 8808
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2297560453314853,
      "learning_rate": 1.1150402989254183e-07,
      "loss": 0.0303,
      "step": 8809
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19112317893431216,
      "learning_rate": 1.1075511771822423e-07,
      "loss": 0.0374,
      "step": 8810
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23386695157101786,
      "learning_rate": 1.1000872206117452e-07,
      "loss": 0.0373,
      "step": 8811
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19713922737970394,
      "learning_rate": 1.0926484301583273e-07,
      "loss": 0.0394,
      "step": 8812
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18110095629371134,
      "learning_rate": 1.0852348067631913e-07,
      "loss": 0.0274,
      "step": 8813
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2928962617842024,
      "learning_rate": 1.0778463513643645e-07,
      "loss": 0.0668,
      "step": 8814
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2546459573670888,
      "learning_rate": 1.0704830648966769e-07,
      "loss": 0.0356,
      "step": 8815
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21231603513149047,
      "learning_rate": 1.0631449482917833e-07,
      "loss": 0.0299,
      "step": 8816
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1979718643732338,
      "learning_rate": 1.055832002478141e-07,
      "loss": 0.0209,
      "step": 8817
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21945028101543912,
      "learning_rate": 1.0485442283810321e-07,
      "loss": 0.0401,
      "step": 8818
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1639425656768194,
      "learning_rate": 1.0412816269225635e-07,
      "loss": 0.0138,
      "step": 8819
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2373730517840274,
      "learning_rate": 1.0340441990216443e-07,
      "loss": 0.0338,
      "step": 8820
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19370221079304215,
      "learning_rate": 1.026831945594009e-07,
      "loss": 0.0307,
      "step": 8821
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23440798686271105,
      "learning_rate": 1.019644867552172e-07,
      "loss": 0.0259,
      "step": 8822
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20228400729248017,
      "learning_rate": 1.0124829658055168e-07,
      "loss": 0.0276,
      "step": 8823
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18871557073288675,
      "learning_rate": 1.0053462412601855e-07,
      "loss": 0.0263,
      "step": 8824
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21704169360163283,
      "learning_rate": 9.982346948192112e-08,
      "loss": 0.0321,
      "step": 8825
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17353412325180773,
      "learning_rate": 9.911483273823408e-08,
      "loss": 0.0208,
      "step": 8826
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2056149495633992,
      "learning_rate": 9.840871398462126e-08,
      "loss": 0.0387,
      "step": 8827
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20205207038862702,
      "learning_rate": 9.770511331042454e-08,
      "loss": 0.042,
      "step": 8828
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.24749035618999599,
      "learning_rate": 9.700403080467046e-08,
      "loss": 0.029,
      "step": 8829
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.206954698985052,
      "learning_rate": 9.630546655606365e-08,
      "loss": 0.0182,
      "step": 8830
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2722647996668299,
      "learning_rate": 9.560942065298673e-08,
      "loss": 0.0422,
      "step": 8831
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2047944331944346,
      "learning_rate": 9.491589318351368e-08,
      "loss": 0.0364,
      "step": 8832
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19281538146600172,
      "learning_rate": 9.422488423539211e-08,
      "loss": 0.0195,
      "step": 8833
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.16511098434459562,
      "learning_rate": 9.353639389605207e-08,
      "loss": 0.0229,
      "step": 8834
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20503108983846863,
      "learning_rate": 9.285042225260832e-08,
      "loss": 0.0335,
      "step": 8835
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18161791997947183,
      "learning_rate": 9.216696939184922e-08,
      "loss": 0.0232,
      "step": 8836
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.39087788978901583,
      "learning_rate": 9.14860354002567e-08,
      "loss": 0.041,
      "step": 8837
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.15480180404137683,
      "learning_rate": 9.080762036398184e-08,
      "loss": 0.0298,
      "step": 8838
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.22438641137934368,
      "learning_rate": 9.013172436886263e-08,
      "loss": 0.0341,
      "step": 8839
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23703619599132775,
      "learning_rate": 8.94583475004196e-08,
      "loss": 0.043,
      "step": 8840
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18334841061135454,
      "learning_rate": 8.878748984385344e-08,
      "loss": 0.0392,
      "step": 8841
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23021026437188277,
      "learning_rate": 8.811915148404294e-08,
      "loss": 0.026,
      "step": 8842
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1827860067195035,
      "learning_rate": 8.745333250555377e-08,
      "loss": 0.0271,
      "step": 8843
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2202341532224017,
      "learning_rate": 8.679003299262523e-08,
      "loss": 0.0365,
      "step": 8844
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23085318039072167,
      "learning_rate": 8.612925302918796e-08,
      "loss": 0.0376,
      "step": 8845
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.29691353347062704,
      "learning_rate": 8.547099269884396e-08,
      "loss": 0.0333,
      "step": 8846
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.208201573350002,
      "learning_rate": 8.481525208487995e-08,
      "loss": 0.0268,
      "step": 8847
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2416686221706812,
      "learning_rate": 8.416203127026734e-08,
      "loss": 0.0364,
      "step": 8848
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.15994587316132045,
      "learning_rate": 8.351133033765557e-08,
      "loss": 0.0166,
      "step": 8849
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1608993283622438,
      "learning_rate": 8.286314936937434e-08,
      "loss": 0.0375,
      "step": 8850
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19504197459049552,
      "learning_rate": 8.221748844743804e-08,
      "loss": 0.032,
      "step": 8851
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2631298286284889,
      "learning_rate": 8.157434765353466e-08,
      "loss": 0.0229,
      "step": 8852
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.22562095238233013,
      "learning_rate": 8.093372706904579e-08,
      "loss": 0.0234,
      "step": 8853
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.240873154336681,
      "learning_rate": 8.029562677502212e-08,
      "loss": 0.0336,
      "step": 8854
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1875751662586995,
      "learning_rate": 7.966004685219908e-08,
      "loss": 0.0302,
      "step": 8855
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.255440492846282,
      "learning_rate": 7.902698738099901e-08,
      "loss": 0.0292,
      "step": 8856
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1964251392934732,
      "learning_rate": 7.839644844151784e-08,
      "loss": 0.0333,
      "step": 8857
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.24668400987959496,
      "learning_rate": 7.776843011353619e-08,
      "loss": 0.0301,
      "step": 8858
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.22160146400238329,
      "learning_rate": 7.714293247651494e-08,
      "loss": 0.0326,
      "step": 8859
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21657704041673662,
      "learning_rate": 7.651995560959525e-08,
      "loss": 0.0274,
      "step": 8860
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20973267678300314,
      "learning_rate": 7.589949959160292e-08,
      "loss": 0.0336,
      "step": 8861
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19163122384999404,
      "learning_rate": 7.528156450103963e-08,
      "loss": 0.0263,
      "step": 8862
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.24958776222230894,
      "learning_rate": 7.46661504160895e-08,
      "loss": 0.0268,
      "step": 8863
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.16567642677578373,
      "learning_rate": 7.405325741462354e-08,
      "loss": 0.0212,
      "step": 8864
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17346485089816543,
      "learning_rate": 7.344288557418422e-08,
      "loss": 0.0154,
      "step": 8865
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.16755537799026465,
      "learning_rate": 7.283503497200083e-08,
      "loss": 0.0179,
      "step": 8866
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.22721290132856814,
      "learning_rate": 7.222970568498299e-08,
      "loss": 0.0323,
      "step": 8867
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2073917399353256,
      "learning_rate": 7.162689778972276e-08,
      "loss": 0.0316,
      "step": 8868
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.22171986421677445,
      "learning_rate": 7.102661136248801e-08,
      "loss": 0.0195,
      "step": 8869
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2027553349062341,
      "learning_rate": 7.042884647923353e-08,
      "loss": 0.0422,
      "step": 8870
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2404274281101992,
      "learning_rate": 6.983360321559218e-08,
      "loss": 0.0398,
      "step": 8871
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18298409062942378,
      "learning_rate": 6.924088164687703e-08,
      "loss": 0.024,
      "step": 8872
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2195366950394955,
      "learning_rate": 6.865068184808366e-08,
      "loss": 0.0329,
      "step": 8873
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23303050005089862,
      "learning_rate": 6.806300389388565e-08,
      "loss": 0.0502,
      "step": 8874
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23381320060720864,
      "learning_rate": 6.747784785864354e-08,
      "loss": 0.0161,
      "step": 8875
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21115008873706176,
      "learning_rate": 6.689521381639363e-08,
      "loss": 0.0324,
      "step": 8876
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2722526355638607,
      "learning_rate": 6.631510184085477e-08,
      "loss": 0.0275,
      "step": 8877
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2771019754854928,
      "learning_rate": 6.573751200542599e-08,
      "loss": 0.0277,
      "step": 8878
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.17865040280213682,
      "learning_rate": 6.516244438318664e-08,
      "loss": 0.0205,
      "step": 8879
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1856223806042282,
      "learning_rate": 6.458989904690072e-08,
      "loss": 0.0196,
      "step": 8880
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2362359000973686,
      "learning_rate": 6.401987606901028e-08,
      "loss": 0.023,
      "step": 8881
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.23464998961370817,
      "learning_rate": 6.345237552163541e-08,
      "loss": 0.0308,
      "step": 8882
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.16618538199456276,
      "learning_rate": 6.288739747658312e-08,
      "loss": 0.0147,
      "step": 8883
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1973434521332318,
      "learning_rate": 6.232494200533623e-08,
      "loss": 0.0191,
      "step": 8884
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.26002566866528776,
      "learning_rate": 6.176500917906003e-08,
      "loss": 0.0413,
      "step": 8885
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4893865663372262,
      "learning_rate": 6.120759906860008e-08,
      "loss": 0.0714,
      "step": 8886
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2434821365700065,
      "learning_rate": 6.065271174448884e-08,
      "loss": 0.0404,
      "step": 8887
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18391031001494496,
      "learning_rate": 6.010034727692792e-08,
      "loss": 0.0274,
      "step": 8888
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.29160401687560084,
      "learning_rate": 5.955050573581034e-08,
      "loss": 0.0313,
      "step": 8889
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19577673100814572,
      "learning_rate": 5.900318719070264e-08,
      "loss": 0.0304,
      "step": 8890
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2033135565992095,
      "learning_rate": 5.84583917108561e-08,
      "loss": 0.0245,
      "step": 8891
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33709778130354817,
      "learning_rate": 5.791611936520447e-08,
      "loss": 0.0557,
      "step": 8892
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1923278207171422,
      "learning_rate": 5.737637022235509e-08,
      "loss": 0.024,
      "step": 8893
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21318792005859344,
      "learning_rate": 5.683914435060445e-08,
      "loss": 0.0357,
      "step": 8894
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19600562132831909,
      "learning_rate": 5.6304441817922606e-08,
      "loss": 0.011,
      "step": 8895
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.17746759661222283,
      "learning_rate": 5.577226269196656e-08,
      "loss": 0.0241,
      "step": 8896
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18966043082755105,
      "learning_rate": 5.5242607040069116e-08,
      "loss": 0.0251,
      "step": 8897
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21884208682325815,
      "learning_rate": 5.471547492924778e-08,
      "loss": 0.0357,
      "step": 8898
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2660924220878005,
      "learning_rate": 5.4190866426195866e-08,
      "loss": 0.0348,
      "step": 8899
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2184540159642352,
      "learning_rate": 5.3668781597291386e-08,
      "loss": 0.0285,
      "step": 8900
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.22007032841693855,
      "learning_rate": 5.314922050859484e-08,
      "loss": 0.0226,
      "step": 8901
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2103459044875899,
      "learning_rate": 5.263218322584029e-08,
      "loss": 0.0357,
      "step": 8902
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.24810193419444404,
      "learning_rate": 5.211766981444877e-08,
      "loss": 0.0512,
      "step": 8903
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.15084836706323024,
      "learning_rate": 5.160568033951929e-08,
      "loss": 0.0223,
      "step": 8904
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20629750087694815,
      "learning_rate": 5.109621486583116e-08,
      "loss": 0.0482,
      "step": 8905
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2717431159608309,
      "learning_rate": 5.058927345784836e-08,
      "loss": 0.0261,
      "step": 8906
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2657327433462176,
      "learning_rate": 5.0084856179708484e-08,
      "loss": 0.0362,
      "step": 8907
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.23068559273184816,
      "learning_rate": 4.9582963095238247e-08,
      "loss": 0.0289,
      "step": 8908
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21004634273455655,
      "learning_rate": 4.908359426793574e-08,
      "loss": 0.0325,
      "step": 8909
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21006131101701378,
      "learning_rate": 4.8586749760985987e-08,
      "loss": 0.0345,
      "step": 8910
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.25108688245401023,
      "learning_rate": 4.809242963725647e-08,
      "loss": 0.0419,
      "step": 8911
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2528863081682947,
      "learning_rate": 4.7600633959286044e-08,
      "loss": 0.0275,
      "step": 8912
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.25668979156396365,
      "learning_rate": 4.7111362789304946e-08,
      "loss": 0.0323,
      "step": 8913
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7035446398236099,
      "learning_rate": 4.6624616189214765e-08,
      "loss": 0.0311,
      "step": 8914
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2318738197800452,
      "learning_rate": 4.614039422060623e-08,
      "loss": 0.0337,
      "step": 8915
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2532999870246074,
      "learning_rate": 4.565869694474367e-08,
      "loss": 0.0375,
      "step": 8916
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18877799878022666,
      "learning_rate": 4.51795244225739e-08,
      "loss": 0.0169,
      "step": 8917
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21006230413546365,
      "learning_rate": 4.470287671472395e-08,
      "loss": 0.0265,
      "step": 8918
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1977127126480708,
      "learning_rate": 4.4228753881507826e-08,
      "loss": 0.0266,
      "step": 8919
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.230548489240827,
      "learning_rate": 4.375715598290864e-08,
      "loss": 0.0476,
      "step": 8920
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19867245251402507,
      "learning_rate": 4.3288083078600886e-08,
      "loss": 0.023,
      "step": 8921
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.17243855915490924,
      "learning_rate": 4.2821535227930424e-08,
      "loss": 0.0278,
      "step": 8922
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2076809513363692,
      "learning_rate": 4.235751248993003e-08,
      "loss": 0.0342,
      "step": 8923
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1839580371252363,
      "learning_rate": 4.1896014923310525e-08,
      "loss": 0.0216,
      "step": 8924
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.196581968334963,
      "learning_rate": 4.1437042586465194e-08,
      "loss": 0.0288,
      "step": 8925
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.14836014438568257,
      "learning_rate": 4.098059553746536e-08,
      "loss": 0.0168,
      "step": 8926
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20459380407756547,
      "learning_rate": 4.0526673834060394e-08,
      "loss": 0.0233,
      "step": 8927
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2334731632781089,
      "learning_rate": 4.0075277533688784e-08,
      "loss": 0.0371,
      "step": 8928
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.27081992806979993,
      "learning_rate": 3.9626406693460405e-08,
      "loss": 0.028,
      "step": 8929
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.17549744053004038,
      "learning_rate": 3.918006137017205e-08,
      "loss": 0.0305,
      "step": 8930
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1689352241988022,
      "learning_rate": 3.87362416202941e-08,
      "loss": 0.0235,
      "step": 8931
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2101084523637817,
      "learning_rate": 3.829494749998608e-08,
      "loss": 0.014,
      "step": 8932
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18920444606137404,
      "learning_rate": 3.78561790650811e-08,
      "loss": 0.0325,
      "step": 8933
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1607385957114975,
      "learning_rate": 3.7419936371094756e-08,
      "loss": 0.0282,
      "step": 8934
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20258937420489143,
      "learning_rate": 3.6986219473225116e-08,
      "loss": 0.0444,
      "step": 8935
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18878558623927508,
      "learning_rate": 3.655502842634606e-08,
      "loss": 0.042,
      "step": 8936
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21752551211437238,
      "learning_rate": 3.6126363285018396e-08,
      "loss": 0.0411,
      "step": 8937
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21324645935911626,
      "learning_rate": 3.570022410347651e-08,
      "loss": 0.021,
      "step": 8938
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2698570024453317,
      "learning_rate": 3.52766109356395e-08,
      "loss": 0.026,
      "step": 8939
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.15491729655500275,
      "learning_rate": 3.485552383510671e-08,
      "loss": 0.0215,
      "step": 8940
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.17882249820801224,
      "learning_rate": 3.44369628551533e-08,
      "loss": 0.0298,
      "step": 8941
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.22690129435099873,
      "learning_rate": 3.402092804874357e-08,
      "loss": 0.0233,
      "step": 8942
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18710486181625777,
      "learning_rate": 3.3607419468510985e-08,
      "loss": 0.0342,
      "step": 8943
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.23569667411577044,
      "learning_rate": 3.3196437166780336e-08,
      "loss": 0.0411,
      "step": 8944
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21990746574559097,
      "learning_rate": 3.278798119555004e-08,
      "loss": 0.0282,
      "step": 8945
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2083865644081224,
      "learning_rate": 3.2382051606500986e-08,
      "loss": 0.022,
      "step": 8946
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.17548957043037844,
      "learning_rate": 3.197864845099208e-08,
      "loss": 0.0295,
      "step": 8947
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.22755816792841235,
      "learning_rate": 3.1577771780066936e-08,
      "loss": 0.0258,
      "step": 8948
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2109330494729064,
      "learning_rate": 3.117942164444498e-08,
      "loss": 0.0519,
      "step": 8949
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2344377354265016,
      "learning_rate": 3.078359809453257e-08,
      "loss": 0.0291,
      "step": 8950
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1685005297684336,
      "learning_rate": 3.03903011804052e-08,
      "loss": 0.0188,
      "step": 8951
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20011099924009323,
      "learning_rate": 2.999953095182972e-08,
      "loss": 0.0148,
      "step": 8952
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2352683525669655,
      "learning_rate": 2.9611287458248817e-08,
      "loss": 0.036,
      "step": 8953
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.234208111155559,
      "learning_rate": 2.9225570748785402e-08,
      "loss": 0.039,
      "step": 8954
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.16460965515460493,
      "learning_rate": 2.884238087224045e-08,
      "loss": 0.0306,
      "step": 8955
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19101480433125886,
      "learning_rate": 2.8461717877099615e-08,
      "loss": 0.0108,
      "step": 8956
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18602885838859567,
      "learning_rate": 2.8083581811526594e-08,
      "loss": 0.0178,
      "step": 8957
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1981284280233722,
      "learning_rate": 2.770797272336756e-08,
      "loss": 0.04,
      "step": 8958
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20061913264836562,
      "learning_rate": 2.73348906601445e-08,
      "loss": 0.0397,
      "step": 8959
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.22872502766508018,
      "learning_rate": 2.696433566905965e-08,
      "loss": 0.0354,
      "step": 8960
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18010970974392435,
      "learning_rate": 2.6596307797004396e-08,
      "loss": 0.0331,
      "step": 8961
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.27038988923967716,
      "learning_rate": 2.623080709054149e-08,
      "loss": 0.0369,
      "step": 8962
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.26163690625313313,
      "learning_rate": 2.5867833595913937e-08,
      "loss": 0.038,
      "step": 8963
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19029744765225817,
      "learning_rate": 2.550738735905167e-08,
      "loss": 0.0295,
      "step": 8964
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20746219516115086,
      "learning_rate": 2.514946842555821e-08,
      "loss": 0.0324,
      "step": 8965
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.280971455394375,
      "learning_rate": 2.479407684071733e-08,
      "loss": 0.0333,
      "step": 8966
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20197736716358233,
      "learning_rate": 2.4441212649501944e-08,
      "loss": 0.0271,
      "step": 8967
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.16926910815215676,
      "learning_rate": 2.4090875896551903e-08,
      "loss": 0.0176,
      "step": 8968
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.24804209907319322,
      "learning_rate": 2.374306662619841e-08,
      "loss": 0.0491,
      "step": 8969
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21520106339954145,
      "learning_rate": 2.3397784882448483e-08,
      "loss": 0.0247,
      "step": 8970
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19443187550006327,
      "learning_rate": 2.305503070898718e-08,
      "loss": 0.0329,
      "step": 8971
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.196732158537213,
      "learning_rate": 2.2714804149184256e-08,
      "loss": 0.0285,
      "step": 8972
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2632305832283681,
      "learning_rate": 2.2377105246085272e-08,
      "loss": 0.0266,
      "step": 8973
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1710043224923331,
      "learning_rate": 2.2041934042420497e-08,
      "loss": 0.0201,
      "step": 8974
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2837634101925336,
      "learning_rate": 2.170929058059379e-08,
      "loss": 0.038,
      "step": 8975
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.17693066443913338,
      "learning_rate": 2.137917490269814e-08,
      "loss": 0.0274,
      "step": 8976
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2245182515936549,
      "learning_rate": 2.1051587050500144e-08,
      "loss": 0.0353,
      "step": 8977
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2584886354552909,
      "learning_rate": 2.0726527065448865e-08,
      "loss": 0.0351,
      "step": 8978
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1553510256118456,
      "learning_rate": 2.0403994988671404e-08,
      "loss": 0.0269,
      "step": 8979
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20004508732172224,
      "learning_rate": 2.0083990860977343e-08,
      "loss": 0.0353,
      "step": 8980
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22401136317552417,
      "learning_rate": 1.9766514722856512e-08,
      "loss": 0.0457,
      "step": 8981
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20792340639106136,
      "learning_rate": 1.9451566614479e-08,
      "loss": 0.0359,
      "step": 8982
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20801362401159895,
      "learning_rate": 1.9139146575690714e-08,
      "loss": 0.0284,
      "step": 8983
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.17488369142040103,
      "learning_rate": 1.8829254646022256e-08,
      "loss": 0.03,
      "step": 8984
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18400830277273272,
      "learning_rate": 1.85218908646867e-08,
      "loss": 0.0233,
      "step": 8985
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2610193415459683,
      "learning_rate": 1.8217055270568497e-08,
      "loss": 0.0293,
      "step": 8986
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.24891932686017573,
      "learning_rate": 1.7914747902241235e-08,
      "loss": 0.0313,
      "step": 8987
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21294748789274465,
      "learning_rate": 1.7614968797952102e-08,
      "loss": 0.0252,
      "step": 8988
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1993302799762338,
      "learning_rate": 1.7317717995632975e-08,
      "loss": 0.0352,
      "step": 8989
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19925219124255597,
      "learning_rate": 1.702299553289377e-08,
      "loss": 0.0404,
      "step": 8990
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.23484811714055948,
      "learning_rate": 1.6730801447024658e-08,
      "loss": 0.035,
      "step": 8991
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.23870087242973842,
      "learning_rate": 1.6441135774996066e-08,
      "loss": 0.0402,
      "step": 8992
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2903157978671264,
      "learning_rate": 1.6153998553456453e-08,
      "loss": 0.0285,
      "step": 8993
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2329017029833701,
      "learning_rate": 1.586938981873898e-08,
      "loss": 0.0358,
      "step": 8994
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19562284608575325,
      "learning_rate": 1.5587309606852618e-08,
      "loss": 0.0196,
      "step": 8995
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.14443401983996534,
      "learning_rate": 1.530775795348882e-08,
      "loss": 0.0158,
      "step": 8996
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.24249420434377358,
      "learning_rate": 1.5030734894019293e-08,
      "loss": 0.0318,
      "step": 8997
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18572297021909712,
      "learning_rate": 1.4756240463491555e-08,
      "loss": 0.0225,
      "step": 8998
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21422832080877244,
      "learning_rate": 1.4484274696637823e-08,
      "loss": 0.0382,
      "step": 8999
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2144251125919366,
      "learning_rate": 1.421483762787057e-08,
      "loss": 0.0309,
      "step": 9000
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3272001766456183,
      "learning_rate": 1.3947929291280305e-08,
      "loss": 0.0538,
      "step": 9001
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20169835502415198,
      "learning_rate": 1.368354972063557e-08,
      "loss": 0.026,
      "step": 9002
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1968762643712936,
      "learning_rate": 1.3421698949389605e-08,
      "loss": 0.0355,
      "step": 9003
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.17513244419490553,
      "learning_rate": 1.3162377010673689e-08,
      "loss": 0.0303,
      "step": 9004
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21097209434813735,
      "learning_rate": 1.2905583937299349e-08,
      "loss": 0.0308,
      "step": 9005
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28172995093871017,
      "learning_rate": 1.2651319761753933e-08,
      "loss": 0.066,
      "step": 9006
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18943717732114126,
      "learning_rate": 1.2399584516213925e-08,
      "loss": 0.0341,
      "step": 9007
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37573119244489,
      "learning_rate": 1.2150378232527183e-08,
      "loss": 0.0477,
      "step": 9008
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20770890514814155,
      "learning_rate": 1.1903700942224039e-08,
      "loss": 0.0384,
      "step": 9009
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22860660535083552,
      "learning_rate": 1.1659552676519525e-08,
      "loss": 0.0218,
      "step": 9010
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.15668467023711313,
      "learning_rate": 1.1417933466302267e-08,
      "loss": 0.0221,
      "step": 9011
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2610500675370838,
      "learning_rate": 1.1178843342143363e-08,
      "loss": 0.037,
      "step": 9012
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1800054970537677,
      "learning_rate": 1.0942282334294174e-08,
      "loss": 0.0258,
      "step": 9013
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.16902715982596964,
      "learning_rate": 1.070825047268631e-08,
      "loss": 0.0221,
      "step": 9014
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21877754412671224,
      "learning_rate": 1.047674778693164e-08,
      "loss": 0.032,
      "step": 9015
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19546011115117934,
      "learning_rate": 1.024777430632229e-08,
      "loss": 0.0313,
      "step": 9016
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.24489220708733828,
      "learning_rate": 1.0021330059826195e-08,
      "loss": 0.0347,
      "step": 9017
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.23904698483602776,
      "learning_rate": 9.797415076095996e-09,
      "loss": 0.0338,
      "step": 9018
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2555471509968717,
      "learning_rate": 9.57602938346458e-09,
      "loss": 0.037,
      "step": 9019
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.27180622425759465,
      "learning_rate": 9.357173009942878e-09,
      "loss": 0.0317,
      "step": 9020
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18345631873318244,
      "learning_rate": 9.140845983219848e-09,
      "loss": 0.0329,
      "step": 9021
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.16651174800898588,
      "learning_rate": 8.927048330666932e-09,
      "loss": 0.0183,
      "step": 9022
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.15833267747174332,
      "learning_rate": 8.715780079335823e-09,
      "loss": 0.0309,
      "step": 9023
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20065714900883727,
      "learning_rate": 8.50704125595847e-09,
      "loss": 0.0368,
      "step": 9024
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1898271565190854,
      "learning_rate": 8.30083188694486e-09,
      "loss": 0.0293,
      "step": 9025
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2175329616935779,
      "learning_rate": 8.097151998387453e-09,
      "loss": 0.0248,
      "step": 9026
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22140314270940384,
      "learning_rate": 7.896001616054527e-09,
      "loss": 0.0264,
      "step": 9027
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19572558588207523,
      "learning_rate": 7.697380765399053e-09,
      "loss": 0.0241,
      "step": 9028
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22397959778251336,
      "learning_rate": 7.501289471549821e-09,
      "loss": 0.0345,
      "step": 9029
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21757753396082377,
      "learning_rate": 7.3077277593203155e-09,
      "loss": 0.0431,
      "step": 9030
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.27201894917468206,
      "learning_rate": 7.116695653199834e-09,
      "loss": 0.0414,
      "step": 9031
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18455419520493435,
      "learning_rate": 6.928193177360154e-09,
      "loss": 0.0278,
      "step": 9032
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1934247532528776,
      "learning_rate": 6.742220355648865e-09,
      "loss": 0.0245,
      "step": 9033
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2428767640387947,
      "learning_rate": 6.558777211598255e-09,
      "loss": 0.0342,
      "step": 9034
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2293864473272074,
      "learning_rate": 6.377863768418646e-09,
      "loss": 0.0412,
      "step": 9035
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.24727667230934006,
      "learning_rate": 6.199480049000617e-09,
      "loss": 0.0312,
      "step": 9036
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2483851086657334,
      "learning_rate": 6.023626075915001e-09,
      "loss": 0.0309,
      "step": 9037
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2175901693849103,
      "learning_rate": 5.850301871410668e-09,
      "loss": 0.0413,
      "step": 9038
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2690359900481581,
      "learning_rate": 5.679507457418964e-09,
      "loss": 0.0363,
      "step": 9039
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.16839307593772218,
      "learning_rate": 5.51124285554927e-09,
      "loss": 0.0314,
      "step": 9040
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.17465031056518474,
      "learning_rate": 5.345508087093443e-09,
      "loss": 0.0305,
      "step": 9041
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22130723998113816,
      "learning_rate": 5.182303173016934e-09,
      "loss": 0.0392,
      "step": 9042
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20169247236703886,
      "learning_rate": 5.021628133972112e-09,
      "loss": 0.0154,
      "step": 9043
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18799913053045197,
      "learning_rate": 4.8634829902893806e-09,
      "loss": 0.033,
      "step": 9044
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2050763448002161,
      "learning_rate": 4.707867761977181e-09,
      "loss": 0.0337,
      "step": 9045
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21454403518954293,
      "learning_rate": 4.554782468726426e-09,
      "loss": 0.0366,
      "step": 9046
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22219991768636732,
      "learning_rate": 4.4042271299038486e-09,
      "loss": 0.0372,
      "step": 9047
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18634081183533258,
      "learning_rate": 4.256201764560874e-09,
      "loss": 0.0279,
      "step": 9048
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.27014418030622855,
      "learning_rate": 4.110706391424745e-09,
      "loss": 0.0298,
      "step": 9049
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.27799029945954745,
      "learning_rate": 3.967741028907401e-09,
      "loss": 0.0396,
      "step": 9050
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20356836898525382,
      "learning_rate": 3.827305695094375e-09,
      "loss": 0.0312,
      "step": 9051
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20310897030303288,
      "learning_rate": 3.6894004077558942e-09,
      "loss": 0.0309,
      "step": 9052
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19356217971297343,
      "learning_rate": 3.5540251843424468e-09,
      "loss": 0.0389,
      "step": 9053
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21035311245856395,
      "learning_rate": 3.421180041980332e-09,
      "loss": 0.0244,
      "step": 9054
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22289592826195415,
      "learning_rate": 3.290864997478327e-09,
      "loss": 0.0368,
      "step": 9055
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.16281274653652217,
      "learning_rate": 3.1630800673254636e-09,
      "loss": 0.0247,
      "step": 9056
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2661407596505153,
      "learning_rate": 3.0378252676910302e-09,
      "loss": 0.0268,
      "step": 9057
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3523807645754501,
      "learning_rate": 2.9151006144201298e-09,
      "loss": 0.0479,
      "step": 9058
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.34567161459555834,
      "learning_rate": 2.794906123042562e-09,
      "loss": 0.0401,
      "step": 9059
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1806135880353218,
      "learning_rate": 2.6772418087639417e-09,
      "loss": 0.024,
      "step": 9060
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19636057978983132,
      "learning_rate": 2.5621076864745796e-09,
      "loss": 0.0112,
      "step": 9061
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.17979448699359868,
      "learning_rate": 2.4495037707428226e-09,
      "loss": 0.0271,
      "step": 9062
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.25195775286542077,
      "learning_rate": 2.339430075812832e-09,
      "loss": 0.0334,
      "step": 9063
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2423025519365941,
      "learning_rate": 2.2318866156134654e-09,
      "loss": 0.026,
      "step": 9064
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2530028484093671,
      "learning_rate": 2.126873403751617e-09,
      "loss": 0.0564,
      "step": 9065
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.24952642468377387,
      "learning_rate": 2.0243904535144353e-09,
      "loss": 0.0349,
      "step": 9066
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.22973490115623516,
      "learning_rate": 1.924437777869326e-09,
      "loss": 0.0322,
      "step": 9067
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.17827245983148415,
      "learning_rate": 1.8270153894617282e-09,
      "loss": 0.0273,
      "step": 9068
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19810679467912204,
      "learning_rate": 1.7321233006173388e-09,
      "loss": 0.0231,
      "step": 9069
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.24555635055916267,
      "learning_rate": 1.6397615233465503e-09,
      "loss": 0.0375,
      "step": 9070
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2149386716440784,
      "learning_rate": 1.5499300693311293e-09,
      "loss": 0.0269,
      "step": 9071
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.18254295921288924,
      "learning_rate": 1.4626289499397596e-09,
      "loss": 0.0275,
      "step": 9072
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.17111133937911638,
      "learning_rate": 1.3778581762169397e-09,
      "loss": 0.0242,
      "step": 9073
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2517839647534739,
      "learning_rate": 1.2956177588896445e-09,
      "loss": 0.0399,
      "step": 9074
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.17626863305029444,
      "learning_rate": 1.215907708360664e-09,
      "loss": 0.021,
      "step": 9075
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2206262268864394,
      "learning_rate": 1.138728034719705e-09,
      "loss": 0.0423,
      "step": 9076
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2505197784073812,
      "learning_rate": 1.0640787477300685e-09,
      "loss": 0.0334,
      "step": 9077
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.24630506465030078,
      "learning_rate": 9.919598568353118e-10,
      "loss": 0.0364,
      "step": 9078
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19157908868761192,
      "learning_rate": 9.223713711636883e-10,
      "loss": 0.0273,
      "step": 9079
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3398292911128397,
      "learning_rate": 8.553132995170466e-10,
      "loss": 0.0352,
      "step": 9080
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.41283786259721683,
      "learning_rate": 7.907856503797106e-10,
      "loss": 0.0417,
      "step": 9081
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19454371783888974,
      "learning_rate": 7.287884319184813e-10,
      "loss": 0.0294,
      "step": 9082
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.22095453239841353,
      "learning_rate": 6.693216519781942e-10,
      "loss": 0.0402,
      "step": 9083
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.24136782189091516,
      "learning_rate": 6.123853180795003e-10,
      "loss": 0.0391,
      "step": 9084
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.18302363283545298,
      "learning_rate": 5.579794374299674e-10,
      "loss": 0.0127,
      "step": 9085
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19306094326380838,
      "learning_rate": 5.061040169107578e-10,
      "loss": 0.0238,
      "step": 9086
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.308372104093069,
      "learning_rate": 4.567590630855101e-10,
      "loss": 0.053,
      "step": 9087
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.20385748583398602,
      "learning_rate": 4.0994458220033896e-10,
      "loss": 0.0195,
      "step": 9088
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.17134609990906197,
      "learning_rate": 3.65660580177174e-10,
      "loss": 0.0251,
      "step": 9089
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2556640520088537,
      "learning_rate": 3.2390706261598015e-10,
      "loss": 0.0499,
      "step": 9090
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1805423532808625,
      "learning_rate": 2.846840348058599e-10,
      "loss": 0.0358,
      "step": 9091
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.20602725962832472,
      "learning_rate": 2.479915017028489e-10,
      "loss": 0.033,
      "step": 9092
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.16231528384142585,
      "learning_rate": 2.1382946795434067e-10,
      "loss": 0.0201,
      "step": 9093
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1820822341476219,
      "learning_rate": 1.8219793788132322e-10,
      "loss": 0.0226,
      "step": 9094
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.21737306509588128,
      "learning_rate": 1.530969154872608e-10,
      "loss": 0.0281,
      "step": 9095
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.18921095324076428,
      "learning_rate": 1.265264044514325e-10,
      "loss": 0.0298,
      "step": 9096
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.23190776480357483,
      "learning_rate": 1.0248640813781407e-10,
      "loss": 0.0323,
      "step": 9097
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.17537146593218583,
      "learning_rate": 8.097692958619619e-11,
      "loss": 0.027,
      "step": 9098
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19028406688718302,
      "learning_rate": 6.199797151884568e-11,
      "loss": 0.0175,
      "step": 9099
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.26877648644450103,
      "learning_rate": 4.5549536340505627e-11,
      "loss": 0.0268,
      "step": 9100
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.18589950982842812,
      "learning_rate": 3.163162612729309e-11,
      "loss": 0.0249,
      "step": 9101
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1556310015501666,
      "learning_rate": 2.024424264224223e-11,
      "loss": 0.025,
      "step": 9102
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.20509211174634498,
      "learning_rate": 1.1387387326422528e-11,
      "loss": 0.0379,
      "step": 9103
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1919688015546051,
      "learning_rate": 5.061061301159242e-12,
      "loss": 0.0141,
      "step": 9104
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.23409397443303054,
      "learning_rate": 1.265265365812951e-12,
      "loss": 0.0448,
      "step": 9105
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1942362122116247,
      "learning_rate": 0.0,
      "loss": 0.0345,
      "step": 9106
    },
    {
      "epoch": 1.0,
      "step": 9106,
      "total_flos": 8.962615037143621e+21,
      "train_loss": 0.14248651401592877,
      "train_runtime": 38094.5224,
      "train_samples_per_second": 7.649,
      "train_steps_per_second": 0.239
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 9106,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 3000,
  "total_flos": 8.962615037143621e+21,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}