{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8,
  "eval_steps": 500,
  "global_step": 3640,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00021978021978021978,
      "grad_norm": 1.2186660196817165,
      "learning_rate": 6.999999249142552e-07,
      "loss": 1.5998,
      "step": 1
    },
    {
      "epoch": 0.00043956043956043956,
      "grad_norm": 1.0439899552298968,
      "learning_rate": 6.999996996570566e-07,
      "loss": 1.6101,
      "step": 2
    },
    {
      "epoch": 0.0006593406593406593,
      "grad_norm": 1.0721695105594953,
      "learning_rate": 6.999993242285117e-07,
      "loss": 1.5086,
      "step": 3
    },
    {
      "epoch": 0.0008791208791208791,
      "grad_norm": 1.1098815548570793,
      "learning_rate": 6.999987986287993e-07,
      "loss": 1.5492,
      "step": 4
    },
    {
      "epoch": 0.001098901098901099,
      "grad_norm": 1.1307841967232446,
      "learning_rate": 6.999981228581701e-07,
      "loss": 1.5344,
      "step": 5
    },
    {
      "epoch": 0.0013186813186813187,
      "grad_norm": 0.9920620937163445,
      "learning_rate": 6.999972969169463e-07,
      "loss": 1.558,
      "step": 6
    },
    {
      "epoch": 0.0015384615384615385,
      "grad_norm": 1.024132361415856,
      "learning_rate": 6.999963208055215e-07,
      "loss": 1.5217,
      "step": 7
    },
    {
      "epoch": 0.0017582417582417582,
      "grad_norm": 0.9309822643703438,
      "learning_rate": 6.999951945243612e-07,
      "loss": 1.541,
      "step": 8
    },
    {
      "epoch": 0.001978021978021978,
      "grad_norm": 0.8249543682534687,
      "learning_rate": 6.999939180740022e-07,
      "loss": 1.5733,
      "step": 9
    },
    {
      "epoch": 0.002197802197802198,
      "grad_norm": 0.9676935755823561,
      "learning_rate": 6.999924914550531e-07,
      "loss": 1.5692,
      "step": 10
    },
    {
      "epoch": 0.0024175824175824176,
      "grad_norm": 0.8980208205863951,
      "learning_rate": 6.999909146681941e-07,
      "loss": 1.565,
      "step": 11
    },
    {
      "epoch": 0.0026373626373626374,
      "grad_norm": 0.9306418951850254,
      "learning_rate": 6.999891877141767e-07,
      "loss": 1.5342,
      "step": 12
    },
    {
      "epoch": 0.002857142857142857,
      "grad_norm": 0.7374753300540053,
      "learning_rate": 6.999873105938243e-07,
      "loss": 1.5431,
      "step": 13
    },
    {
      "epoch": 0.003076923076923077,
      "grad_norm": 0.6068473940709174,
      "learning_rate": 6.99985283308032e-07,
      "loss": 1.5225,
      "step": 14
    },
    {
      "epoch": 0.0032967032967032967,
      "grad_norm": 0.6019875923751878,
      "learning_rate": 6.999831058577659e-07,
      "loss": 1.4669,
      "step": 15
    },
    {
      "epoch": 0.0035164835164835165,
      "grad_norm": 0.6249268931056989,
      "learning_rate": 6.999807782440644e-07,
      "loss": 1.5616,
      "step": 16
    },
    {
      "epoch": 0.0037362637362637363,
      "grad_norm": 0.5835249297903639,
      "learning_rate": 6.99978300468037e-07,
      "loss": 1.5251,
      "step": 17
    },
    {
      "epoch": 0.003956043956043956,
      "grad_norm": 0.6063786110343348,
      "learning_rate": 6.999756725308648e-07,
      "loss": 1.531,
      "step": 18
    },
    {
      "epoch": 0.004175824175824176,
      "grad_norm": 0.5755540396780645,
      "learning_rate": 6.999728944338009e-07,
      "loss": 1.5712,
      "step": 19
    },
    {
      "epoch": 0.004395604395604396,
      "grad_norm": 0.5687868093148271,
      "learning_rate": 6.999699661781696e-07,
      "loss": 1.5716,
      "step": 20
    },
    {
      "epoch": 0.004615384615384616,
      "grad_norm": 0.5622248804783495,
      "learning_rate": 6.999668877653669e-07,
      "loss": 1.4995,
      "step": 21
    },
    {
      "epoch": 0.004835164835164835,
      "grad_norm": 0.5122047997706676,
      "learning_rate": 6.999636591968604e-07,
      "loss": 1.5336,
      "step": 22
    },
    {
      "epoch": 0.005054945054945055,
      "grad_norm": 0.5376748142392922,
      "learning_rate": 6.999602804741893e-07,
      "loss": 1.5249,
      "step": 23
    },
    {
      "epoch": 0.005274725274725275,
      "grad_norm": 0.4960039885042067,
      "learning_rate": 6.999567515989641e-07,
      "loss": 1.5907,
      "step": 24
    },
    {
      "epoch": 0.005494505494505495,
      "grad_norm": 0.4868968983538848,
      "learning_rate": 6.999530725728675e-07,
      "loss": 1.5094,
      "step": 25
    },
    {
      "epoch": 0.005714285714285714,
      "grad_norm": 0.4509570161471638,
      "learning_rate": 6.999492433976532e-07,
      "loss": 1.5193,
      "step": 26
    },
    {
      "epoch": 0.0059340659340659345,
      "grad_norm": 0.43012066136835875,
      "learning_rate": 6.999452640751469e-07,
      "loss": 1.4869,
      "step": 27
    },
    {
      "epoch": 0.006153846153846154,
      "grad_norm": 0.42798833956354815,
      "learning_rate": 6.999411346072455e-07,
      "loss": 1.5537,
      "step": 28
    },
    {
      "epoch": 0.006373626373626374,
      "grad_norm": 0.42110987188751625,
      "learning_rate": 6.999368549959178e-07,
      "loss": 1.4695,
      "step": 29
    },
    {
      "epoch": 0.006593406593406593,
      "grad_norm": 0.42241293601772917,
      "learning_rate": 6.999324252432038e-07,
      "loss": 1.5394,
      "step": 30
    },
    {
      "epoch": 0.006813186813186814,
      "grad_norm": 0.40830105748996753,
      "learning_rate": 6.999278453512156e-07,
      "loss": 1.5436,
      "step": 31
    },
    {
      "epoch": 0.007032967032967033,
      "grad_norm": 0.42088049458463134,
      "learning_rate": 6.999231153221364e-07,
      "loss": 1.517,
      "step": 32
    },
    {
      "epoch": 0.007252747252747253,
      "grad_norm": 0.4254524125760086,
      "learning_rate": 6.999182351582212e-07,
      "loss": 1.4963,
      "step": 33
    },
    {
      "epoch": 0.0074725274725274725,
      "grad_norm": 0.45211208077565246,
      "learning_rate": 6.999132048617966e-07,
      "loss": 1.5334,
      "step": 34
    },
    {
      "epoch": 0.007692307692307693,
      "grad_norm": 0.38954848298052885,
      "learning_rate": 6.999080244352607e-07,
      "loss": 1.4666,
      "step": 35
    },
    {
      "epoch": 0.007912087912087912,
      "grad_norm": 0.37662572294061986,
      "learning_rate": 6.999026938810832e-07,
      "loss": 1.5887,
      "step": 36
    },
    {
      "epoch": 0.008131868131868131,
      "grad_norm": 0.3800537277157674,
      "learning_rate": 6.998972132018054e-07,
      "loss": 1.5428,
      "step": 37
    },
    {
      "epoch": 0.008351648351648353,
      "grad_norm": 0.39327960840253473,
      "learning_rate": 6.998915824000401e-07,
      "loss": 1.5812,
      "step": 38
    },
    {
      "epoch": 0.008571428571428572,
      "grad_norm": 14.112700324145928,
      "learning_rate": 6.998858014784716e-07,
      "loss": 1.5788,
      "step": 39
    },
    {
      "epoch": 0.008791208791208791,
      "grad_norm": 0.35127700103550985,
      "learning_rate": 6.99879870439856e-07,
      "loss": 1.5079,
      "step": 40
    },
    {
      "epoch": 0.00901098901098901,
      "grad_norm": 0.33547200423608775,
      "learning_rate": 6.998737892870208e-07,
      "loss": 1.4971,
      "step": 41
    },
    {
      "epoch": 0.009230769230769232,
      "grad_norm": 0.3463399443779019,
      "learning_rate": 6.99867558022865e-07,
      "loss": 1.5603,
      "step": 42
    },
    {
      "epoch": 0.009450549450549451,
      "grad_norm": 0.5100467180860531,
      "learning_rate": 6.998611766503595e-07,
      "loss": 1.5597,
      "step": 43
    },
    {
      "epoch": 0.00967032967032967,
      "grad_norm": 0.34297872378392785,
      "learning_rate": 6.998546451725461e-07,
      "loss": 1.4844,
      "step": 44
    },
    {
      "epoch": 0.00989010989010989,
      "grad_norm": 0.36743100193234246,
      "learning_rate": 6.99847963592539e-07,
      "loss": 1.5067,
      "step": 45
    },
    {
      "epoch": 0.01010989010989011,
      "grad_norm": 0.34786527855699156,
      "learning_rate": 6.998411319135234e-07,
      "loss": 1.5487,
      "step": 46
    },
    {
      "epoch": 0.01032967032967033,
      "grad_norm": 0.3520019386286944,
      "learning_rate": 6.998341501387561e-07,
      "loss": 1.5092,
      "step": 47
    },
    {
      "epoch": 0.01054945054945055,
      "grad_norm": 0.34687698726669514,
      "learning_rate": 6.998270182715658e-07,
      "loss": 1.5078,
      "step": 48
    },
    {
      "epoch": 0.010769230769230769,
      "grad_norm": 0.33420958814829355,
      "learning_rate": 6.998197363153522e-07,
      "loss": 1.4704,
      "step": 49
    },
    {
      "epoch": 0.01098901098901099,
      "grad_norm": 0.33024025304703036,
      "learning_rate": 6.998123042735871e-07,
      "loss": 1.5104,
      "step": 50
    },
    {
      "epoch": 0.01120879120879121,
      "grad_norm": 13.292323804745312,
      "learning_rate": 6.998047221498137e-07,
      "loss": 1.5458,
      "step": 51
    },
    {
      "epoch": 0.011428571428571429,
      "grad_norm": 0.3231695044826081,
      "learning_rate": 6.997969899476462e-07,
      "loss": 1.5122,
      "step": 52
    },
    {
      "epoch": 0.011648351648351648,
      "grad_norm": 0.331099451474555,
      "learning_rate": 6.997891076707711e-07,
      "loss": 1.5195,
      "step": 53
    },
    {
      "epoch": 0.011868131868131869,
      "grad_norm": 0.34515503143462656,
      "learning_rate": 6.997810753229464e-07,
      "loss": 1.5428,
      "step": 54
    },
    {
      "epoch": 0.012087912087912088,
      "grad_norm": 0.34544004796040934,
      "learning_rate": 6.997728929080011e-07,
      "loss": 1.5598,
      "step": 55
    },
    {
      "epoch": 0.012307692307692308,
      "grad_norm": 0.32630043594715297,
      "learning_rate": 6.99764560429836e-07,
      "loss": 1.5739,
      "step": 56
    },
    {
      "epoch": 0.012527472527472527,
      "grad_norm": 0.33162041156072575,
      "learning_rate": 6.997560778924236e-07,
      "loss": 1.5091,
      "step": 57
    },
    {
      "epoch": 0.012747252747252748,
      "grad_norm": 0.32637573914797435,
      "learning_rate": 6.997474452998078e-07,
      "loss": 1.4946,
      "step": 58
    },
    {
      "epoch": 0.012967032967032967,
      "grad_norm": 0.3241571015698132,
      "learning_rate": 6.997386626561041e-07,
      "loss": 1.4998,
      "step": 59
    },
    {
      "epoch": 0.013186813186813187,
      "grad_norm": 0.3105444297848675,
      "learning_rate": 6.997297299654995e-07,
      "loss": 1.5221,
      "step": 60
    },
    {
      "epoch": 0.013406593406593406,
      "grad_norm": 0.32279738495572,
      "learning_rate": 6.997206472322524e-07,
      "loss": 1.531,
      "step": 61
    },
    {
      "epoch": 0.013626373626373627,
      "grad_norm": 0.321559866825534,
      "learning_rate": 6.99711414460693e-07,
      "loss": 1.5037,
      "step": 62
    },
    {
      "epoch": 0.013846153846153847,
      "grad_norm": 0.31003333500380825,
      "learning_rate": 6.997020316552228e-07,
      "loss": 1.5349,
      "step": 63
    },
    {
      "epoch": 0.014065934065934066,
      "grad_norm": 0.3226878420193051,
      "learning_rate": 6.996924988203149e-07,
      "loss": 1.5072,
      "step": 64
    },
    {
      "epoch": 0.014285714285714285,
      "grad_norm": 0.30960484322916615,
      "learning_rate": 6.99682815960514e-07,
      "loss": 1.4557,
      "step": 65
    },
    {
      "epoch": 0.014505494505494506,
      "grad_norm": 0.31173035549535516,
      "learning_rate": 6.996729830804362e-07,
      "loss": 1.55,
      "step": 66
    },
    {
      "epoch": 0.014725274725274726,
      "grad_norm": 0.33346114160057794,
      "learning_rate": 6.996630001847694e-07,
      "loss": 1.5098,
      "step": 67
    },
    {
      "epoch": 0.014945054945054945,
      "grad_norm": 0.3262332746170292,
      "learning_rate": 6.996528672782724e-07,
      "loss": 1.4954,
      "step": 68
    },
    {
      "epoch": 0.015164835164835164,
      "grad_norm": 0.3066424118028698,
      "learning_rate": 6.996425843657762e-07,
      "loss": 1.5247,
      "step": 69
    },
    {
      "epoch": 0.015384615384615385,
      "grad_norm": 0.29511754230899856,
      "learning_rate": 6.996321514521829e-07,
      "loss": 1.4867,
      "step": 70
    },
    {
      "epoch": 0.015604395604395605,
      "grad_norm": 0.3029790265153383,
      "learning_rate": 6.996215685424661e-07,
      "loss": 1.4409,
      "step": 71
    },
    {
      "epoch": 0.015824175824175824,
      "grad_norm": 0.3215987443977887,
      "learning_rate": 6.996108356416715e-07,
      "loss": 1.482,
      "step": 72
    },
    {
      "epoch": 0.016043956043956045,
      "grad_norm": 0.30330633610502405,
      "learning_rate": 6.995999527549153e-07,
      "loss": 1.5087,
      "step": 73
    },
    {
      "epoch": 0.016263736263736263,
      "grad_norm": 0.3003227941293912,
      "learning_rate": 6.995889198873862e-07,
      "loss": 1.4914,
      "step": 74
    },
    {
      "epoch": 0.016483516483516484,
      "grad_norm": 0.34686838487569577,
      "learning_rate": 6.995777370443436e-07,
      "loss": 1.4711,
      "step": 75
    },
    {
      "epoch": 0.016703296703296705,
      "grad_norm": 0.28588606950024403,
      "learning_rate": 6.995664042311191e-07,
      "loss": 1.5046,
      "step": 76
    },
    {
      "epoch": 0.016923076923076923,
      "grad_norm": 0.3090307833343753,
      "learning_rate": 6.995549214531152e-07,
      "loss": 1.5053,
      "step": 77
    },
    {
      "epoch": 0.017142857142857144,
      "grad_norm": 0.279920066376002,
      "learning_rate": 6.995432887158061e-07,
      "loss": 1.4293,
      "step": 78
    },
    {
      "epoch": 0.01736263736263736,
      "grad_norm": 0.3205459040940839,
      "learning_rate": 6.995315060247377e-07,
      "loss": 1.5683,
      "step": 79
    },
    {
      "epoch": 0.017582417582417582,
      "grad_norm": 0.2933165344735273,
      "learning_rate": 6.995195733855271e-07,
      "loss": 1.496,
      "step": 80
    },
    {
      "epoch": 0.017802197802197803,
      "grad_norm": 0.28611460012927237,
      "learning_rate": 6.995074908038631e-07,
      "loss": 1.5161,
      "step": 81
    },
    {
      "epoch": 0.01802197802197802,
      "grad_norm": 0.281541569510838,
      "learning_rate": 6.994952582855058e-07,
      "loss": 1.5087,
      "step": 82
    },
    {
      "epoch": 0.018241758241758242,
      "grad_norm": 2.608706443624609,
      "learning_rate": 6.99482875836287e-07,
      "loss": 1.5499,
      "step": 83
    },
    {
      "epoch": 0.018461538461538463,
      "grad_norm": 0.30022043627485206,
      "learning_rate": 6.994703434621097e-07,
      "loss": 1.5573,
      "step": 84
    },
    {
      "epoch": 0.01868131868131868,
      "grad_norm": 0.3925486655499724,
      "learning_rate": 6.994576611689486e-07,
      "loss": 1.494,
      "step": 85
    },
    {
      "epoch": 0.018901098901098902,
      "grad_norm": 0.2843534465338799,
      "learning_rate": 6.994448289628498e-07,
      "loss": 1.4763,
      "step": 86
    },
    {
      "epoch": 0.01912087912087912,
      "grad_norm": 0.2869825542786904,
      "learning_rate": 6.994318468499308e-07,
      "loss": 1.51,
      "step": 87
    },
    {
      "epoch": 0.01934065934065934,
      "grad_norm": 0.3116759886396289,
      "learning_rate": 6.994187148363806e-07,
      "loss": 1.4883,
      "step": 88
    },
    {
      "epoch": 0.01956043956043956,
      "grad_norm": 0.297394744452125,
      "learning_rate": 6.994054329284599e-07,
      "loss": 1.5885,
      "step": 89
    },
    {
      "epoch": 0.01978021978021978,
      "grad_norm": 0.2918854080020701,
      "learning_rate": 6.993920011325003e-07,
      "loss": 1.4858,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2940304534340659,
      "learning_rate": 6.993784194549055e-07,
      "loss": 1.4796,
      "step": 91
    },
    {
      "epoch": 0.02021978021978022,
      "grad_norm": 0.32308529480303205,
      "learning_rate": 6.993646879021503e-07,
      "loss": 1.4847,
      "step": 92
    },
    {
      "epoch": 0.02043956043956044,
      "grad_norm": 0.29725992838234777,
      "learning_rate": 6.993508064807809e-07,
      "loss": 1.5481,
      "step": 93
    },
    {
      "epoch": 0.02065934065934066,
      "grad_norm": 0.2959729110272764,
      "learning_rate": 6.993367751974151e-07,
      "loss": 1.4977,
      "step": 94
    },
    {
      "epoch": 0.020879120879120878,
      "grad_norm": 0.29995140162354467,
      "learning_rate": 6.993225940587423e-07,
      "loss": 1.5682,
      "step": 95
    },
    {
      "epoch": 0.0210989010989011,
      "grad_norm": 0.27708359576077696,
      "learning_rate": 6.993082630715229e-07,
      "loss": 1.4819,
      "step": 96
    },
    {
      "epoch": 0.02131868131868132,
      "grad_norm": 0.2858375043856483,
      "learning_rate": 6.99293782242589e-07,
      "loss": 1.4997,
      "step": 97
    },
    {
      "epoch": 0.021538461538461538,
      "grad_norm": 0.3242764772196802,
      "learning_rate": 6.992791515788442e-07,
      "loss": 1.4909,
      "step": 98
    },
    {
      "epoch": 0.02175824175824176,
      "grad_norm": 0.2855918680306477,
      "learning_rate": 6.992643710872633e-07,
      "loss": 1.507,
      "step": 99
    },
    {
      "epoch": 0.02197802197802198,
      "grad_norm": 0.29129408787298827,
      "learning_rate": 6.992494407748931e-07,
      "loss": 1.5033,
      "step": 100
    },
    {
      "epoch": 0.022197802197802197,
      "grad_norm": 0.2831537449986603,
      "learning_rate": 6.992343606488509e-07,
      "loss": 1.4584,
      "step": 101
    },
    {
      "epoch": 0.02241758241758242,
      "grad_norm": 0.30566588359048896,
      "learning_rate": 6.99219130716326e-07,
      "loss": 1.4876,
      "step": 102
    },
    {
      "epoch": 0.022637362637362636,
      "grad_norm": 0.28003441401871154,
      "learning_rate": 6.992037509845793e-07,
      "loss": 1.5287,
      "step": 103
    },
    {
      "epoch": 0.022857142857142857,
      "grad_norm": 0.2776071350189179,
      "learning_rate": 6.991882214609427e-07,
      "loss": 1.5196,
      "step": 104
    },
    {
      "epoch": 0.023076923076923078,
      "grad_norm": 0.28320343050708896,
      "learning_rate": 6.991725421528197e-07,
      "loss": 1.4868,
      "step": 105
    },
    {
      "epoch": 0.023296703296703296,
      "grad_norm": 0.28999603887230935,
      "learning_rate": 6.991567130676851e-07,
      "loss": 1.4722,
      "step": 106
    },
    {
      "epoch": 0.023516483516483517,
      "grad_norm": 0.3264889773497954,
      "learning_rate": 6.991407342130853e-07,
      "loss": 1.5275,
      "step": 107
    },
    {
      "epoch": 0.023736263736263738,
      "grad_norm": 0.2789118306208069,
      "learning_rate": 6.991246055966378e-07,
      "loss": 1.5081,
      "step": 108
    },
    {
      "epoch": 0.023956043956043956,
      "grad_norm": 0.27960937704917793,
      "learning_rate": 6.991083272260319e-07,
      "loss": 1.4807,
      "step": 109
    },
    {
      "epoch": 0.024175824175824177,
      "grad_norm": 0.28191339204154914,
      "learning_rate": 6.990918991090279e-07,
      "loss": 1.5313,
      "step": 110
    },
    {
      "epoch": 0.024395604395604394,
      "grad_norm": 0.2945719571612475,
      "learning_rate": 6.990753212534577e-07,
      "loss": 1.5475,
      "step": 111
    },
    {
      "epoch": 0.024615384615384615,
      "grad_norm": 0.2751109973724785,
      "learning_rate": 6.990585936672245e-07,
      "loss": 1.5284,
      "step": 112
    },
    {
      "epoch": 0.024835164835164836,
      "grad_norm": 0.2824020087608803,
      "learning_rate": 6.990417163583031e-07,
      "loss": 1.5038,
      "step": 113
    },
    {
      "epoch": 0.025054945054945054,
      "grad_norm": 0.3012447042602539,
      "learning_rate": 6.990246893347393e-07,
      "loss": 1.5257,
      "step": 114
    },
    {
      "epoch": 0.025274725274725275,
      "grad_norm": 0.3373183520802013,
      "learning_rate": 6.990075126046506e-07,
      "loss": 1.5136,
      "step": 115
    },
    {
      "epoch": 0.025494505494505496,
      "grad_norm": 0.26947512348674646,
      "learning_rate": 6.989901861762256e-07,
      "loss": 1.4885,
      "step": 116
    },
    {
      "epoch": 0.025714285714285714,
      "grad_norm": 0.30543440043549336,
      "learning_rate": 6.989727100577246e-07,
      "loss": 1.4932,
      "step": 117
    },
    {
      "epoch": 0.025934065934065935,
      "grad_norm": 0.27272901688360696,
      "learning_rate": 6.98955084257479e-07,
      "loss": 1.4918,
      "step": 118
    },
    {
      "epoch": 0.026153846153846153,
      "grad_norm": 0.2864790696227162,
      "learning_rate": 6.989373087838915e-07,
      "loss": 1.5619,
      "step": 119
    },
    {
      "epoch": 0.026373626373626374,
      "grad_norm": 0.2756479522819835,
      "learning_rate": 6.989193836454364e-07,
      "loss": 1.537,
      "step": 120
    },
    {
      "epoch": 0.026593406593406595,
      "grad_norm": 0.27232865699953096,
      "learning_rate": 6.989013088506593e-07,
      "loss": 1.5046,
      "step": 121
    },
    {
      "epoch": 0.026813186813186812,
      "grad_norm": 0.268028106647397,
      "learning_rate": 6.988830844081771e-07,
      "loss": 1.477,
      "step": 122
    },
    {
      "epoch": 0.027032967032967033,
      "grad_norm": 0.27821815601753497,
      "learning_rate": 6.988647103266779e-07,
      "loss": 1.5441,
      "step": 123
    },
    {
      "epoch": 0.027252747252747254,
      "grad_norm": 0.27697543965729127,
      "learning_rate": 6.988461866149214e-07,
      "loss": 1.6002,
      "step": 124
    },
    {
      "epoch": 0.027472527472527472,
      "grad_norm": 0.29538619865773286,
      "learning_rate": 6.988275132817383e-07,
      "loss": 1.5378,
      "step": 125
    },
    {
      "epoch": 0.027692307692307693,
      "grad_norm": 0.2775367321371363,
      "learning_rate": 6.988086903360311e-07,
      "loss": 1.4762,
      "step": 126
    },
    {
      "epoch": 0.02791208791208791,
      "grad_norm": 0.2702662287092996,
      "learning_rate": 6.987897177867731e-07,
      "loss": 1.4599,
      "step": 127
    },
    {
      "epoch": 0.028131868131868132,
      "grad_norm": 0.28656388789553466,
      "learning_rate": 6.987705956430092e-07,
      "loss": 1.5213,
      "step": 128
    },
    {
      "epoch": 0.028351648351648353,
      "grad_norm": 0.2712634686760554,
      "learning_rate": 6.987513239138558e-07,
      "loss": 1.5791,
      "step": 129
    },
    {
      "epoch": 0.02857142857142857,
      "grad_norm": 0.2905484073509738,
      "learning_rate": 6.987319026085003e-07,
      "loss": 1.4899,
      "step": 130
    },
    {
      "epoch": 0.02879120879120879,
      "grad_norm": 0.2846977667808167,
      "learning_rate": 6.987123317362014e-07,
      "loss": 1.4881,
      "step": 131
    },
    {
      "epoch": 0.029010989010989013,
      "grad_norm": 0.2691320851817523,
      "learning_rate": 6.986926113062894e-07,
      "loss": 1.4857,
      "step": 132
    },
    {
      "epoch": 0.02923076923076923,
      "grad_norm": 0.2911894232325037,
      "learning_rate": 6.986727413281656e-07,
      "loss": 1.4855,
      "step": 133
    },
    {
      "epoch": 0.02945054945054945,
      "grad_norm": 0.305082177212358,
      "learning_rate": 6.986527218113028e-07,
      "loss": 1.5369,
      "step": 134
    },
    {
      "epoch": 0.02967032967032967,
      "grad_norm": 0.27036465289461803,
      "learning_rate": 6.986325527652449e-07,
      "loss": 1.5031,
      "step": 135
    },
    {
      "epoch": 0.02989010989010989,
      "grad_norm": 0.2631354750255763,
      "learning_rate": 6.986122341996072e-07,
      "loss": 1.4465,
      "step": 136
    },
    {
      "epoch": 0.03010989010989011,
      "grad_norm": 0.2752652309390088,
      "learning_rate": 6.985917661240763e-07,
      "loss": 1.4614,
      "step": 137
    },
    {
      "epoch": 0.03032967032967033,
      "grad_norm": 0.32996418437558866,
      "learning_rate": 6.985711485484101e-07,
      "loss": 1.5232,
      "step": 138
    },
    {
      "epoch": 0.03054945054945055,
      "grad_norm": 0.2812242833122952,
      "learning_rate": 6.985503814824375e-07,
      "loss": 1.5128,
      "step": 139
    },
    {
      "epoch": 0.03076923076923077,
      "grad_norm": 0.2728686694891554,
      "learning_rate": 6.985294649360593e-07,
      "loss": 1.472,
      "step": 140
    },
    {
      "epoch": 0.03098901098901099,
      "grad_norm": 0.2710365904840655,
      "learning_rate": 6.985083989192469e-07,
      "loss": 1.5455,
      "step": 141
    },
    {
      "epoch": 0.03120879120879121,
      "grad_norm": 0.29439651848838755,
      "learning_rate": 6.984871834420429e-07,
      "loss": 1.5177,
      "step": 142
    },
    {
      "epoch": 0.03142857142857143,
      "grad_norm": 0.26500957178841184,
      "learning_rate": 6.98465818514562e-07,
      "loss": 1.4676,
      "step": 143
    },
    {
      "epoch": 0.03164835164835165,
      "grad_norm": 0.27673963910531124,
      "learning_rate": 6.984443041469894e-07,
      "loss": 1.5072,
      "step": 144
    },
    {
      "epoch": 0.031868131868131866,
      "grad_norm": 0.2750426966896585,
      "learning_rate": 6.984226403495816e-07,
      "loss": 1.4873,
      "step": 145
    },
    {
      "epoch": 0.03208791208791209,
      "grad_norm": 0.28280202577217106,
      "learning_rate": 6.984008271326666e-07,
      "loss": 1.4351,
      "step": 146
    },
    {
      "epoch": 0.03230769230769231,
      "grad_norm": 0.28194677251705746,
      "learning_rate": 6.983788645066436e-07,
      "loss": 1.4858,
      "step": 147
    },
    {
      "epoch": 0.032527472527472526,
      "grad_norm": 0.2818612120674954,
      "learning_rate": 6.983567524819828e-07,
      "loss": 1.4792,
      "step": 148
    },
    {
      "epoch": 0.03274725274725275,
      "grad_norm": 0.25947122206030515,
      "learning_rate": 6.98334491069226e-07,
      "loss": 1.4632,
      "step": 149
    },
    {
      "epoch": 0.03296703296703297,
      "grad_norm": 0.2743375323362592,
      "learning_rate": 6.983120802789857e-07,
      "loss": 1.4429,
      "step": 150
    },
    {
      "epoch": 0.033186813186813185,
      "grad_norm": 0.34639656463234103,
      "learning_rate": 6.98289520121946e-07,
      "loss": 1.4937,
      "step": 151
    },
    {
      "epoch": 0.03340659340659341,
      "grad_norm": 0.305354077564706,
      "learning_rate": 6.982668106088623e-07,
      "loss": 1.4401,
      "step": 152
    },
    {
      "epoch": 0.03362637362637363,
      "grad_norm": 0.26958390231177176,
      "learning_rate": 6.982439517505609e-07,
      "loss": 1.4865,
      "step": 153
    },
    {
      "epoch": 0.033846153846153845,
      "grad_norm": 0.26179721533422456,
      "learning_rate": 6.982209435579393e-07,
      "loss": 1.4573,
      "step": 154
    },
    {
      "epoch": 0.03406593406593406,
      "grad_norm": 0.26746664471911324,
      "learning_rate": 6.981977860419664e-07,
      "loss": 1.47,
      "step": 155
    },
    {
      "epoch": 0.03428571428571429,
      "grad_norm": 0.26825361946569676,
      "learning_rate": 6.981744792136822e-07,
      "loss": 1.5045,
      "step": 156
    },
    {
      "epoch": 0.034505494505494505,
      "grad_norm": 0.4030036122615661,
      "learning_rate": 6.981510230841979e-07,
      "loss": 1.4877,
      "step": 157
    },
    {
      "epoch": 0.03472527472527472,
      "grad_norm": 0.27170584168506157,
      "learning_rate": 6.981274176646958e-07,
      "loss": 1.5494,
      "step": 158
    },
    {
      "epoch": 0.03494505494505495,
      "grad_norm": 0.2672096935421432,
      "learning_rate": 6.981036629664294e-07,
      "loss": 1.4938,
      "step": 159
    },
    {
      "epoch": 0.035164835164835165,
      "grad_norm": 0.2773005433971945,
      "learning_rate": 6.980797590007237e-07,
      "loss": 1.4964,
      "step": 160
    },
    {
      "epoch": 0.03538461538461538,
      "grad_norm": 0.2694855506049262,
      "learning_rate": 6.98055705778974e-07,
      "loss": 1.5027,
      "step": 161
    },
    {
      "epoch": 0.03560439560439561,
      "grad_norm": 0.27266383003600164,
      "learning_rate": 6.980315033126479e-07,
      "loss": 1.5,
      "step": 162
    },
    {
      "epoch": 0.035824175824175825,
      "grad_norm": 0.9420262239251024,
      "learning_rate": 6.980071516132831e-07,
      "loss": 1.4917,
      "step": 163
    },
    {
      "epoch": 0.03604395604395604,
      "grad_norm": 0.26818601052376817,
      "learning_rate": 6.979826506924891e-07,
      "loss": 1.4569,
      "step": 164
    },
    {
      "epoch": 0.03626373626373627,
      "grad_norm": 0.2620913338518106,
      "learning_rate": 6.979580005619464e-07,
      "loss": 1.4557,
      "step": 165
    },
    {
      "epoch": 0.036483516483516484,
      "grad_norm": 0.266182106227132,
      "learning_rate": 6.979332012334064e-07,
      "loss": 1.5161,
      "step": 166
    },
    {
      "epoch": 0.0367032967032967,
      "grad_norm": 0.2662800539649438,
      "learning_rate": 6.97908252718692e-07,
      "loss": 1.4594,
      "step": 167
    },
    {
      "epoch": 0.036923076923076927,
      "grad_norm": 0.26646844653296586,
      "learning_rate": 6.978831550296969e-07,
      "loss": 1.5261,
      "step": 168
    },
    {
      "epoch": 0.037142857142857144,
      "grad_norm": 0.29971802724846996,
      "learning_rate": 6.978579081783861e-07,
      "loss": 1.5052,
      "step": 169
    },
    {
      "epoch": 0.03736263736263736,
      "grad_norm": 0.27368819727195265,
      "learning_rate": 6.978325121767956e-07,
      "loss": 1.5034,
      "step": 170
    },
    {
      "epoch": 0.03758241758241758,
      "grad_norm": 0.258761161700573,
      "learning_rate": 6.978069670370325e-07,
      "loss": 1.5086,
      "step": 171
    },
    {
      "epoch": 0.037802197802197804,
      "grad_norm": 0.2600022072426382,
      "learning_rate": 6.977812727712753e-07,
      "loss": 1.5493,
      "step": 172
    },
    {
      "epoch": 0.03802197802197802,
      "grad_norm": 0.27417700518264077,
      "learning_rate": 6.977554293917731e-07,
      "loss": 1.5198,
      "step": 173
    },
    {
      "epoch": 0.03824175824175824,
      "grad_norm": 0.27524991954024314,
      "learning_rate": 6.977294369108466e-07,
      "loss": 1.4635,
      "step": 174
    },
    {
      "epoch": 0.038461538461538464,
      "grad_norm": 0.28804055695078534,
      "learning_rate": 6.977032953408869e-07,
      "loss": 1.5334,
      "step": 175
    },
    {
      "epoch": 0.03868131868131868,
      "grad_norm": 0.27813313998794814,
      "learning_rate": 6.976770046943571e-07,
      "loss": 1.4716,
      "step": 176
    },
    {
      "epoch": 0.0389010989010989,
      "grad_norm": 0.25675207046610016,
      "learning_rate": 6.976505649837905e-07,
      "loss": 1.4714,
      "step": 177
    },
    {
      "epoch": 0.03912087912087912,
      "grad_norm": 0.2591895815458457,
      "learning_rate": 6.976239762217921e-07,
      "loss": 1.4909,
      "step": 178
    },
    {
      "epoch": 0.03934065934065934,
      "grad_norm": 0.2674694439581417,
      "learning_rate": 6.975972384210375e-07,
      "loss": 1.5039,
      "step": 179
    },
    {
      "epoch": 0.03956043956043956,
      "grad_norm": 0.26704045084258143,
      "learning_rate": 6.975703515942736e-07,
      "loss": 1.4712,
      "step": 180
    },
    {
      "epoch": 0.03978021978021978,
      "grad_norm": 0.2706980099916859,
      "learning_rate": 6.975433157543184e-07,
      "loss": 1.4654,
      "step": 181
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2613703776781457,
      "learning_rate": 6.975161309140605e-07,
      "loss": 1.5339,
      "step": 182
    },
    {
      "epoch": 0.04021978021978022,
      "grad_norm": 0.26132683662049644,
      "learning_rate": 6.974887970864602e-07,
      "loss": 1.4638,
      "step": 183
    },
    {
      "epoch": 0.04043956043956044,
      "grad_norm": 0.31062354073651693,
      "learning_rate": 6.974613142845483e-07,
      "loss": 1.4769,
      "step": 184
    },
    {
      "epoch": 0.04065934065934066,
      "grad_norm": 0.2677559086305453,
      "learning_rate": 6.974336825214269e-07,
      "loss": 1.527,
      "step": 185
    },
    {
      "epoch": 0.04087912087912088,
      "grad_norm": 0.2614383595103056,
      "learning_rate": 6.974059018102691e-07,
      "loss": 1.5394,
      "step": 186
    },
    {
      "epoch": 0.041098901098901096,
      "grad_norm": 0.2803770763369622,
      "learning_rate": 6.973779721643187e-07,
      "loss": 1.4823,
      "step": 187
    },
    {
      "epoch": 0.04131868131868132,
      "grad_norm": 0.27070270266026036,
      "learning_rate": 6.97349893596891e-07,
      "loss": 1.4939,
      "step": 188
    },
    {
      "epoch": 0.04153846153846154,
      "grad_norm": 0.25333287693602247,
      "learning_rate": 6.973216661213718e-07,
      "loss": 1.4866,
      "step": 189
    },
    {
      "epoch": 0.041758241758241756,
      "grad_norm": 0.4573189538131726,
      "learning_rate": 6.972932897512184e-07,
      "loss": 1.4048,
      "step": 190
    },
    {
      "epoch": 0.04197802197802198,
      "grad_norm": 0.3000858279410423,
      "learning_rate": 6.972647644999584e-07,
      "loss": 1.4757,
      "step": 191
    },
    {
      "epoch": 0.0421978021978022,
      "grad_norm": 0.2678269758785627,
      "learning_rate": 6.972360903811911e-07,
      "loss": 1.4553,
      "step": 192
    },
    {
      "epoch": 0.042417582417582415,
      "grad_norm": 0.25922908064411393,
      "learning_rate": 6.972072674085864e-07,
      "loss": 1.463,
      "step": 193
    },
    {
      "epoch": 0.04263736263736264,
      "grad_norm": 0.26416460454952456,
      "learning_rate": 6.971782955958853e-07,
      "loss": 1.498,
      "step": 194
    },
    {
      "epoch": 0.04285714285714286,
      "grad_norm": 0.2636566769664566,
      "learning_rate": 6.971491749568994e-07,
      "loss": 1.4274,
      "step": 195
    },
    {
      "epoch": 0.043076923076923075,
      "grad_norm": 0.2613977552257632,
      "learning_rate": 6.971199055055118e-07,
      "loss": 1.4527,
      "step": 196
    },
    {
      "epoch": 0.0432967032967033,
      "grad_norm": 0.7037865083710388,
      "learning_rate": 6.970904872556761e-07,
      "loss": 1.4665,
      "step": 197
    },
    {
      "epoch": 0.04351648351648352,
      "grad_norm": 0.2939214889537684,
      "learning_rate": 6.970609202214169e-07,
      "loss": 1.5244,
      "step": 198
    },
    {
      "epoch": 0.043736263736263735,
      "grad_norm": 0.26993117023093754,
      "learning_rate": 6.970312044168303e-07,
      "loss": 1.4594,
      "step": 199
    },
    {
      "epoch": 0.04395604395604396,
      "grad_norm": 0.29335910638815493,
      "learning_rate": 6.970013398560824e-07,
      "loss": 1.5345,
      "step": 200
    },
    {
      "epoch": 0.04417582417582418,
      "grad_norm": 0.2673832762745383,
      "learning_rate": 6.969713265534109e-07,
      "loss": 1.5341,
      "step": 201
    },
    {
      "epoch": 0.044395604395604395,
      "grad_norm": 0.33706349568368016,
      "learning_rate": 6.96941164523124e-07,
      "loss": 1.5072,
      "step": 202
    },
    {
      "epoch": 0.04461538461538461,
      "grad_norm": 1.7527104868913561,
      "learning_rate": 6.969108537796012e-07,
      "loss": 1.4623,
      "step": 203
    },
    {
      "epoch": 0.04483516483516484,
      "grad_norm": 0.2758247656438932,
      "learning_rate": 6.968803943372925e-07,
      "loss": 1.5642,
      "step": 204
    },
    {
      "epoch": 0.045054945054945054,
      "grad_norm": 0.2811418878265,
      "learning_rate": 6.968497862107193e-07,
      "loss": 1.4691,
      "step": 205
    },
    {
      "epoch": 0.04527472527472527,
      "grad_norm": 0.2718143932122558,
      "learning_rate": 6.968190294144732e-07,
      "loss": 1.5535,
      "step": 206
    },
    {
      "epoch": 0.0454945054945055,
      "grad_norm": 0.2728985339929895,
      "learning_rate": 6.967881239632172e-07,
      "loss": 1.5283,
      "step": 207
    },
    {
      "epoch": 0.045714285714285714,
      "grad_norm": 0.26533504546058206,
      "learning_rate": 6.96757069871685e-07,
      "loss": 1.4883,
      "step": 208
    },
    {
      "epoch": 0.04593406593406593,
      "grad_norm": 0.2638565050651618,
      "learning_rate": 6.967258671546811e-07,
      "loss": 1.5245,
      "step": 209
    },
    {
      "epoch": 0.046153846153846156,
      "grad_norm": 0.2646371262308997,
      "learning_rate": 6.96694515827081e-07,
      "loss": 1.472,
      "step": 210
    },
    {
      "epoch": 0.046373626373626374,
      "grad_norm": 0.2674406666090845,
      "learning_rate": 6.966630159038311e-07,
      "loss": 1.4214,
      "step": 211
    },
    {
      "epoch": 0.04659340659340659,
      "grad_norm": 0.2739895654283086,
      "learning_rate": 6.966313673999482e-07,
      "loss": 1.5281,
      "step": 212
    },
    {
      "epoch": 0.046813186813186816,
      "grad_norm": 0.2578629141398095,
      "learning_rate": 6.965995703305205e-07,
      "loss": 1.4542,
      "step": 213
    },
    {
      "epoch": 0.047032967032967034,
      "grad_norm": 0.265806832621302,
      "learning_rate": 6.965676247107067e-07,
      "loss": 1.497,
      "step": 214
    },
    {
      "epoch": 0.04725274725274725,
      "grad_norm": 0.2747953041522572,
      "learning_rate": 6.965355305557364e-07,
      "loss": 1.4623,
      "step": 215
    },
    {
      "epoch": 0.047472527472527476,
      "grad_norm": 0.2654747254930102,
      "learning_rate": 6.965032878809099e-07,
      "loss": 1.4393,
      "step": 216
    },
    {
      "epoch": 0.047692307692307694,
      "grad_norm": 0.26140208597905573,
      "learning_rate": 6.964708967015985e-07,
      "loss": 1.5312,
      "step": 217
    },
    {
      "epoch": 0.04791208791208791,
      "grad_norm": 0.30084804983668306,
      "learning_rate": 6.964383570332442e-07,
      "loss": 1.475,
      "step": 218
    },
    {
      "epoch": 0.04813186813186813,
      "grad_norm": 0.3027050045757752,
      "learning_rate": 6.964056688913597e-07,
      "loss": 1.5316,
      "step": 219
    },
    {
      "epoch": 0.04835164835164835,
      "grad_norm": 0.2609356819024029,
      "learning_rate": 6.963728322915288e-07,
      "loss": 1.5129,
      "step": 220
    },
    {
      "epoch": 0.04857142857142857,
      "grad_norm": 0.28897418836738115,
      "learning_rate": 6.963398472494057e-07,
      "loss": 1.5621,
      "step": 221
    },
    {
      "epoch": 0.04879120879120879,
      "grad_norm": 0.25256520476991495,
      "learning_rate": 6.963067137807155e-07,
      "loss": 1.4543,
      "step": 222
    },
    {
      "epoch": 0.04901098901098901,
      "grad_norm": 0.2635781648831903,
      "learning_rate": 6.96273431901254e-07,
      "loss": 1.5291,
      "step": 223
    },
    {
      "epoch": 0.04923076923076923,
      "grad_norm": 0.2823487055962416,
      "learning_rate": 6.962400016268882e-07,
      "loss": 1.4998,
      "step": 224
    },
    {
      "epoch": 0.04945054945054945,
      "grad_norm": 0.26684083629955946,
      "learning_rate": 6.962064229735551e-07,
      "loss": 1.4757,
      "step": 225
    },
    {
      "epoch": 0.04967032967032967,
      "grad_norm": 0.32736400051695624,
      "learning_rate": 6.96172695957263e-07,
      "loss": 1.4646,
      "step": 226
    },
    {
      "epoch": 0.04989010989010989,
      "grad_norm": 0.2673703631683696,
      "learning_rate": 6.961388205940907e-07,
      "loss": 1.445,
      "step": 227
    },
    {
      "epoch": 0.05010989010989011,
      "grad_norm": 2.2958669167730785,
      "learning_rate": 6.961047969001877e-07,
      "loss": 1.5246,
      "step": 228
    },
    {
      "epoch": 0.05032967032967033,
      "grad_norm": 0.2591744019844466,
      "learning_rate": 6.960706248917743e-07,
      "loss": 1.4993,
      "step": 229
    },
    {
      "epoch": 0.05054945054945055,
      "grad_norm": 0.272023775827462,
      "learning_rate": 6.960363045851415e-07,
      "loss": 1.4788,
      "step": 230
    },
    {
      "epoch": 0.05076923076923077,
      "grad_norm": 0.2881378705174016,
      "learning_rate": 6.960018359966512e-07,
      "loss": 1.4825,
      "step": 231
    },
    {
      "epoch": 0.05098901098901099,
      "grad_norm": 0.2871013170100303,
      "learning_rate": 6.959672191427355e-07,
      "loss": 1.4743,
      "step": 232
    },
    {
      "epoch": 0.05120879120879121,
      "grad_norm": 0.29645734733942863,
      "learning_rate": 6.959324540398975e-07,
      "loss": 1.4721,
      "step": 233
    },
    {
      "epoch": 0.05142857142857143,
      "grad_norm": 0.25854948360200714,
      "learning_rate": 6.958975407047111e-07,
      "loss": 1.563,
      "step": 234
    },
    {
      "epoch": 0.051648351648351645,
      "grad_norm": 0.28516775227487257,
      "learning_rate": 6.958624791538206e-07,
      "loss": 1.4549,
      "step": 235
    },
    {
      "epoch": 0.05186813186813187,
      "grad_norm": 0.2715940213280979,
      "learning_rate": 6.95827269403941e-07,
      "loss": 1.4767,
      "step": 236
    },
    {
      "epoch": 0.05208791208791209,
      "grad_norm": 0.25990619084628314,
      "learning_rate": 6.957919114718581e-07,
      "loss": 1.4342,
      "step": 237
    },
    {
      "epoch": 0.052307692307692305,
      "grad_norm": 0.258840154005006,
      "learning_rate": 6.957564053744282e-07,
      "loss": 1.486,
      "step": 238
    },
    {
      "epoch": 0.05252747252747253,
      "grad_norm": 0.2535593709769906,
      "learning_rate": 6.957207511285784e-07,
      "loss": 1.4045,
      "step": 239
    },
    {
      "epoch": 0.05274725274725275,
      "grad_norm": 0.29933148068654547,
      "learning_rate": 6.956849487513061e-07,
      "loss": 1.4696,
      "step": 240
    },
    {
      "epoch": 0.052967032967032965,
      "grad_norm": 0.26986020751770984,
      "learning_rate": 6.9564899825968e-07,
      "loss": 1.4899,
      "step": 241
    },
    {
      "epoch": 0.05318681318681319,
      "grad_norm": 0.2516728317277496,
      "learning_rate": 6.956128996708385e-07,
      "loss": 1.4936,
      "step": 242
    },
    {
      "epoch": 0.05340659340659341,
      "grad_norm": 0.2556011054356573,
      "learning_rate": 6.955766530019911e-07,
      "loss": 1.4764,
      "step": 243
    },
    {
      "epoch": 0.053626373626373625,
      "grad_norm": 0.2660489284065188,
      "learning_rate": 6.955402582704181e-07,
      "loss": 1.5266,
      "step": 244
    },
    {
      "epoch": 0.05384615384615385,
      "grad_norm": 0.2598598971211793,
      "learning_rate": 6.955037154934699e-07,
      "loss": 1.5083,
      "step": 245
    },
    {
      "epoch": 0.05406593406593407,
      "grad_norm": 0.2576200896121239,
      "learning_rate": 6.95467024688568e-07,
      "loss": 1.501,
      "step": 246
    },
    {
      "epoch": 0.054285714285714284,
      "grad_norm": 0.2715207303985658,
      "learning_rate": 6.954301858732039e-07,
      "loss": 1.4701,
      "step": 247
    },
    {
      "epoch": 0.05450549450549451,
      "grad_norm": 0.2547077363611182,
      "learning_rate": 6.9539319906494e-07,
      "loss": 1.4653,
      "step": 248
    },
    {
      "epoch": 0.054725274725274727,
      "grad_norm": 0.26056877427332575,
      "learning_rate": 6.953560642814092e-07,
      "loss": 1.5032,
      "step": 249
    },
    {
      "epoch": 0.054945054945054944,
      "grad_norm": 0.2679576435566933,
      "learning_rate": 6.953187815403152e-07,
      "loss": 1.478,
      "step": 250
    },
    {
      "epoch": 0.05516483516483516,
      "grad_norm": 0.2578705664542964,
      "learning_rate": 6.952813508594318e-07,
      "loss": 1.5082,
      "step": 251
    },
    {
      "epoch": 0.055384615384615386,
      "grad_norm": 0.26029743030106334,
      "learning_rate": 6.952437722566032e-07,
      "loss": 1.5029,
      "step": 252
    },
    {
      "epoch": 0.055604395604395604,
      "grad_norm": 0.26016894646276006,
      "learning_rate": 6.95206045749745e-07,
      "loss": 1.4943,
      "step": 253
    },
    {
      "epoch": 0.05582417582417582,
      "grad_norm": 0.2536811282304101,
      "learning_rate": 6.951681713568424e-07,
      "loss": 1.4562,
      "step": 254
    },
    {
      "epoch": 0.056043956043956046,
      "grad_norm": 0.25971527603024036,
      "learning_rate": 6.951301490959514e-07,
      "loss": 1.4685,
      "step": 255
    },
    {
      "epoch": 0.056263736263736264,
      "grad_norm": 0.26670319176313756,
      "learning_rate": 6.950919789851988e-07,
      "loss": 1.5116,
      "step": 256
    },
    {
      "epoch": 0.05648351648351648,
      "grad_norm": 0.25735075761994464,
      "learning_rate": 6.950536610427815e-07,
      "loss": 1.4561,
      "step": 257
    },
    {
      "epoch": 0.056703296703296706,
      "grad_norm": 0.2736555478779235,
      "learning_rate": 6.950151952869669e-07,
      "loss": 1.5029,
      "step": 258
    },
    {
      "epoch": 0.05692307692307692,
      "grad_norm": 0.254819786458596,
      "learning_rate": 6.94976581736093e-07,
      "loss": 1.5071,
      "step": 259
    },
    {
      "epoch": 0.05714285714285714,
      "grad_norm": 0.2603206938823192,
      "learning_rate": 6.949378204085683e-07,
      "loss": 1.4552,
      "step": 260
    },
    {
      "epoch": 0.057362637362637366,
      "grad_norm": 0.26917748367705185,
      "learning_rate": 6.948989113228717e-07,
      "loss": 1.513,
      "step": 261
    },
    {
      "epoch": 0.05758241758241758,
      "grad_norm": 0.26502851157044544,
      "learning_rate": 6.948598544975524e-07,
      "loss": 1.5196,
      "step": 262
    },
    {
      "epoch": 0.0578021978021978,
      "grad_norm": 0.27055374354698203,
      "learning_rate": 6.948206499512302e-07,
      "loss": 1.5044,
      "step": 263
    },
    {
      "epoch": 0.058021978021978025,
      "grad_norm": 0.3208603270233443,
      "learning_rate": 6.947812977025953e-07,
      "loss": 1.5076,
      "step": 264
    },
    {
      "epoch": 0.05824175824175824,
      "grad_norm": 0.2596677957877133,
      "learning_rate": 6.947417977704084e-07,
      "loss": 1.4997,
      "step": 265
    },
    {
      "epoch": 0.05846153846153846,
      "grad_norm": 0.26397969261767407,
      "learning_rate": 6.947021501735003e-07,
      "loss": 1.5694,
      "step": 266
    },
    {
      "epoch": 0.05868131868131868,
      "grad_norm": 0.2855200095017623,
      "learning_rate": 6.946623549307725e-07,
      "loss": 1.4998,
      "step": 267
    },
    {
      "epoch": 0.0589010989010989,
      "grad_norm": 0.28246859273897623,
      "learning_rate": 6.946224120611967e-07,
      "loss": 1.4881,
      "step": 268
    },
    {
      "epoch": 0.05912087912087912,
      "grad_norm": 0.24907848933117777,
      "learning_rate": 6.945823215838151e-07,
      "loss": 1.4923,
      "step": 269
    },
    {
      "epoch": 0.05934065934065934,
      "grad_norm": 0.27291804363093825,
      "learning_rate": 6.945420835177403e-07,
      "loss": 1.4702,
      "step": 270
    },
    {
      "epoch": 0.05956043956043956,
      "grad_norm": 0.2673109560476995,
      "learning_rate": 6.945016978821552e-07,
      "loss": 1.523,
      "step": 271
    },
    {
      "epoch": 0.05978021978021978,
      "grad_norm": 0.26555807629974576,
      "learning_rate": 6.944611646963128e-07,
      "loss": 1.4903,
      "step": 272
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.26523653033189487,
      "learning_rate": 6.944204839795369e-07,
      "loss": 1.431,
      "step": 273
    },
    {
      "epoch": 0.06021978021978022,
      "grad_norm": 0.26418183818079705,
      "learning_rate": 6.943796557512214e-07,
      "loss": 1.4915,
      "step": 274
    },
    {
      "epoch": 0.06043956043956044,
      "grad_norm": 0.26251672112394603,
      "learning_rate": 6.943386800308305e-07,
      "loss": 1.5133,
      "step": 275
    },
    {
      "epoch": 0.06065934065934066,
      "grad_norm": 0.2601657148483874,
      "learning_rate": 6.942975568378989e-07,
      "loss": 1.4393,
      "step": 276
    },
    {
      "epoch": 0.06087912087912088,
      "grad_norm": 0.25383464049284915,
      "learning_rate": 6.94256286192031e-07,
      "loss": 1.5193,
      "step": 277
    },
    {
      "epoch": 0.0610989010989011,
      "grad_norm": 0.2527707719641187,
      "learning_rate": 6.942148681129026e-07,
      "loss": 1.502,
      "step": 278
    },
    {
      "epoch": 0.06131868131868132,
      "grad_norm": 0.2550257292963019,
      "learning_rate": 6.941733026202587e-07,
      "loss": 1.5577,
      "step": 279
    },
    {
      "epoch": 0.06153846153846154,
      "grad_norm": 0.26800959630930515,
      "learning_rate": 6.941315897339153e-07,
      "loss": 1.4685,
      "step": 280
    },
    {
      "epoch": 0.06175824175824176,
      "grad_norm": 0.2639661087333859,
      "learning_rate": 6.94089729473758e-07,
      "loss": 1.4654,
      "step": 281
    },
    {
      "epoch": 0.06197802197802198,
      "grad_norm": 0.3044655793163802,
      "learning_rate": 6.940477218597434e-07,
      "loss": 1.5145,
      "step": 282
    },
    {
      "epoch": 0.062197802197802195,
      "grad_norm": 0.26644488846843883,
      "learning_rate": 6.940055669118979e-07,
      "loss": 1.4279,
      "step": 283
    },
    {
      "epoch": 0.06241758241758242,
      "grad_norm": 0.2531821057834337,
      "learning_rate": 6.939632646503181e-07,
      "loss": 1.4871,
      "step": 284
    },
    {
      "epoch": 0.06263736263736264,
      "grad_norm": 0.26515730237558094,
      "learning_rate": 6.939208150951713e-07,
      "loss": 1.4509,
      "step": 285
    },
    {
      "epoch": 0.06285714285714286,
      "grad_norm": 0.27598146747073726,
      "learning_rate": 6.938782182666943e-07,
      "loss": 1.4871,
      "step": 286
    },
    {
      "epoch": 0.06307692307692307,
      "grad_norm": 0.2661969469148146,
      "learning_rate": 6.938354741851947e-07,
      "loss": 1.396,
      "step": 287
    },
    {
      "epoch": 0.0632967032967033,
      "grad_norm": 0.29842965989187764,
      "learning_rate": 6.9379258287105e-07,
      "loss": 1.4267,
      "step": 288
    },
    {
      "epoch": 0.06351648351648352,
      "grad_norm": 0.25858035630649373,
      "learning_rate": 6.93749544344708e-07,
      "loss": 1.5154,
      "step": 289
    },
    {
      "epoch": 0.06373626373626373,
      "grad_norm": 0.2593531590218418,
      "learning_rate": 6.937063586266869e-07,
      "loss": 1.4901,
      "step": 290
    },
    {
      "epoch": 0.06395604395604396,
      "grad_norm": 0.25858425487633924,
      "learning_rate": 6.936630257375745e-07,
      "loss": 1.457,
      "step": 291
    },
    {
      "epoch": 0.06417582417582418,
      "grad_norm": 0.40235054361683753,
      "learning_rate": 6.936195456980291e-07,
      "loss": 1.4533,
      "step": 292
    },
    {
      "epoch": 0.06439560439560439,
      "grad_norm": 0.25704692703086185,
      "learning_rate": 6.935759185287796e-07,
      "loss": 1.4939,
      "step": 293
    },
    {
      "epoch": 0.06461538461538462,
      "grad_norm": 0.26084450550432003,
      "learning_rate": 6.935321442506241e-07,
      "loss": 1.4745,
      "step": 294
    },
    {
      "epoch": 0.06483516483516484,
      "grad_norm": 0.3204320723154116,
      "learning_rate": 6.934882228844315e-07,
      "loss": 1.4833,
      "step": 295
    },
    {
      "epoch": 0.06505494505494505,
      "grad_norm": 0.2925379455584677,
      "learning_rate": 6.934441544511407e-07,
      "loss": 1.5301,
      "step": 296
    },
    {
      "epoch": 0.06527472527472528,
      "grad_norm": 0.2830144001441581,
      "learning_rate": 6.933999389717608e-07,
      "loss": 1.4863,
      "step": 297
    },
    {
      "epoch": 0.0654945054945055,
      "grad_norm": 0.2893794244145113,
      "learning_rate": 6.933555764673704e-07,
      "loss": 1.5546,
      "step": 298
    },
    {
      "epoch": 0.06571428571428571,
      "grad_norm": 0.26853198296167635,
      "learning_rate": 6.933110669591191e-07,
      "loss": 1.4265,
      "step": 299
    },
    {
      "epoch": 0.06593406593406594,
      "grad_norm": 0.2579360349174298,
      "learning_rate": 6.932664104682259e-07,
      "loss": 1.4675,
      "step": 300
    },
    {
      "epoch": 0.06615384615384616,
      "grad_norm": 0.2634299933254918,
      "learning_rate": 6.932216070159802e-07,
      "loss": 1.4939,
      "step": 301
    },
    {
      "epoch": 0.06637362637362637,
      "grad_norm": 0.255533756023376,
      "learning_rate": 6.931766566237413e-07,
      "loss": 1.4855,
      "step": 302
    },
    {
      "epoch": 0.0665934065934066,
      "grad_norm": 0.2525279781357927,
      "learning_rate": 6.931315593129387e-07,
      "loss": 1.4807,
      "step": 303
    },
    {
      "epoch": 0.06681318681318682,
      "grad_norm": 0.2606954135874035,
      "learning_rate": 6.930863151050719e-07,
      "loss": 1.4852,
      "step": 304
    },
    {
      "epoch": 0.06703296703296703,
      "grad_norm": 0.6770030951043836,
      "learning_rate": 6.930409240217102e-07,
      "loss": 1.4383,
      "step": 305
    },
    {
      "epoch": 0.06725274725274726,
      "grad_norm": 0.31236654031247846,
      "learning_rate": 6.929953860844933e-07,
      "loss": 1.4198,
      "step": 306
    },
    {
      "epoch": 0.06747252747252747,
      "grad_norm": 0.27722631806332093,
      "learning_rate": 6.929497013151307e-07,
      "loss": 1.5056,
      "step": 307
    },
    {
      "epoch": 0.06769230769230769,
      "grad_norm": 0.2537696257468583,
      "learning_rate": 6.929038697354018e-07,
      "loss": 1.5424,
      "step": 308
    },
    {
      "epoch": 0.06791208791208792,
      "grad_norm": 0.25301268518741477,
      "learning_rate": 6.928578913671562e-07,
      "loss": 1.5102,
      "step": 309
    },
    {
      "epoch": 0.06813186813186813,
      "grad_norm": 0.35098513331568687,
      "learning_rate": 6.928117662323135e-07,
      "loss": 1.4889,
      "step": 310
    },
    {
      "epoch": 0.06835164835164835,
      "grad_norm": 0.6935830624249227,
      "learning_rate": 6.927654943528629e-07,
      "loss": 1.462,
      "step": 311
    },
    {
      "epoch": 0.06857142857142857,
      "grad_norm": 0.2617719858115404,
      "learning_rate": 6.92719075750864e-07,
      "loss": 1.5075,
      "step": 312
    },
    {
      "epoch": 0.06879120879120879,
      "grad_norm": 0.2502886113171237,
      "learning_rate": 6.926725104484462e-07,
      "loss": 1.4976,
      "step": 313
    },
    {
      "epoch": 0.06901098901098901,
      "grad_norm": 0.26401853218329807,
      "learning_rate": 6.926257984678086e-07,
      "loss": 1.5236,
      "step": 314
    },
    {
      "epoch": 0.06923076923076923,
      "grad_norm": 0.26577736446286343,
      "learning_rate": 6.925789398312206e-07,
      "loss": 1.4676,
      "step": 315
    },
    {
      "epoch": 0.06945054945054945,
      "grad_norm": 0.27638635956465635,
      "learning_rate": 6.925319345610214e-07,
      "loss": 1.4941,
      "step": 316
    },
    {
      "epoch": 0.06967032967032967,
      "grad_norm": 0.29321419443219365,
      "learning_rate": 6.9248478267962e-07,
      "loss": 1.5329,
      "step": 317
    },
    {
      "epoch": 0.0698901098901099,
      "grad_norm": 0.26616921944946587,
      "learning_rate": 6.924374842094952e-07,
      "loss": 1.511,
      "step": 318
    },
    {
      "epoch": 0.0701098901098901,
      "grad_norm": 0.3436957907926617,
      "learning_rate": 6.92390039173196e-07,
      "loss": 1.4564,
      "step": 319
    },
    {
      "epoch": 0.07032967032967033,
      "grad_norm": 0.2827992534635701,
      "learning_rate": 6.923424475933411e-07,
      "loss": 1.4817,
      "step": 320
    },
    {
      "epoch": 0.07054945054945055,
      "grad_norm": 0.2553823593207318,
      "learning_rate": 6.922947094926189e-07,
      "loss": 1.5147,
      "step": 321
    },
    {
      "epoch": 0.07076923076923076,
      "grad_norm": 8.134751377702411,
      "learning_rate": 6.922468248937879e-07,
      "loss": 1.4417,
      "step": 322
    },
    {
      "epoch": 0.07098901098901099,
      "grad_norm": 0.26967755651153225,
      "learning_rate": 6.921987938196766e-07,
      "loss": 1.4807,
      "step": 323
    },
    {
      "epoch": 0.07120879120879121,
      "grad_norm": 3.5383722866014304,
      "learning_rate": 6.921506162931828e-07,
      "loss": 1.4948,
      "step": 324
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 0.2775193997301435,
      "learning_rate": 6.921022923372744e-07,
      "loss": 1.5318,
      "step": 325
    },
    {
      "epoch": 0.07164835164835165,
      "grad_norm": 0.2713185922837225,
      "learning_rate": 6.920538219749892e-07,
      "loss": 1.4545,
      "step": 326
    },
    {
      "epoch": 0.07186813186813187,
      "grad_norm": 0.25034860762400624,
      "learning_rate": 6.920052052294348e-07,
      "loss": 1.4889,
      "step": 327
    },
    {
      "epoch": 0.07208791208791208,
      "grad_norm": 0.26622578973377264,
      "learning_rate": 6.919564421237884e-07,
      "loss": 1.5255,
      "step": 328
    },
    {
      "epoch": 0.07230769230769231,
      "grad_norm": 0.2514601292842668,
      "learning_rate": 6.919075326812971e-07,
      "loss": 1.4863,
      "step": 329
    },
    {
      "epoch": 0.07252747252747253,
      "grad_norm": 0.24965703280073318,
      "learning_rate": 6.918584769252778e-07,
      "loss": 1.4945,
      "step": 330
    },
    {
      "epoch": 0.07274725274725274,
      "grad_norm": 0.26038456628413265,
      "learning_rate": 6.918092748791168e-07,
      "loss": 1.5358,
      "step": 331
    },
    {
      "epoch": 0.07296703296703297,
      "grad_norm": 0.30474496747461144,
      "learning_rate": 6.917599265662707e-07,
      "loss": 1.5037,
      "step": 332
    },
    {
      "epoch": 0.0731868131868132,
      "grad_norm": 0.2704338207014198,
      "learning_rate": 6.917104320102656e-07,
      "loss": 1.5216,
      "step": 333
    },
    {
      "epoch": 0.0734065934065934,
      "grad_norm": 0.2724832631325722,
      "learning_rate": 6.916607912346972e-07,
      "loss": 1.5369,
      "step": 334
    },
    {
      "epoch": 0.07362637362637363,
      "grad_norm": 0.2592734429969151,
      "learning_rate": 6.916110042632309e-07,
      "loss": 1.5068,
      "step": 335
    },
    {
      "epoch": 0.07384615384615385,
      "grad_norm": 0.26663545112069037,
      "learning_rate": 6.915610711196021e-07,
      "loss": 1.5046,
      "step": 336
    },
    {
      "epoch": 0.07406593406593406,
      "grad_norm": 0.2577845369442996,
      "learning_rate": 6.915109918276155e-07,
      "loss": 1.4669,
      "step": 337
    },
    {
      "epoch": 0.07428571428571429,
      "grad_norm": 0.26062999277192267,
      "learning_rate": 6.914607664111456e-07,
      "loss": 1.4814,
      "step": 338
    },
    {
      "epoch": 0.0745054945054945,
      "grad_norm": 0.4468615031479458,
      "learning_rate": 6.914103948941367e-07,
      "loss": 1.499,
      "step": 339
    },
    {
      "epoch": 0.07472527472527472,
      "grad_norm": 0.2556666659553541,
      "learning_rate": 6.913598773006028e-07,
      "loss": 1.4797,
      "step": 340
    },
    {
      "epoch": 0.07494505494505495,
      "grad_norm": 0.26674340467264024,
      "learning_rate": 6.913092136546271e-07,
      "loss": 1.4582,
      "step": 341
    },
    {
      "epoch": 0.07516483516483516,
      "grad_norm": 0.2584509725069121,
      "learning_rate": 6.912584039803629e-07,
      "loss": 1.4678,
      "step": 342
    },
    {
      "epoch": 0.07538461538461538,
      "grad_norm": 0.256005850312379,
      "learning_rate": 6.91207448302033e-07,
      "loss": 1.4405,
      "step": 343
    },
    {
      "epoch": 0.07560439560439561,
      "grad_norm": 0.37548276718717116,
      "learning_rate": 6.911563466439296e-07,
      "loss": 1.4787,
      "step": 344
    },
    {
      "epoch": 0.07582417582417582,
      "grad_norm": 0.25131728724393987,
      "learning_rate": 6.911050990304147e-07,
      "loss": 1.4951,
      "step": 345
    },
    {
      "epoch": 0.07604395604395604,
      "grad_norm": 0.28063430218461827,
      "learning_rate": 6.9105370548592e-07,
      "loss": 1.4831,
      "step": 346
    },
    {
      "epoch": 0.07626373626373627,
      "grad_norm": 0.2671512942188966,
      "learning_rate": 6.910021660349463e-07,
      "loss": 1.4849,
      "step": 347
    },
    {
      "epoch": 0.07648351648351648,
      "grad_norm": 0.25460261140041673,
      "learning_rate": 6.909504807020644e-07,
      "loss": 1.4724,
      "step": 348
    },
    {
      "epoch": 0.0767032967032967,
      "grad_norm": 0.2700059778290506,
      "learning_rate": 6.908986495119145e-07,
      "loss": 1.5273,
      "step": 349
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 0.25645752419926465,
      "learning_rate": 6.908466724892063e-07,
      "loss": 1.4982,
      "step": 350
    },
    {
      "epoch": 0.07714285714285714,
      "grad_norm": 0.25959474579877234,
      "learning_rate": 6.907945496587191e-07,
      "loss": 1.4803,
      "step": 351
    },
    {
      "epoch": 0.07736263736263736,
      "grad_norm": 0.2620479009009138,
      "learning_rate": 6.907422810453017e-07,
      "loss": 1.4585,
      "step": 352
    },
    {
      "epoch": 0.07758241758241759,
      "grad_norm": 0.2527696861757479,
      "learning_rate": 6.906898666738724e-07,
      "loss": 1.5057,
      "step": 353
    },
    {
      "epoch": 0.0778021978021978,
      "grad_norm": 0.2657712564207219,
      "learning_rate": 6.906373065694189e-07,
      "loss": 1.5272,
      "step": 354
    },
    {
      "epoch": 0.07802197802197802,
      "grad_norm": 0.2811559522852078,
      "learning_rate": 6.905846007569983e-07,
      "loss": 1.4933,
      "step": 355
    },
    {
      "epoch": 0.07824175824175825,
      "grad_norm": 0.2508390205169112,
      "learning_rate": 6.905317492617375e-07,
      "loss": 1.4854,
      "step": 356
    },
    {
      "epoch": 0.07846153846153846,
      "grad_norm": 0.26744348357494585,
      "learning_rate": 6.904787521088325e-07,
      "loss": 1.5196,
      "step": 357
    },
    {
      "epoch": 0.07868131868131868,
      "grad_norm": 0.24717717010181148,
      "learning_rate": 6.90425609323549e-07,
      "loss": 1.4946,
      "step": 358
    },
    {
      "epoch": 0.0789010989010989,
      "grad_norm": 0.2624486339764091,
      "learning_rate": 6.903723209312222e-07,
      "loss": 1.4666,
      "step": 359
    },
    {
      "epoch": 0.07912087912087912,
      "grad_norm": 0.2643903815888448,
      "learning_rate": 6.903188869572561e-07,
      "loss": 1.459,
      "step": 360
    },
    {
      "epoch": 0.07934065934065934,
      "grad_norm": 0.25849130265367753,
      "learning_rate": 6.902653074271248e-07,
      "loss": 1.4532,
      "step": 361
    },
    {
      "epoch": 0.07956043956043957,
      "grad_norm": 0.2620958518337077,
      "learning_rate": 6.902115823663716e-07,
      "loss": 1.5331,
      "step": 362
    },
    {
      "epoch": 0.07978021978021978,
      "grad_norm": 0.2655735580121207,
      "learning_rate": 6.901577118006088e-07,
      "loss": 1.4818,
      "step": 363
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48161085497749717,
      "learning_rate": 6.901036957555188e-07,
      "loss": 1.4941,
      "step": 364
    },
    {
      "epoch": 0.08021978021978023,
      "grad_norm": 0.25350645445740005,
      "learning_rate": 6.900495342568526e-07,
      "loss": 1.5258,
      "step": 365
    },
    {
      "epoch": 0.08043956043956044,
      "grad_norm": 0.275420280044783,
      "learning_rate": 6.89995227330431e-07,
      "loss": 1.4297,
      "step": 366
    },
    {
      "epoch": 0.08065934065934066,
      "grad_norm": 0.262688338922607,
      "learning_rate": 6.899407750021441e-07,
      "loss": 1.5446,
      "step": 367
    },
    {
      "epoch": 0.08087912087912089,
      "grad_norm": 0.25907908219518205,
      "learning_rate": 6.898861772979509e-07,
      "loss": 1.5029,
      "step": 368
    },
    {
      "epoch": 0.0810989010989011,
      "grad_norm": 0.29498380574337696,
      "learning_rate": 6.898314342438806e-07,
      "loss": 1.5394,
      "step": 369
    },
    {
      "epoch": 0.08131868131868132,
      "grad_norm": 0.26017168929918727,
      "learning_rate": 6.897765458660305e-07,
      "loss": 1.5135,
      "step": 370
    },
    {
      "epoch": 0.08153846153846153,
      "grad_norm": 0.27229468992527034,
      "learning_rate": 6.897215121905683e-07,
      "loss": 1.4793,
      "step": 371
    },
    {
      "epoch": 0.08175824175824176,
      "grad_norm": 0.2560454187610276,
      "learning_rate": 6.896663332437302e-07,
      "loss": 1.4526,
      "step": 372
    },
    {
      "epoch": 0.08197802197802198,
      "grad_norm": 0.24895478307108312,
      "learning_rate": 6.896110090518217e-07,
      "loss": 1.426,
      "step": 373
    },
    {
      "epoch": 0.08219780219780219,
      "grad_norm": 0.2577451016138694,
      "learning_rate": 6.895555396412185e-07,
      "loss": 1.5201,
      "step": 374
    },
    {
      "epoch": 0.08241758241758242,
      "grad_norm": 0.24907178992928014,
      "learning_rate": 6.894999250383641e-07,
      "loss": 1.4061,
      "step": 375
    },
    {
      "epoch": 0.08263736263736264,
      "grad_norm": 0.28771569194362406,
      "learning_rate": 6.894441652697723e-07,
      "loss": 1.5247,
      "step": 376
    },
    {
      "epoch": 0.08285714285714285,
      "grad_norm": 0.24045256690137964,
      "learning_rate": 6.893882603620256e-07,
      "loss": 1.4698,
      "step": 377
    },
    {
      "epoch": 0.08307692307692308,
      "grad_norm": 0.2551056428383154,
      "learning_rate": 6.893322103417756e-07,
      "loss": 1.4939,
      "step": 378
    },
    {
      "epoch": 0.0832967032967033,
      "grad_norm": 0.2570930834107385,
      "learning_rate": 6.892760152357438e-07,
      "loss": 1.4728,
      "step": 379
    },
    {
      "epoch": 0.08351648351648351,
      "grad_norm": 0.2566016081577767,
      "learning_rate": 6.892196750707201e-07,
      "loss": 1.4977,
      "step": 380
    },
    {
      "epoch": 0.08373626373626374,
      "grad_norm": 0.25658848478050506,
      "learning_rate": 6.891631898735637e-07,
      "loss": 1.4985,
      "step": 381
    },
    {
      "epoch": 0.08395604395604396,
      "grad_norm": 0.2706460825991662,
      "learning_rate": 6.891065596712032e-07,
      "loss": 1.4121,
      "step": 382
    },
    {
      "epoch": 0.08417582417582417,
      "grad_norm": 0.3365445903118446,
      "learning_rate": 6.890497844906363e-07,
      "loss": 1.4926,
      "step": 383
    },
    {
      "epoch": 0.0843956043956044,
      "grad_norm": 0.2652784557456132,
      "learning_rate": 6.889928643589294e-07,
      "loss": 1.5019,
      "step": 384
    },
    {
      "epoch": 0.08461538461538462,
      "grad_norm": 0.268707146755482,
      "learning_rate": 6.889357993032188e-07,
      "loss": 1.5447,
      "step": 385
    },
    {
      "epoch": 0.08483516483516483,
      "grad_norm": 0.28103291812562514,
      "learning_rate": 6.888785893507087e-07,
      "loss": 1.5017,
      "step": 386
    },
    {
      "epoch": 0.08505494505494506,
      "grad_norm": 0.25304645911943247,
      "learning_rate": 6.888212345286738e-07,
      "loss": 1.4316,
      "step": 387
    },
    {
      "epoch": 0.08527472527472528,
      "grad_norm": 0.25568766747949867,
      "learning_rate": 6.887637348644568e-07,
      "loss": 1.4708,
      "step": 388
    },
    {
      "epoch": 0.08549450549450549,
      "grad_norm": 0.27101945950458134,
      "learning_rate": 6.887060903854696e-07,
      "loss": 1.5363,
      "step": 389
    },
    {
      "epoch": 0.08571428571428572,
      "grad_norm": 0.24996110145343797,
      "learning_rate": 6.886483011191937e-07,
      "loss": 1.4691,
      "step": 390
    },
    {
      "epoch": 0.08593406593406594,
      "grad_norm": 0.2559733439879277,
      "learning_rate": 6.885903670931792e-07,
      "loss": 1.4421,
      "step": 391
    },
    {
      "epoch": 0.08615384615384615,
      "grad_norm": 0.24899329109658566,
      "learning_rate": 6.885322883350451e-07,
      "loss": 1.4893,
      "step": 392
    },
    {
      "epoch": 0.08637362637362637,
      "grad_norm": 0.2549324828735496,
      "learning_rate": 6.884740648724797e-07,
      "loss": 1.4932,
      "step": 393
    },
    {
      "epoch": 0.0865934065934066,
      "grad_norm": 0.25896256441327564,
      "learning_rate": 6.8841569673324e-07,
      "loss": 1.5253,
      "step": 394
    },
    {
      "epoch": 0.08681318681318681,
      "grad_norm": 0.25886906458445796,
      "learning_rate": 6.883571839451523e-07,
      "loss": 1.4753,
      "step": 395
    },
    {
      "epoch": 0.08703296703296703,
      "grad_norm": 0.25318443906922317,
      "learning_rate": 6.882985265361118e-07,
      "loss": 1.448,
      "step": 396
    },
    {
      "epoch": 0.08725274725274726,
      "grad_norm": 0.2512113574590466,
      "learning_rate": 6.882397245340823e-07,
      "loss": 1.4394,
      "step": 397
    },
    {
      "epoch": 0.08747252747252747,
      "grad_norm": 0.26137115546154455,
      "learning_rate": 6.881807779670967e-07,
      "loss": 1.4473,
      "step": 398
    },
    {
      "epoch": 0.0876923076923077,
      "grad_norm": 0.2593229842385263,
      "learning_rate": 6.88121686863257e-07,
      "loss": 1.4174,
      "step": 399
    },
    {
      "epoch": 0.08791208791208792,
      "grad_norm": 0.25189941519194164,
      "learning_rate": 6.880624512507341e-07,
      "loss": 1.5331,
      "step": 400
    },
    {
      "epoch": 0.08813186813186813,
      "grad_norm": 0.26325410031477786,
      "learning_rate": 6.880030711577677e-07,
      "loss": 1.5015,
      "step": 401
    },
    {
      "epoch": 0.08835164835164835,
      "grad_norm": 0.2649848528360352,
      "learning_rate": 6.87943546612666e-07,
      "loss": 1.4982,
      "step": 402
    },
    {
      "epoch": 0.08857142857142856,
      "grad_norm": 0.2504726168721806,
      "learning_rate": 6.87883877643807e-07,
      "loss": 1.4844,
      "step": 403
    },
    {
      "epoch": 0.08879120879120879,
      "grad_norm": 0.28735708709213764,
      "learning_rate": 6.878240642796363e-07,
      "loss": 1.4894,
      "step": 404
    },
    {
      "epoch": 0.08901098901098901,
      "grad_norm": 0.2747049804656409,
      "learning_rate": 6.877641065486695e-07,
      "loss": 1.431,
      "step": 405
    },
    {
      "epoch": 0.08923076923076922,
      "grad_norm": 0.2478607752612214,
      "learning_rate": 6.877040044794905e-07,
      "loss": 1.4954,
      "step": 406
    },
    {
      "epoch": 0.08945054945054945,
      "grad_norm": 0.2624854251600358,
      "learning_rate": 6.87643758100752e-07,
      "loss": 1.47,
      "step": 407
    },
    {
      "epoch": 0.08967032967032967,
      "grad_norm": 0.24467138857512993,
      "learning_rate": 6.875833674411754e-07,
      "loss": 1.4333,
      "step": 408
    },
    {
      "epoch": 0.08989010989010988,
      "grad_norm": 0.3695908168101729,
      "learning_rate": 6.875228325295512e-07,
      "loss": 1.5326,
      "step": 409
    },
    {
      "epoch": 0.09010989010989011,
      "grad_norm": 0.3967591539462583,
      "learning_rate": 6.874621533947384e-07,
      "loss": 1.4744,
      "step": 410
    },
    {
      "epoch": 0.09032967032967033,
      "grad_norm": 0.29037431096431854,
      "learning_rate": 6.874013300656651e-07,
      "loss": 1.4998,
      "step": 411
    },
    {
      "epoch": 0.09054945054945054,
      "grad_norm": 0.26813923373713594,
      "learning_rate": 6.873403625713275e-07,
      "loss": 1.4229,
      "step": 412
    },
    {
      "epoch": 0.09076923076923077,
      "grad_norm": 0.25477108101131696,
      "learning_rate": 6.872792509407913e-07,
      "loss": 1.4572,
      "step": 413
    },
    {
      "epoch": 0.090989010989011,
      "grad_norm": 0.2533629057160468,
      "learning_rate": 6.872179952031905e-07,
      "loss": 1.5205,
      "step": 414
    },
    {
      "epoch": 0.0912087912087912,
      "grad_norm": 0.25823996012280015,
      "learning_rate": 6.871565953877276e-07,
      "loss": 1.5479,
      "step": 415
    },
    {
      "epoch": 0.09142857142857143,
      "grad_norm": 0.2502767404385213,
      "learning_rate": 6.870950515236742e-07,
      "loss": 1.4669,
      "step": 416
    },
    {
      "epoch": 0.09164835164835165,
      "grad_norm": 0.25614080286852986,
      "learning_rate": 6.870333636403703e-07,
      "loss": 1.4709,
      "step": 417
    },
    {
      "epoch": 0.09186813186813186,
      "grad_norm": 0.25423154440920465,
      "learning_rate": 6.869715317672249e-07,
      "loss": 1.4614,
      "step": 418
    },
    {
      "epoch": 0.09208791208791209,
      "grad_norm": 0.2820748505144496,
      "learning_rate": 6.869095559337152e-07,
      "loss": 1.4711,
      "step": 419
    },
    {
      "epoch": 0.09230769230769231,
      "grad_norm": 0.38836643198774956,
      "learning_rate": 6.868474361693873e-07,
      "loss": 1.4723,
      "step": 420
    },
    {
      "epoch": 0.09252747252747252,
      "grad_norm": 0.2622776612555053,
      "learning_rate": 6.867851725038559e-07,
      "loss": 1.4931,
      "step": 421
    },
    {
      "epoch": 0.09274725274725275,
      "grad_norm": 0.2638414257200676,
      "learning_rate": 6.867227649668043e-07,
      "loss": 1.4804,
      "step": 422
    },
    {
      "epoch": 0.09296703296703297,
      "grad_norm": 0.2538323772231728,
      "learning_rate": 6.866602135879841e-07,
      "loss": 1.4465,
      "step": 423
    },
    {
      "epoch": 0.09318681318681318,
      "grad_norm": 0.25220892076182044,
      "learning_rate": 6.86597518397216e-07,
      "loss": 1.5155,
      "step": 424
    },
    {
      "epoch": 0.09340659340659341,
      "grad_norm": 0.25877234719644127,
      "learning_rate": 6.865346794243889e-07,
      "loss": 1.4924,
      "step": 425
    },
    {
      "epoch": 0.09362637362637363,
      "grad_norm": 0.254282371919666,
      "learning_rate": 6.864716966994603e-07,
      "loss": 1.5142,
      "step": 426
    },
    {
      "epoch": 0.09384615384615384,
      "grad_norm": 0.2699039585810529,
      "learning_rate": 6.864085702524563e-07,
      "loss": 1.5293,
      "step": 427
    },
    {
      "epoch": 0.09406593406593407,
      "grad_norm": 0.26421931723195685,
      "learning_rate": 6.863453001134715e-07,
      "loss": 1.4517,
      "step": 428
    },
    {
      "epoch": 0.09428571428571429,
      "grad_norm": 0.2642785692878364,
      "learning_rate": 6.862818863126689e-07,
      "loss": 1.4724,
      "step": 429
    },
    {
      "epoch": 0.0945054945054945,
      "grad_norm": 0.2503119645036744,
      "learning_rate": 6.862183288802801e-07,
      "loss": 1.4846,
      "step": 430
    },
    {
      "epoch": 0.09472527472527473,
      "grad_norm": 0.2540401369329786,
      "learning_rate": 6.861546278466052e-07,
      "loss": 1.4668,
      "step": 431
    },
    {
      "epoch": 0.09494505494505495,
      "grad_norm": 0.2524060265141518,
      "learning_rate": 6.860907832420127e-07,
      "loss": 1.4974,
      "step": 432
    },
    {
      "epoch": 0.09516483516483516,
      "grad_norm": 0.2533493015247761,
      "learning_rate": 6.860267950969394e-07,
      "loss": 1.4708,
      "step": 433
    },
    {
      "epoch": 0.09538461538461539,
      "grad_norm": 0.3374431842558201,
      "learning_rate": 6.859626634418909e-07,
      "loss": 1.4815,
      "step": 434
    },
    {
      "epoch": 0.0956043956043956,
      "grad_norm": 0.2581770852497789,
      "learning_rate": 6.858983883074408e-07,
      "loss": 1.4708,
      "step": 435
    },
    {
      "epoch": 0.09582417582417582,
      "grad_norm": 0.2522574673385085,
      "learning_rate": 6.858339697242315e-07,
      "loss": 1.566,
      "step": 436
    },
    {
      "epoch": 0.09604395604395605,
      "grad_norm": 0.42697293913143436,
      "learning_rate": 6.857694077229734e-07,
      "loss": 1.4585,
      "step": 437
    },
    {
      "epoch": 0.09626373626373626,
      "grad_norm": 0.26177314911984273,
      "learning_rate": 6.857047023344456e-07,
      "loss": 1.4889,
      "step": 438
    },
    {
      "epoch": 0.09648351648351648,
      "grad_norm": 0.2610139100609117,
      "learning_rate": 6.856398535894953e-07,
      "loss": 1.4891,
      "step": 439
    },
    {
      "epoch": 0.0967032967032967,
      "grad_norm": 0.2495838633528667,
      "learning_rate": 6.855748615190382e-07,
      "loss": 1.466,
      "step": 440
    },
    {
      "epoch": 0.09692307692307692,
      "grad_norm": 0.24574730034749598,
      "learning_rate": 6.855097261540584e-07,
      "loss": 1.4826,
      "step": 441
    },
    {
      "epoch": 0.09714285714285714,
      "grad_norm": 0.2763109377977699,
      "learning_rate": 6.85444447525608e-07,
      "loss": 1.5334,
      "step": 442
    },
    {
      "epoch": 0.09736263736263737,
      "grad_norm": 0.28745607094795184,
      "learning_rate": 6.853790256648076e-07,
      "loss": 1.4575,
      "step": 443
    },
    {
      "epoch": 0.09758241758241758,
      "grad_norm": 0.24475023289100628,
      "learning_rate": 6.853134606028465e-07,
      "loss": 1.4609,
      "step": 444
    },
    {
      "epoch": 0.0978021978021978,
      "grad_norm": 0.2504850925462208,
      "learning_rate": 6.852477523709813e-07,
      "loss": 1.4629,
      "step": 445
    },
    {
      "epoch": 0.09802197802197803,
      "grad_norm": 0.24049767078730194,
      "learning_rate": 6.851819010005377e-07,
      "loss": 1.4344,
      "step": 446
    },
    {
      "epoch": 0.09824175824175824,
      "grad_norm": 0.25202798311689356,
      "learning_rate": 6.851159065229092e-07,
      "loss": 1.4478,
      "step": 447
    },
    {
      "epoch": 0.09846153846153846,
      "grad_norm": 0.2735054530824979,
      "learning_rate": 6.850497689695579e-07,
      "loss": 1.5428,
      "step": 448
    },
    {
      "epoch": 0.09868131868131869,
      "grad_norm": 0.26669562560985255,
      "learning_rate": 6.849834883720137e-07,
      "loss": 1.5291,
      "step": 449
    },
    {
      "epoch": 0.0989010989010989,
      "grad_norm": 0.28016580215239933,
      "learning_rate": 6.849170647618749e-07,
      "loss": 1.4678,
      "step": 450
    },
    {
      "epoch": 0.09912087912087912,
      "grad_norm": 0.2477907212755894,
      "learning_rate": 6.848504981708079e-07,
      "loss": 1.4914,
      "step": 451
    },
    {
      "epoch": 0.09934065934065935,
      "grad_norm": 0.29689288202333614,
      "learning_rate": 6.847837886305475e-07,
      "loss": 1.4912,
      "step": 452
    },
    {
      "epoch": 0.09956043956043956,
      "grad_norm": 0.25483214112335445,
      "learning_rate": 6.847169361728964e-07,
      "loss": 1.445,
      "step": 453
    },
    {
      "epoch": 0.09978021978021978,
      "grad_norm": 0.2593492089307687,
      "learning_rate": 6.846499408297253e-07,
      "loss": 1.5105,
      "step": 454
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2511983653558936,
      "learning_rate": 6.845828026329734e-07,
      "loss": 1.4672,
      "step": 455
    },
    {
      "epoch": 0.10021978021978022,
      "grad_norm": 0.25016435696472505,
      "learning_rate": 6.845155216146477e-07,
      "loss": 1.5156,
      "step": 456
    },
    {
      "epoch": 0.10043956043956044,
      "grad_norm": 0.2491019652028722,
      "learning_rate": 6.844480978068236e-07,
      "loss": 1.5027,
      "step": 457
    },
    {
      "epoch": 0.10065934065934067,
      "grad_norm": 0.2532570455267634,
      "learning_rate": 6.843805312416443e-07,
      "loss": 1.4765,
      "step": 458
    },
    {
      "epoch": 0.10087912087912088,
      "grad_norm": 0.2513622077483454,
      "learning_rate": 6.84312821951321e-07,
      "loss": 1.4851,
      "step": 459
    },
    {
      "epoch": 0.1010989010989011,
      "grad_norm": 0.299920254513179,
      "learning_rate": 6.842449699681332e-07,
      "loss": 1.4649,
      "step": 460
    },
    {
      "epoch": 0.10131868131868133,
      "grad_norm": 0.25464179536962583,
      "learning_rate": 6.841769753244283e-07,
      "loss": 1.5068,
      "step": 461
    },
    {
      "epoch": 0.10153846153846154,
      "grad_norm": 0.27790285861164804,
      "learning_rate": 6.841088380526217e-07,
      "loss": 1.4857,
      "step": 462
    },
    {
      "epoch": 0.10175824175824176,
      "grad_norm": 0.27302067653462836,
      "learning_rate": 6.840405581851971e-07,
      "loss": 1.5504,
      "step": 463
    },
    {
      "epoch": 0.10197802197802198,
      "grad_norm": 0.27219458588952417,
      "learning_rate": 6.839721357547054e-07,
      "loss": 1.5409,
      "step": 464
    },
    {
      "epoch": 0.1021978021978022,
      "grad_norm": 0.24724735358444033,
      "learning_rate": 6.839035707937662e-07,
      "loss": 1.4748,
      "step": 465
    },
    {
      "epoch": 0.10241758241758242,
      "grad_norm": 0.25664320111975036,
      "learning_rate": 6.838348633350668e-07,
      "loss": 1.4888,
      "step": 466
    },
    {
      "epoch": 0.10263736263736263,
      "grad_norm": 0.27292257984434426,
      "learning_rate": 6.837660134113625e-07,
      "loss": 1.507,
      "step": 467
    },
    {
      "epoch": 0.10285714285714286,
      "grad_norm": 0.25023308630494506,
      "learning_rate": 6.836970210554762e-07,
      "loss": 1.4296,
      "step": 468
    },
    {
      "epoch": 0.10307692307692308,
      "grad_norm": 0.25185551572310205,
      "learning_rate": 6.836278863002994e-07,
      "loss": 1.4775,
      "step": 469
    },
    {
      "epoch": 0.10329670329670329,
      "grad_norm": 0.2536090712788748,
      "learning_rate": 6.835586091787907e-07,
      "loss": 1.4453,
      "step": 470
    },
    {
      "epoch": 0.10351648351648352,
      "grad_norm": 0.2598915530603082,
      "learning_rate": 6.83489189723977e-07,
      "loss": 1.4433,
      "step": 471
    },
    {
      "epoch": 0.10373626373626374,
      "grad_norm": 0.3671535169290183,
      "learning_rate": 6.834196279689531e-07,
      "loss": 1.4908,
      "step": 472
    },
    {
      "epoch": 0.10395604395604395,
      "grad_norm": 0.25483324688785886,
      "learning_rate": 6.833499239468814e-07,
      "loss": 1.5017,
      "step": 473
    },
    {
      "epoch": 0.10417582417582417,
      "grad_norm": 0.29012032430794304,
      "learning_rate": 6.832800776909922e-07,
      "loss": 1.4441,
      "step": 474
    },
    {
      "epoch": 0.1043956043956044,
      "grad_norm": 0.2689171943652517,
      "learning_rate": 6.832100892345839e-07,
      "loss": 1.504,
      "step": 475
    },
    {
      "epoch": 0.10461538461538461,
      "grad_norm": 0.25222277720179376,
      "learning_rate": 6.83139958611022e-07,
      "loss": 1.4755,
      "step": 476
    },
    {
      "epoch": 0.10483516483516483,
      "grad_norm": 0.2669393257393482,
      "learning_rate": 6.830696858537406e-07,
      "loss": 1.4486,
      "step": 477
    },
    {
      "epoch": 0.10505494505494506,
      "grad_norm": 0.28494549510793676,
      "learning_rate": 6.82999270996241e-07,
      "loss": 1.5624,
      "step": 478
    },
    {
      "epoch": 0.10527472527472527,
      "grad_norm": 0.253055576786351,
      "learning_rate": 6.829287140720925e-07,
      "loss": 1.465,
      "step": 479
    },
    {
      "epoch": 0.1054945054945055,
      "grad_norm": 0.2727270214833426,
      "learning_rate": 6.82858015114932e-07,
      "loss": 1.4986,
      "step": 480
    },
    {
      "epoch": 0.10571428571428572,
      "grad_norm": 0.2616853189251014,
      "learning_rate": 6.827871741584641e-07,
      "loss": 1.4361,
      "step": 481
    },
    {
      "epoch": 0.10593406593406593,
      "grad_norm": 0.26911600692056026,
      "learning_rate": 6.827161912364613e-07,
      "loss": 1.4822,
      "step": 482
    },
    {
      "epoch": 0.10615384615384615,
      "grad_norm": 0.2506542803052069,
      "learning_rate": 6.826450663827636e-07,
      "loss": 1.4666,
      "step": 483
    },
    {
      "epoch": 0.10637362637362638,
      "grad_norm": 0.2792852427027902,
      "learning_rate": 6.825737996312786e-07,
      "loss": 1.4945,
      "step": 484
    },
    {
      "epoch": 0.10659340659340659,
      "grad_norm": 0.2485019231447162,
      "learning_rate": 6.825023910159816e-07,
      "loss": 1.4846,
      "step": 485
    },
    {
      "epoch": 0.10681318681318681,
      "grad_norm": 0.2844425014032852,
      "learning_rate": 6.824308405709159e-07,
      "loss": 1.4581,
      "step": 486
    },
    {
      "epoch": 0.10703296703296704,
      "grad_norm": 0.24777904804958625,
      "learning_rate": 6.823591483301917e-07,
      "loss": 1.4519,
      "step": 487
    },
    {
      "epoch": 0.10725274725274725,
      "grad_norm": 0.2449120015806373,
      "learning_rate": 6.822873143279874e-07,
      "loss": 1.434,
      "step": 488
    },
    {
      "epoch": 0.10747252747252747,
      "grad_norm": 0.2651587756968944,
      "learning_rate": 6.822153385985487e-07,
      "loss": 1.4872,
      "step": 489
    },
    {
      "epoch": 0.1076923076923077,
      "grad_norm": 0.2647908048951012,
      "learning_rate": 6.82143221176189e-07,
      "loss": 1.4457,
      "step": 490
    },
    {
      "epoch": 0.10791208791208791,
      "grad_norm": 0.25353637829195075,
      "learning_rate": 6.820709620952893e-07,
      "loss": 1.4286,
      "step": 491
    },
    {
      "epoch": 0.10813186813186813,
      "grad_norm": 0.2700291227118415,
      "learning_rate": 6.819985613902977e-07,
      "loss": 1.5072,
      "step": 492
    },
    {
      "epoch": 0.10835164835164836,
      "grad_norm": 0.2505039082672865,
      "learning_rate": 6.819260190957305e-07,
      "loss": 1.4755,
      "step": 493
    },
    {
      "epoch": 0.10857142857142857,
      "grad_norm": 0.2564507325461924,
      "learning_rate": 6.81853335246171e-07,
      "loss": 1.464,
      "step": 494
    },
    {
      "epoch": 0.1087912087912088,
      "grad_norm": 0.2714965940036086,
      "learning_rate": 6.8178050987627e-07,
      "loss": 1.4287,
      "step": 495
    },
    {
      "epoch": 0.10901098901098902,
      "grad_norm": 0.38581712340759083,
      "learning_rate": 6.817075430207461e-07,
      "loss": 1.5313,
      "step": 496
    },
    {
      "epoch": 0.10923076923076923,
      "grad_norm": 0.26480519030228705,
      "learning_rate": 6.816344347143849e-07,
      "loss": 1.436,
      "step": 497
    },
    {
      "epoch": 0.10945054945054945,
      "grad_norm": 0.25247005640015824,
      "learning_rate": 6.815611849920399e-07,
      "loss": 1.4192,
      "step": 498
    },
    {
      "epoch": 0.10967032967032966,
      "grad_norm": 0.24479858855327907,
      "learning_rate": 6.814877938886317e-07,
      "loss": 1.5108,
      "step": 499
    },
    {
      "epoch": 0.10989010989010989,
      "grad_norm": 0.25245686600030026,
      "learning_rate": 6.814142614391486e-07,
      "loss": 1.4044,
      "step": 500
    },
    {
      "epoch": 0.11010989010989011,
      "grad_norm": 0.2779562368830177,
      "learning_rate": 6.813405876786458e-07,
      "loss": 1.4583,
      "step": 501
    },
    {
      "epoch": 0.11032967032967032,
      "grad_norm": 0.2517484372257996,
      "learning_rate": 6.812667726422462e-07,
      "loss": 1.511,
      "step": 502
    },
    {
      "epoch": 0.11054945054945055,
      "grad_norm": 0.24830708842372914,
      "learning_rate": 6.811928163651402e-07,
      "loss": 1.5059,
      "step": 503
    },
    {
      "epoch": 0.11076923076923077,
      "grad_norm": 0.2596011658819083,
      "learning_rate": 6.811187188825851e-07,
      "loss": 1.4828,
      "step": 504
    },
    {
      "epoch": 0.11098901098901098,
      "grad_norm": 0.25591987109679937,
      "learning_rate": 6.810444802299058e-07,
      "loss": 1.4684,
      "step": 505
    },
    {
      "epoch": 0.11120879120879121,
      "grad_norm": 0.24595512205028225,
      "learning_rate": 6.809701004424947e-07,
      "loss": 1.4691,
      "step": 506
    },
    {
      "epoch": 0.11142857142857143,
      "grad_norm": 0.26389582638107206,
      "learning_rate": 6.808955795558109e-07,
      "loss": 1.4806,
      "step": 507
    },
    {
      "epoch": 0.11164835164835164,
      "grad_norm": 0.24042051551057392,
      "learning_rate": 6.808209176053813e-07,
      "loss": 1.3643,
      "step": 508
    },
    {
      "epoch": 0.11186813186813187,
      "grad_norm": 0.23954361062400112,
      "learning_rate": 6.807461146267999e-07,
      "loss": 1.409,
      "step": 509
    },
    {
      "epoch": 0.11208791208791209,
      "grad_norm": 0.3501248630351167,
      "learning_rate": 6.806711706557276e-07,
      "loss": 1.5127,
      "step": 510
    },
    {
      "epoch": 0.1123076923076923,
      "grad_norm": 0.255586536969081,
      "learning_rate": 6.805960857278933e-07,
      "loss": 1.5027,
      "step": 511
    },
    {
      "epoch": 0.11252747252747253,
      "grad_norm": 0.2604156168407272,
      "learning_rate": 6.805208598790921e-07,
      "loss": 1.4921,
      "step": 512
    },
    {
      "epoch": 0.11274725274725275,
      "grad_norm": 0.25777522687505616,
      "learning_rate": 6.80445493145187e-07,
      "loss": 1.4881,
      "step": 513
    },
    {
      "epoch": 0.11296703296703296,
      "grad_norm": 0.2545150835243321,
      "learning_rate": 6.803699855621081e-07,
      "loss": 1.511,
      "step": 514
    },
    {
      "epoch": 0.11318681318681319,
      "grad_norm": 0.2787260228637967,
      "learning_rate": 6.802943371658523e-07,
      "loss": 1.4887,
      "step": 515
    },
    {
      "epoch": 0.11340659340659341,
      "grad_norm": 0.2783140483999986,
      "learning_rate": 6.802185479924839e-07,
      "loss": 1.5523,
      "step": 516
    },
    {
      "epoch": 0.11362637362637362,
      "grad_norm": 0.2536161073592539,
      "learning_rate": 6.801426180781342e-07,
      "loss": 1.4825,
      "step": 517
    },
    {
      "epoch": 0.11384615384615385,
      "grad_norm": 0.2605049130206773,
      "learning_rate": 6.800665474590016e-07,
      "loss": 1.4176,
      "step": 518
    },
    {
      "epoch": 0.11406593406593407,
      "grad_norm": 0.24106833753743503,
      "learning_rate": 6.799903361713517e-07,
      "loss": 1.4833,
      "step": 519
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 0.25180718354253395,
      "learning_rate": 6.799139842515172e-07,
      "loss": 1.5119,
      "step": 520
    },
    {
      "epoch": 0.1145054945054945,
      "grad_norm": 0.24719275326999504,
      "learning_rate": 6.798374917358974e-07,
      "loss": 1.4587,
      "step": 521
    },
    {
      "epoch": 0.11472527472527473,
      "grad_norm": 0.374620782444536,
      "learning_rate": 6.797608586609593e-07,
      "loss": 1.4872,
      "step": 522
    },
    {
      "epoch": 0.11494505494505494,
      "grad_norm": 0.24043316281503388,
      "learning_rate": 6.796840850632363e-07,
      "loss": 1.3724,
      "step": 523
    },
    {
      "epoch": 0.11516483516483517,
      "grad_norm": 0.24894792679030192,
      "learning_rate": 6.796071709793292e-07,
      "loss": 1.4977,
      "step": 524
    },
    {
      "epoch": 0.11538461538461539,
      "grad_norm": 0.25757890500804637,
      "learning_rate": 6.795301164459057e-07,
      "loss": 1.585,
      "step": 525
    },
    {
      "epoch": 0.1156043956043956,
      "grad_norm": 0.2680353943018703,
      "learning_rate": 6.794529214997e-07,
      "loss": 1.4679,
      "step": 526
    },
    {
      "epoch": 0.11582417582417583,
      "grad_norm": 0.28430578711166316,
      "learning_rate": 6.79375586177514e-07,
      "loss": 1.4856,
      "step": 527
    },
    {
      "epoch": 0.11604395604395605,
      "grad_norm": 0.24729281847640566,
      "learning_rate": 6.792981105162161e-07,
      "loss": 1.4634,
      "step": 528
    },
    {
      "epoch": 0.11626373626373626,
      "grad_norm": 0.26446540265063584,
      "learning_rate": 6.792204945527416e-07,
      "loss": 1.5078,
      "step": 529
    },
    {
      "epoch": 0.11648351648351649,
      "grad_norm": 0.2562857706514639,
      "learning_rate": 6.791427383240926e-07,
      "loss": 1.4887,
      "step": 530
    },
    {
      "epoch": 0.1167032967032967,
      "grad_norm": 0.2699076739620769,
      "learning_rate": 6.790648418673385e-07,
      "loss": 1.5422,
      "step": 531
    },
    {
      "epoch": 0.11692307692307692,
      "grad_norm": 0.24833698988720773,
      "learning_rate": 6.78986805219615e-07,
      "loss": 1.4167,
      "step": 532
    },
    {
      "epoch": 0.11714285714285715,
      "grad_norm": 0.2599219400232677,
      "learning_rate": 6.789086284181249e-07,
      "loss": 1.473,
      "step": 533
    },
    {
      "epoch": 0.11736263736263736,
      "grad_norm": 0.25646038902365914,
      "learning_rate": 6.788303115001379e-07,
      "loss": 1.4754,
      "step": 534
    },
    {
      "epoch": 0.11758241758241758,
      "grad_norm": 0.2570833200962951,
      "learning_rate": 6.787518545029905e-07,
      "loss": 1.5218,
      "step": 535
    },
    {
      "epoch": 0.1178021978021978,
      "grad_norm": 0.24369640759526123,
      "learning_rate": 6.786732574640857e-07,
      "loss": 1.4993,
      "step": 536
    },
    {
      "epoch": 0.11802197802197802,
      "grad_norm": 0.2996102711505013,
      "learning_rate": 6.785945204208935e-07,
      "loss": 1.4707,
      "step": 537
    },
    {
      "epoch": 0.11824175824175824,
      "grad_norm": 4.836968981601045,
      "learning_rate": 6.785156434109508e-07,
      "loss": 1.4633,
      "step": 538
    },
    {
      "epoch": 0.11846153846153847,
      "grad_norm": 0.2521721845885586,
      "learning_rate": 6.784366264718607e-07,
      "loss": 1.495,
      "step": 539
    },
    {
      "epoch": 0.11868131868131868,
      "grad_norm": 0.24759360173294137,
      "learning_rate": 6.783574696412935e-07,
      "loss": 1.4902,
      "step": 540
    },
    {
      "epoch": 0.1189010989010989,
      "grad_norm": 0.2506082057658365,
      "learning_rate": 6.782781729569861e-07,
      "loss": 1.4489,
      "step": 541
    },
    {
      "epoch": 0.11912087912087913,
      "grad_norm": 0.247922029109496,
      "learning_rate": 6.781987364567418e-07,
      "loss": 1.5067,
      "step": 542
    },
    {
      "epoch": 0.11934065934065934,
      "grad_norm": 0.2596877891802961,
      "learning_rate": 6.781191601784308e-07,
      "loss": 1.5452,
      "step": 543
    },
    {
      "epoch": 0.11956043956043956,
      "grad_norm": 0.26471453771120845,
      "learning_rate": 6.780394441599901e-07,
      "loss": 1.4403,
      "step": 544
    },
    {
      "epoch": 0.11978021978021978,
      "grad_norm": 0.26119129531553703,
      "learning_rate": 6.779595884394228e-07,
      "loss": 1.4563,
      "step": 545
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.24934977562527547,
      "learning_rate": 6.778795930547992e-07,
      "loss": 1.4777,
      "step": 546
    },
    {
      "epoch": 0.12021978021978022,
      "grad_norm": 0.32445647857339116,
      "learning_rate": 6.777994580442556e-07,
      "loss": 1.4674,
      "step": 547
    },
    {
      "epoch": 0.12043956043956044,
      "grad_norm": 0.2735917636144104,
      "learning_rate": 6.777191834459955e-07,
      "loss": 1.4359,
      "step": 548
    },
    {
      "epoch": 0.12065934065934066,
      "grad_norm": 0.25710108499917056,
      "learning_rate": 6.776387692982883e-07,
      "loss": 1.4893,
      "step": 549
    },
    {
      "epoch": 0.12087912087912088,
      "grad_norm": 0.2568323094996253,
      "learning_rate": 6.775582156394702e-07,
      "loss": 1.4989,
      "step": 550
    },
    {
      "epoch": 0.1210989010989011,
      "grad_norm": 0.2503252992520465,
      "learning_rate": 6.774775225079441e-07,
      "loss": 1.5192,
      "step": 551
    },
    {
      "epoch": 0.12131868131868132,
      "grad_norm": 0.25233039820591163,
      "learning_rate": 6.773966899421793e-07,
      "loss": 1.4429,
      "step": 552
    },
    {
      "epoch": 0.12153846153846154,
      "grad_norm": 0.26672210036338045,
      "learning_rate": 6.773157179807115e-07,
      "loss": 1.4522,
      "step": 553
    },
    {
      "epoch": 0.12175824175824176,
      "grad_norm": 0.27208276025497646,
      "learning_rate": 6.772346066621426e-07,
      "loss": 1.4366,
      "step": 554
    },
    {
      "epoch": 0.12197802197802197,
      "grad_norm": 0.2494064492885206,
      "learning_rate": 6.771533560251415e-07,
      "loss": 1.5145,
      "step": 555
    },
    {
      "epoch": 0.1221978021978022,
      "grad_norm": 0.26432423181770315,
      "learning_rate": 6.77071966108443e-07,
      "loss": 1.4245,
      "step": 556
    },
    {
      "epoch": 0.12241758241758242,
      "grad_norm": 0.26234707648890465,
      "learning_rate": 6.769904369508486e-07,
      "loss": 1.4391,
      "step": 557
    },
    {
      "epoch": 0.12263736263736263,
      "grad_norm": 0.26429009678438037,
      "learning_rate": 6.76908768591226e-07,
      "loss": 1.4905,
      "step": 558
    },
    {
      "epoch": 0.12285714285714286,
      "grad_norm": 0.32733980284574415,
      "learning_rate": 6.768269610685095e-07,
      "loss": 1.5061,
      "step": 559
    },
    {
      "epoch": 0.12307692307692308,
      "grad_norm": 0.2604183503592994,
      "learning_rate": 6.767450144216995e-07,
      "loss": 1.4344,
      "step": 560
    },
    {
      "epoch": 0.1232967032967033,
      "grad_norm": 0.2545024083799796,
      "learning_rate": 6.76662928689863e-07,
      "loss": 1.4102,
      "step": 561
    },
    {
      "epoch": 0.12351648351648352,
      "grad_norm": 0.2529655914581244,
      "learning_rate": 6.765807039121328e-07,
      "loss": 1.4566,
      "step": 562
    },
    {
      "epoch": 0.12373626373626373,
      "grad_norm": 0.25955218773939964,
      "learning_rate": 6.764983401277088e-07,
      "loss": 1.4717,
      "step": 563
    },
    {
      "epoch": 0.12395604395604395,
      "grad_norm": 0.2560272633314233,
      "learning_rate": 6.764158373758561e-07,
      "loss": 1.4886,
      "step": 564
    },
    {
      "epoch": 0.12417582417582418,
      "grad_norm": 0.24216328593221215,
      "learning_rate": 6.76333195695907e-07,
      "loss": 1.4722,
      "step": 565
    },
    {
      "epoch": 0.12439560439560439,
      "grad_norm": 0.27914603169370267,
      "learning_rate": 6.762504151272598e-07,
      "loss": 1.5085,
      "step": 566
    },
    {
      "epoch": 0.12461538461538461,
      "grad_norm": 0.2672713341139715,
      "learning_rate": 6.761674957093786e-07,
      "loss": 1.4944,
      "step": 567
    },
    {
      "epoch": 0.12483516483516484,
      "grad_norm": 0.2601506203345243,
      "learning_rate": 6.760844374817941e-07,
      "loss": 1.5202,
      "step": 568
    },
    {
      "epoch": 0.12505494505494505,
      "grad_norm": 0.2515362004640614,
      "learning_rate": 6.76001240484103e-07,
      "loss": 1.4805,
      "step": 569
    },
    {
      "epoch": 0.12527472527472527,
      "grad_norm": 0.2608813528356316,
      "learning_rate": 6.759179047559682e-07,
      "loss": 1.486,
      "step": 570
    },
    {
      "epoch": 0.1254945054945055,
      "grad_norm": 0.2589014573963635,
      "learning_rate": 6.758344303371189e-07,
      "loss": 1.4856,
      "step": 571
    },
    {
      "epoch": 0.12571428571428572,
      "grad_norm": 0.24920654811289786,
      "learning_rate": 6.757508172673502e-07,
      "loss": 1.4637,
      "step": 572
    },
    {
      "epoch": 0.12593406593406595,
      "grad_norm": 0.2549398474073354,
      "learning_rate": 6.756670655865232e-07,
      "loss": 1.5162,
      "step": 573
    },
    {
      "epoch": 0.12615384615384614,
      "grad_norm": 0.24662331267290583,
      "learning_rate": 6.755831753345654e-07,
      "loss": 1.5235,
      "step": 574
    },
    {
      "epoch": 0.12637362637362637,
      "grad_norm": 0.26311319396900035,
      "learning_rate": 6.754991465514703e-07,
      "loss": 1.4975,
      "step": 575
    },
    {
      "epoch": 0.1265934065934066,
      "grad_norm": 0.27166576107579626,
      "learning_rate": 6.754149792772971e-07,
      "loss": 1.447,
      "step": 576
    },
    {
      "epoch": 0.12681318681318682,
      "grad_norm": 0.2818718536050693,
      "learning_rate": 6.753306735521715e-07,
      "loss": 1.4835,
      "step": 577
    },
    {
      "epoch": 0.12703296703296704,
      "grad_norm": 0.2538419957229311,
      "learning_rate": 6.752462294162848e-07,
      "loss": 1.4437,
      "step": 578
    },
    {
      "epoch": 0.12725274725274724,
      "grad_norm": 0.25876807777467264,
      "learning_rate": 6.751616469098947e-07,
      "loss": 1.4824,
      "step": 579
    },
    {
      "epoch": 0.12747252747252746,
      "grad_norm": 0.2555996009280035,
      "learning_rate": 6.750769260733244e-07,
      "loss": 1.5044,
      "step": 580
    },
    {
      "epoch": 0.1276923076923077,
      "grad_norm": 0.2573739323380264,
      "learning_rate": 6.749920669469635e-07,
      "loss": 1.4234,
      "step": 581
    },
    {
      "epoch": 0.1279120879120879,
      "grad_norm": 0.2852857286268162,
      "learning_rate": 6.749070695712671e-07,
      "loss": 1.4695,
      "step": 582
    },
    {
      "epoch": 0.12813186813186814,
      "grad_norm": 0.254237091514892,
      "learning_rate": 6.748219339867565e-07,
      "loss": 1.5095,
      "step": 583
    },
    {
      "epoch": 0.12835164835164836,
      "grad_norm": 0.25060909727802133,
      "learning_rate": 6.747366602340187e-07,
      "loss": 1.4565,
      "step": 584
    },
    {
      "epoch": 0.12857142857142856,
      "grad_norm": 0.26823526977337603,
      "learning_rate": 6.746512483537069e-07,
      "loss": 1.4477,
      "step": 585
    },
    {
      "epoch": 0.12879120879120878,
      "grad_norm": 0.2575369033243557,
      "learning_rate": 6.745656983865397e-07,
      "loss": 1.5267,
      "step": 586
    },
    {
      "epoch": 0.129010989010989,
      "grad_norm": 0.25692016695936587,
      "learning_rate": 6.74480010373302e-07,
      "loss": 1.6099,
      "step": 587
    },
    {
      "epoch": 0.12923076923076923,
      "grad_norm": 0.24798901677537402,
      "learning_rate": 6.74394184354844e-07,
      "loss": 1.4586,
      "step": 588
    },
    {
      "epoch": 0.12945054945054946,
      "grad_norm": 0.24918359117104827,
      "learning_rate": 6.743082203720821e-07,
      "loss": 1.4707,
      "step": 589
    },
    {
      "epoch": 0.12967032967032968,
      "grad_norm": 0.3580205505412506,
      "learning_rate": 6.742221184659983e-07,
      "loss": 1.4628,
      "step": 590
    },
    {
      "epoch": 0.12989010989010988,
      "grad_norm": 0.2482170366026315,
      "learning_rate": 6.741358786776405e-07,
      "loss": 1.4678,
      "step": 591
    },
    {
      "epoch": 0.1301098901098901,
      "grad_norm": 0.2621541503303289,
      "learning_rate": 6.740495010481218e-07,
      "loss": 1.4722,
      "step": 592
    },
    {
      "epoch": 0.13032967032967033,
      "grad_norm": 0.2609206727854163,
      "learning_rate": 6.739629856186219e-07,
      "loss": 1.4373,
      "step": 593
    },
    {
      "epoch": 0.13054945054945055,
      "grad_norm": 0.24730244786886074,
      "learning_rate": 6.738763324303856e-07,
      "loss": 1.5107,
      "step": 594
    },
    {
      "epoch": 0.13076923076923078,
      "grad_norm": 0.25360527024214685,
      "learning_rate": 6.737895415247234e-07,
      "loss": 1.4751,
      "step": 595
    },
    {
      "epoch": 0.130989010989011,
      "grad_norm": 0.25135232062801177,
      "learning_rate": 6.737026129430117e-07,
      "loss": 1.4944,
      "step": 596
    },
    {
      "epoch": 0.1312087912087912,
      "grad_norm": 0.24739770278957182,
      "learning_rate": 6.736155467266922e-07,
      "loss": 1.4992,
      "step": 597
    },
    {
      "epoch": 0.13142857142857142,
      "grad_norm": 0.248993357785481,
      "learning_rate": 6.735283429172724e-07,
      "loss": 1.4297,
      "step": 598
    },
    {
      "epoch": 0.13164835164835165,
      "grad_norm": 0.2577381150715164,
      "learning_rate": 6.734410015563257e-07,
      "loss": 1.4672,
      "step": 599
    },
    {
      "epoch": 0.13186813186813187,
      "grad_norm": 0.25352728890385245,
      "learning_rate": 6.733535226854904e-07,
      "loss": 1.4939,
      "step": 600
    },
    {
      "epoch": 0.1320879120879121,
      "grad_norm": 0.28282549033235815,
      "learning_rate": 6.732659063464709e-07,
      "loss": 1.4625,
      "step": 601
    },
    {
      "epoch": 0.13230769230769232,
      "grad_norm": 0.25118396488694805,
      "learning_rate": 6.73178152581037e-07,
      "loss": 1.4724,
      "step": 602
    },
    {
      "epoch": 0.13252747252747252,
      "grad_norm": 0.2643496786353885,
      "learning_rate": 6.730902614310239e-07,
      "loss": 1.4778,
      "step": 603
    },
    {
      "epoch": 0.13274725274725274,
      "grad_norm": 0.2750885070990114,
      "learning_rate": 6.730022329383325e-07,
      "loss": 1.4611,
      "step": 604
    },
    {
      "epoch": 0.13296703296703297,
      "grad_norm": 0.2458917515347148,
      "learning_rate": 6.72914067144929e-07,
      "loss": 1.4808,
      "step": 605
    },
    {
      "epoch": 0.1331868131868132,
      "grad_norm": 0.25874215122652794,
      "learning_rate": 6.72825764092845e-07,
      "loss": 1.4708,
      "step": 606
    },
    {
      "epoch": 0.13340659340659342,
      "grad_norm": 0.3584027575706209,
      "learning_rate": 6.727373238241778e-07,
      "loss": 1.5036,
      "step": 607
    },
    {
      "epoch": 0.13362637362637364,
      "grad_norm": 0.24828060838826704,
      "learning_rate": 6.726487463810898e-07,
      "loss": 1.4932,
      "step": 608
    },
    {
      "epoch": 0.13384615384615384,
      "grad_norm": 0.2464631893675043,
      "learning_rate": 6.725600318058091e-07,
      "loss": 1.4668,
      "step": 609
    },
    {
      "epoch": 0.13406593406593406,
      "grad_norm": 0.24528219819876454,
      "learning_rate": 6.724711801406289e-07,
      "loss": 1.4714,
      "step": 610
    },
    {
      "epoch": 0.13428571428571429,
      "grad_norm": 0.2562755203983173,
      "learning_rate": 6.723821914279081e-07,
      "loss": 1.4575,
      "step": 611
    },
    {
      "epoch": 0.1345054945054945,
      "grad_norm": 0.25387672406745615,
      "learning_rate": 6.722930657100705e-07,
      "loss": 1.4835,
      "step": 612
    },
    {
      "epoch": 0.13472527472527474,
      "grad_norm": 0.2345831827522563,
      "learning_rate": 6.722038030296056e-07,
      "loss": 1.4116,
      "step": 613
    },
    {
      "epoch": 0.13494505494505493,
      "grad_norm": 0.2495181884989296,
      "learning_rate": 6.721144034290679e-07,
      "loss": 1.4706,
      "step": 614
    },
    {
      "epoch": 0.13516483516483516,
      "grad_norm": 0.25425273610397525,
      "learning_rate": 6.720248669510775e-07,
      "loss": 1.4549,
      "step": 615
    },
    {
      "epoch": 0.13538461538461538,
      "grad_norm": 0.25143139259916875,
      "learning_rate": 6.719351936383192e-07,
      "loss": 1.4558,
      "step": 616
    },
    {
      "epoch": 0.1356043956043956,
      "grad_norm": 0.25472415540476356,
      "learning_rate": 6.718453835335438e-07,
      "loss": 1.5004,
      "step": 617
    },
    {
      "epoch": 0.13582417582417583,
      "grad_norm": 0.2622890240277487,
      "learning_rate": 6.717554366795667e-07,
      "loss": 1.4762,
      "step": 618
    },
    {
      "epoch": 0.13604395604395605,
      "grad_norm": 0.24338816027078966,
      "learning_rate": 6.716653531192689e-07,
      "loss": 1.4259,
      "step": 619
    },
    {
      "epoch": 0.13626373626373625,
      "grad_norm": 1.9265731543962334,
      "learning_rate": 6.71575132895596e-07,
      "loss": 1.4509,
      "step": 620
    },
    {
      "epoch": 0.13648351648351648,
      "grad_norm": 0.2442154939987973,
      "learning_rate": 6.714847760515595e-07,
      "loss": 1.5357,
      "step": 621
    },
    {
      "epoch": 0.1367032967032967,
      "grad_norm": 0.25491896578455076,
      "learning_rate": 6.713942826302354e-07,
      "loss": 1.4408,
      "step": 622
    },
    {
      "epoch": 0.13692307692307693,
      "grad_norm": 0.25440094853557665,
      "learning_rate": 6.713036526747652e-07,
      "loss": 1.4512,
      "step": 623
    },
    {
      "epoch": 0.13714285714285715,
      "grad_norm": 0.25297524476493,
      "learning_rate": 6.712128862283555e-07,
      "loss": 1.4955,
      "step": 624
    },
    {
      "epoch": 0.13736263736263737,
      "grad_norm": 0.25034624748662054,
      "learning_rate": 6.711219833342775e-07,
      "loss": 1.5209,
      "step": 625
    },
    {
      "epoch": 0.13758241758241757,
      "grad_norm": 0.30715221340878235,
      "learning_rate": 6.71030944035868e-07,
      "loss": 1.4604,
      "step": 626
    },
    {
      "epoch": 0.1378021978021978,
      "grad_norm": 0.24297220322870838,
      "learning_rate": 6.709397683765286e-07,
      "loss": 1.4935,
      "step": 627
    },
    {
      "epoch": 0.13802197802197802,
      "grad_norm": 0.25807837266666944,
      "learning_rate": 6.708484563997259e-07,
      "loss": 1.4928,
      "step": 628
    },
    {
      "epoch": 0.13824175824175824,
      "grad_norm": 0.25352319005316637,
      "learning_rate": 6.707570081489913e-07,
      "loss": 1.4607,
      "step": 629
    },
    {
      "epoch": 0.13846153846153847,
      "grad_norm": 0.27527236640863634,
      "learning_rate": 6.706654236679217e-07,
      "loss": 1.4641,
      "step": 630
    },
    {
      "epoch": 0.1386813186813187,
      "grad_norm": 0.30420842682976595,
      "learning_rate": 6.705737030001786e-07,
      "loss": 1.509,
      "step": 631
    },
    {
      "epoch": 0.1389010989010989,
      "grad_norm": 0.2635889063342168,
      "learning_rate": 6.704818461894882e-07,
      "loss": 1.5144,
      "step": 632
    },
    {
      "epoch": 0.13912087912087912,
      "grad_norm": 0.27409945062160546,
      "learning_rate": 6.703898532796419e-07,
      "loss": 1.5491,
      "step": 633
    },
    {
      "epoch": 0.13934065934065934,
      "grad_norm": 0.2567530630574621,
      "learning_rate": 6.702977243144962e-07,
      "loss": 1.5287,
      "step": 634
    },
    {
      "epoch": 0.13956043956043956,
      "grad_norm": 0.2605688055972312,
      "learning_rate": 6.702054593379717e-07,
      "loss": 1.4475,
      "step": 635
    },
    {
      "epoch": 0.1397802197802198,
      "grad_norm": 0.3781097534457521,
      "learning_rate": 6.701130583940549e-07,
      "loss": 1.4951,
      "step": 636
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2885118240134864,
      "learning_rate": 6.700205215267962e-07,
      "loss": 1.4084,
      "step": 637
    },
    {
      "epoch": 0.1402197802197802,
      "grad_norm": 0.27675262439117465,
      "learning_rate": 6.699278487803111e-07,
      "loss": 1.4321,
      "step": 638
    },
    {
      "epoch": 0.14043956043956043,
      "grad_norm": 0.30865092076123496,
      "learning_rate": 6.698350401987802e-07,
      "loss": 1.4863,
      "step": 639
    },
    {
      "epoch": 0.14065934065934066,
      "grad_norm": 0.2591974374403393,
      "learning_rate": 6.697420958264484e-07,
      "loss": 1.462,
      "step": 640
    },
    {
      "epoch": 0.14087912087912088,
      "grad_norm": 0.26120262286985,
      "learning_rate": 6.696490157076255e-07,
      "loss": 1.5168,
      "step": 641
    },
    {
      "epoch": 0.1410989010989011,
      "grad_norm": 0.25326512305755383,
      "learning_rate": 6.695557998866861e-07,
      "loss": 1.4377,
      "step": 642
    },
    {
      "epoch": 0.1413186813186813,
      "grad_norm": 0.2521425114224541,
      "learning_rate": 6.694624484080696e-07,
      "loss": 1.4993,
      "step": 643
    },
    {
      "epoch": 0.14153846153846153,
      "grad_norm": 0.2535659939459957,
      "learning_rate": 6.693689613162797e-07,
      "loss": 1.4512,
      "step": 644
    },
    {
      "epoch": 0.14175824175824175,
      "grad_norm": 0.2561199850385403,
      "learning_rate": 6.692753386558849e-07,
      "loss": 1.6039,
      "step": 645
    },
    {
      "epoch": 0.14197802197802198,
      "grad_norm": 0.255464856674977,
      "learning_rate": 6.691815804715187e-07,
      "loss": 1.4365,
      "step": 646
    },
    {
      "epoch": 0.1421978021978022,
      "grad_norm": 0.25958850730486166,
      "learning_rate": 6.690876868078785e-07,
      "loss": 1.5047,
      "step": 647
    },
    {
      "epoch": 0.14241758241758243,
      "grad_norm": 0.2500783111746471,
      "learning_rate": 6.689936577097269e-07,
      "loss": 1.4696,
      "step": 648
    },
    {
      "epoch": 0.14263736263736262,
      "grad_norm": 0.25760812371648933,
      "learning_rate": 6.68899493221891e-07,
      "loss": 1.4919,
      "step": 649
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.26538763002665106,
      "learning_rate": 6.68805193389262e-07,
      "loss": 1.4507,
      "step": 650
    },
    {
      "epoch": 0.14307692307692307,
      "grad_norm": 0.26528655371984755,
      "learning_rate": 6.68710758256796e-07,
      "loss": 1.4743,
      "step": 651
    },
    {
      "epoch": 0.1432967032967033,
      "grad_norm": 0.25406466482981116,
      "learning_rate": 6.686161878695137e-07,
      "loss": 1.5057,
      "step": 652
    },
    {
      "epoch": 0.14351648351648352,
      "grad_norm": 0.247850890598129,
      "learning_rate": 6.685214822724999e-07,
      "loss": 1.5243,
      "step": 653
    },
    {
      "epoch": 0.14373626373626375,
      "grad_norm": 0.2750472401396167,
      "learning_rate": 6.684266415109042e-07,
      "loss": 1.476,
      "step": 654
    },
    {
      "epoch": 0.14395604395604394,
      "grad_norm": 0.2578699382533185,
      "learning_rate": 6.683316656299404e-07,
      "loss": 1.5319,
      "step": 655
    },
    {
      "epoch": 0.14417582417582417,
      "grad_norm": 0.2527601291042012,
      "learning_rate": 6.682365546748869e-07,
      "loss": 1.4707,
      "step": 656
    },
    {
      "epoch": 0.1443956043956044,
      "grad_norm": 0.254402693589395,
      "learning_rate": 6.681413086910863e-07,
      "loss": 1.4655,
      "step": 657
    },
    {
      "epoch": 0.14461538461538462,
      "grad_norm": 0.2578331305103979,
      "learning_rate": 6.680459277239458e-07,
      "loss": 1.5281,
      "step": 658
    },
    {
      "epoch": 0.14483516483516484,
      "grad_norm": 0.2459575646762834,
      "learning_rate": 6.679504118189368e-07,
      "loss": 1.4185,
      "step": 659
    },
    {
      "epoch": 0.14505494505494507,
      "grad_norm": 0.24860288790531948,
      "learning_rate": 6.678547610215949e-07,
      "loss": 1.4984,
      "step": 660
    },
    {
      "epoch": 0.14527472527472526,
      "grad_norm": 0.3380853140654683,
      "learning_rate": 6.677589753775204e-07,
      "loss": 1.4717,
      "step": 661
    },
    {
      "epoch": 0.1454945054945055,
      "grad_norm": 0.27199503422431626,
      "learning_rate": 6.676630549323777e-07,
      "loss": 1.4495,
      "step": 662
    },
    {
      "epoch": 0.1457142857142857,
      "grad_norm": 0.25037691212991126,
      "learning_rate": 6.67566999731895e-07,
      "loss": 1.4454,
      "step": 663
    },
    {
      "epoch": 0.14593406593406594,
      "grad_norm": 0.25256060774128214,
      "learning_rate": 6.674708098218657e-07,
      "loss": 1.5024,
      "step": 664
    },
    {
      "epoch": 0.14615384615384616,
      "grad_norm": 0.2538059396923407,
      "learning_rate": 6.673744852481466e-07,
      "loss": 1.4831,
      "step": 665
    },
    {
      "epoch": 0.1463736263736264,
      "grad_norm": 2.7562989218274083,
      "learning_rate": 6.672780260566591e-07,
      "loss": 1.4373,
      "step": 666
    },
    {
      "epoch": 0.14659340659340658,
      "grad_norm": 0.2552285623766303,
      "learning_rate": 6.671814322933885e-07,
      "loss": 1.4442,
      "step": 667
    },
    {
      "epoch": 0.1468131868131868,
      "grad_norm": 0.26199679489610755,
      "learning_rate": 6.670847040043846e-07,
      "loss": 1.4983,
      "step": 668
    },
    {
      "epoch": 0.14703296703296703,
      "grad_norm": 0.25533200148769086,
      "learning_rate": 6.66987841235761e-07,
      "loss": 1.4984,
      "step": 669
    },
    {
      "epoch": 0.14725274725274726,
      "grad_norm": 0.2481720448897788,
      "learning_rate": 6.668908440336957e-07,
      "loss": 1.4718,
      "step": 670
    },
    {
      "epoch": 0.14747252747252748,
      "grad_norm": 0.25046968974867,
      "learning_rate": 6.667937124444304e-07,
      "loss": 1.5561,
      "step": 671
    },
    {
      "epoch": 0.1476923076923077,
      "grad_norm": 0.25262852955013854,
      "learning_rate": 6.666964465142714e-07,
      "loss": 1.4741,
      "step": 672
    },
    {
      "epoch": 0.1479120879120879,
      "grad_norm": 0.2567999837458799,
      "learning_rate": 6.665990462895887e-07,
      "loss": 1.4582,
      "step": 673
    },
    {
      "epoch": 0.14813186813186813,
      "grad_norm": 0.2782472953324953,
      "learning_rate": 6.665015118168163e-07,
      "loss": 1.4744,
      "step": 674
    },
    {
      "epoch": 0.14835164835164835,
      "grad_norm": 0.25883556005102304,
      "learning_rate": 6.664038431424524e-07,
      "loss": 1.4983,
      "step": 675
    },
    {
      "epoch": 0.14857142857142858,
      "grad_norm": 0.4116410723136531,
      "learning_rate": 6.663060403130589e-07,
      "loss": 1.4946,
      "step": 676
    },
    {
      "epoch": 0.1487912087912088,
      "grad_norm": 0.2521008943237067,
      "learning_rate": 6.662081033752619e-07,
      "loss": 1.4725,
      "step": 677
    },
    {
      "epoch": 0.149010989010989,
      "grad_norm": 0.3065605400222554,
      "learning_rate": 6.661100323757514e-07,
      "loss": 1.4687,
      "step": 678
    },
    {
      "epoch": 0.14923076923076922,
      "grad_norm": 0.3131134849596052,
      "learning_rate": 6.660118273612812e-07,
      "loss": 1.5064,
      "step": 679
    },
    {
      "epoch": 0.14945054945054945,
      "grad_norm": 0.24729196101336837,
      "learning_rate": 6.659134883786692e-07,
      "loss": 1.3626,
      "step": 680
    },
    {
      "epoch": 0.14967032967032967,
      "grad_norm": 0.26349791006173723,
      "learning_rate": 6.658150154747968e-07,
      "loss": 1.4969,
      "step": 681
    },
    {
      "epoch": 0.1498901098901099,
      "grad_norm": 0.2494494282378353,
      "learning_rate": 6.657164086966096e-07,
      "loss": 1.4553,
      "step": 682
    },
    {
      "epoch": 0.15010989010989012,
      "grad_norm": 0.25362773053729387,
      "learning_rate": 6.656176680911169e-07,
      "loss": 1.5267,
      "step": 683
    },
    {
      "epoch": 0.15032967032967032,
      "grad_norm": 0.24941857861908245,
      "learning_rate": 6.655187937053918e-07,
      "loss": 1.516,
      "step": 684
    },
    {
      "epoch": 0.15054945054945054,
      "grad_norm": 0.2523332345790184,
      "learning_rate": 6.65419785586571e-07,
      "loss": 1.451,
      "step": 685
    },
    {
      "epoch": 0.15076923076923077,
      "grad_norm": 0.44281741281359693,
      "learning_rate": 6.653206437818554e-07,
      "loss": 1.4983,
      "step": 686
    },
    {
      "epoch": 0.150989010989011,
      "grad_norm": 0.2580148489364453,
      "learning_rate": 6.652213683385091e-07,
      "loss": 1.4659,
      "step": 687
    },
    {
      "epoch": 0.15120879120879122,
      "grad_norm": 0.27388267317813153,
      "learning_rate": 6.651219593038603e-07,
      "loss": 1.4542,
      "step": 688
    },
    {
      "epoch": 0.15142857142857144,
      "grad_norm": 0.27601761546622605,
      "learning_rate": 6.650224167253007e-07,
      "loss": 1.5113,
      "step": 689
    },
    {
      "epoch": 0.15164835164835164,
      "grad_norm": 0.2540050759966249,
      "learning_rate": 6.64922740650286e-07,
      "loss": 1.4595,
      "step": 690
    },
    {
      "epoch": 0.15186813186813186,
      "grad_norm": 0.2536128366598746,
      "learning_rate": 6.648229311263348e-07,
      "loss": 1.461,
      "step": 691
    },
    {
      "epoch": 0.15208791208791209,
      "grad_norm": 1.173833552368354,
      "learning_rate": 6.647229882010302e-07,
      "loss": 1.5149,
      "step": 692
    },
    {
      "epoch": 0.1523076923076923,
      "grad_norm": 0.2524183503518466,
      "learning_rate": 6.646229119220181e-07,
      "loss": 1.463,
      "step": 693
    },
    {
      "epoch": 0.15252747252747254,
      "grad_norm": 0.25023878320877235,
      "learning_rate": 6.645227023370085e-07,
      "loss": 1.4703,
      "step": 694
    },
    {
      "epoch": 0.15274725274725276,
      "grad_norm": 0.3083160506620809,
      "learning_rate": 6.644223594937749e-07,
      "loss": 1.4707,
      "step": 695
    },
    {
      "epoch": 0.15296703296703296,
      "grad_norm": 0.24693806141253166,
      "learning_rate": 6.643218834401542e-07,
      "loss": 1.4681,
      "step": 696
    },
    {
      "epoch": 0.15318681318681318,
      "grad_norm": 0.24889949450072432,
      "learning_rate": 6.642212742240469e-07,
      "loss": 1.4359,
      "step": 697
    },
    {
      "epoch": 0.1534065934065934,
      "grad_norm": 0.252501631871603,
      "learning_rate": 6.641205318934168e-07,
      "loss": 1.463,
      "step": 698
    },
    {
      "epoch": 0.15362637362637363,
      "grad_norm": 0.2599127968116426,
      "learning_rate": 6.640196564962912e-07,
      "loss": 1.4921,
      "step": 699
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 0.26192494393437693,
      "learning_rate": 6.639186480807611e-07,
      "loss": 1.5068,
      "step": 700
    },
    {
      "epoch": 0.15406593406593408,
      "grad_norm": 0.25497868036668103,
      "learning_rate": 6.638175066949808e-07,
      "loss": 1.4696,
      "step": 701
    },
    {
      "epoch": 0.15428571428571428,
      "grad_norm": 0.2460989820027386,
      "learning_rate": 6.637162323871675e-07,
      "loss": 1.4699,
      "step": 702
    },
    {
      "epoch": 0.1545054945054945,
      "grad_norm": 0.24982022505936596,
      "learning_rate": 6.636148252056028e-07,
      "loss": 1.4955,
      "step": 703
    },
    {
      "epoch": 0.15472527472527473,
      "grad_norm": 0.27317865675245767,
      "learning_rate": 6.635132851986306e-07,
      "loss": 1.53,
      "step": 704
    },
    {
      "epoch": 0.15494505494505495,
      "grad_norm": 0.25048302165674363,
      "learning_rate": 6.634116124146587e-07,
      "loss": 1.5136,
      "step": 705
    },
    {
      "epoch": 0.15516483516483517,
      "grad_norm": 0.26192135430394875,
      "learning_rate": 6.633098069021581e-07,
      "loss": 1.4924,
      "step": 706
    },
    {
      "epoch": 0.15538461538461537,
      "grad_norm": 0.281509310709272,
      "learning_rate": 6.632078687096632e-07,
      "loss": 1.4446,
      "step": 707
    },
    {
      "epoch": 0.1556043956043956,
      "grad_norm": 0.2794053992288377,
      "learning_rate": 6.631057978857711e-07,
      "loss": 1.4437,
      "step": 708
    },
    {
      "epoch": 0.15582417582417582,
      "grad_norm": 0.2520586104467259,
      "learning_rate": 6.630035944791427e-07,
      "loss": 1.4722,
      "step": 709
    },
    {
      "epoch": 0.15604395604395604,
      "grad_norm": 0.24110474613088786,
      "learning_rate": 6.62901258538502e-07,
      "loss": 1.472,
      "step": 710
    },
    {
      "epoch": 0.15626373626373627,
      "grad_norm": 0.24988753659951027,
      "learning_rate": 6.627987901126361e-07,
      "loss": 1.4646,
      "step": 711
    },
    {
      "epoch": 0.1564835164835165,
      "grad_norm": 0.2446989179836303,
      "learning_rate": 6.626961892503952e-07,
      "loss": 1.5106,
      "step": 712
    },
    {
      "epoch": 0.1567032967032967,
      "grad_norm": 0.2916677429885161,
      "learning_rate": 6.625934560006927e-07,
      "loss": 1.5192,
      "step": 713
    },
    {
      "epoch": 0.15692307692307692,
      "grad_norm": 0.32650457597663507,
      "learning_rate": 6.624905904125053e-07,
      "loss": 1.3976,
      "step": 714
    },
    {
      "epoch": 0.15714285714285714,
      "grad_norm": 0.2601942019447628,
      "learning_rate": 6.623875925348725e-07,
      "loss": 1.5088,
      "step": 715
    },
    {
      "epoch": 0.15736263736263736,
      "grad_norm": 0.25807366853009023,
      "learning_rate": 6.622844624168969e-07,
      "loss": 1.4225,
      "step": 716
    },
    {
      "epoch": 0.1575824175824176,
      "grad_norm": 0.30106958162934105,
      "learning_rate": 6.621812001077443e-07,
      "loss": 1.467,
      "step": 717
    },
    {
      "epoch": 0.1578021978021978,
      "grad_norm": 0.26586260319881483,
      "learning_rate": 6.620778056566433e-07,
      "loss": 1.5472,
      "step": 718
    },
    {
      "epoch": 0.158021978021978,
      "grad_norm": 0.2964729009625348,
      "learning_rate": 6.61974279112886e-07,
      "loss": 1.5158,
      "step": 719
    },
    {
      "epoch": 0.15824175824175823,
      "grad_norm": 0.45552473289423245,
      "learning_rate": 6.618706205258267e-07,
      "loss": 1.4683,
      "step": 720
    },
    {
      "epoch": 0.15846153846153846,
      "grad_norm": 0.29935869809134935,
      "learning_rate": 6.617668299448836e-07,
      "loss": 1.415,
      "step": 721
    },
    {
      "epoch": 0.15868131868131868,
      "grad_norm": 0.282983557354629,
      "learning_rate": 6.616629074195368e-07,
      "loss": 1.5263,
      "step": 722
    },
    {
      "epoch": 0.1589010989010989,
      "grad_norm": 0.24751519284333773,
      "learning_rate": 6.615588529993298e-07,
      "loss": 1.4847,
      "step": 723
    },
    {
      "epoch": 0.15912087912087913,
      "grad_norm": 0.2567702507643263,
      "learning_rate": 6.614546667338693e-07,
      "loss": 1.4965,
      "step": 724
    },
    {
      "epoch": 0.15934065934065933,
      "grad_norm": 0.2801887945105601,
      "learning_rate": 6.613503486728242e-07,
      "loss": 1.4991,
      "step": 725
    },
    {
      "epoch": 0.15956043956043955,
      "grad_norm": 0.2607227955687198,
      "learning_rate": 6.612458988659267e-07,
      "loss": 1.5175,
      "step": 726
    },
    {
      "epoch": 0.15978021978021978,
      "grad_norm": 0.25962241272880177,
      "learning_rate": 6.611413173629719e-07,
      "loss": 1.4961,
      "step": 727
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.26264978162906727,
      "learning_rate": 6.61036604213817e-07,
      "loss": 1.5129,
      "step": 728
    },
    {
      "epoch": 0.16021978021978023,
      "grad_norm": 0.2547698297401557,
      "learning_rate": 6.609317594683826e-07,
      "loss": 1.5095,
      "step": 729
    },
    {
      "epoch": 0.16043956043956045,
      "grad_norm": 0.2627553310032375,
      "learning_rate": 6.608267831766522e-07,
      "loss": 1.4882,
      "step": 730
    },
    {
      "epoch": 0.16065934065934065,
      "grad_norm": 0.24763594330752617,
      "learning_rate": 6.607216753886711e-07,
      "loss": 1.4772,
      "step": 731
    },
    {
      "epoch": 0.16087912087912087,
      "grad_norm": 0.2661925378097158,
      "learning_rate": 6.606164361545481e-07,
      "loss": 1.4555,
      "step": 732
    },
    {
      "epoch": 0.1610989010989011,
      "grad_norm": 0.2768756160373414,
      "learning_rate": 6.605110655244544e-07,
      "loss": 1.5111,
      "step": 733
    },
    {
      "epoch": 0.16131868131868132,
      "grad_norm": 0.27331559761307034,
      "learning_rate": 6.60405563548624e-07,
      "loss": 1.5633,
      "step": 734
    },
    {
      "epoch": 0.16153846153846155,
      "grad_norm": 0.27237573228019885,
      "learning_rate": 6.602999302773531e-07,
      "loss": 1.4361,
      "step": 735
    },
    {
      "epoch": 0.16175824175824177,
      "grad_norm": 0.28206585637636306,
      "learning_rate": 6.601941657610009e-07,
      "loss": 1.4718,
      "step": 736
    },
    {
      "epoch": 0.16197802197802197,
      "grad_norm": 0.26533934720128444,
      "learning_rate": 6.600882700499892e-07,
      "loss": 1.4645,
      "step": 737
    },
    {
      "epoch": 0.1621978021978022,
      "grad_norm": 0.26097854908852147,
      "learning_rate": 6.599822431948019e-07,
      "loss": 1.5212,
      "step": 738
    },
    {
      "epoch": 0.16241758241758242,
      "grad_norm": 0.25046327644171784,
      "learning_rate": 6.598760852459858e-07,
      "loss": 1.4502,
      "step": 739
    },
    {
      "epoch": 0.16263736263736264,
      "grad_norm": 0.2576758519604668,
      "learning_rate": 6.597697962541503e-07,
      "loss": 1.4647,
      "step": 740
    },
    {
      "epoch": 0.16285714285714287,
      "grad_norm": 0.2683827822946962,
      "learning_rate": 6.596633762699666e-07,
      "loss": 1.4973,
      "step": 741
    },
    {
      "epoch": 0.16307692307692306,
      "grad_norm": 0.2742508338439112,
      "learning_rate": 6.595568253441693e-07,
      "loss": 1.4685,
      "step": 742
    },
    {
      "epoch": 0.1632967032967033,
      "grad_norm": 0.2531576711174202,
      "learning_rate": 6.594501435275547e-07,
      "loss": 1.4534,
      "step": 743
    },
    {
      "epoch": 0.1635164835164835,
      "grad_norm": 0.3885874890573479,
      "learning_rate": 6.593433308709817e-07,
      "loss": 1.5096,
      "step": 744
    },
    {
      "epoch": 0.16373626373626374,
      "grad_norm": 0.2560369856654694,
      "learning_rate": 6.592363874253718e-07,
      "loss": 1.5082,
      "step": 745
    },
    {
      "epoch": 0.16395604395604396,
      "grad_norm": 0.25046593167757836,
      "learning_rate": 6.591293132417087e-07,
      "loss": 1.4847,
      "step": 746
    },
    {
      "epoch": 0.1641758241758242,
      "grad_norm": 0.251520227691982,
      "learning_rate": 6.590221083710381e-07,
      "loss": 1.4886,
      "step": 747
    },
    {
      "epoch": 0.16439560439560438,
      "grad_norm": 0.2557773067570504,
      "learning_rate": 6.589147728644686e-07,
      "loss": 1.4163,
      "step": 748
    },
    {
      "epoch": 0.1646153846153846,
      "grad_norm": 0.25572096390591553,
      "learning_rate": 6.588073067731707e-07,
      "loss": 1.4783,
      "step": 749
    },
    {
      "epoch": 0.16483516483516483,
      "grad_norm": 0.25899297494207973,
      "learning_rate": 6.58699710148377e-07,
      "loss": 1.4259,
      "step": 750
    },
    {
      "epoch": 0.16505494505494506,
      "grad_norm": 0.2526263606274654,
      "learning_rate": 6.585919830413829e-07,
      "loss": 1.494,
      "step": 751
    },
    {
      "epoch": 0.16527472527472528,
      "grad_norm": 0.2565425985518416,
      "learning_rate": 6.584841255035456e-07,
      "loss": 1.4129,
      "step": 752
    },
    {
      "epoch": 0.1654945054945055,
      "grad_norm": 0.24088172078974396,
      "learning_rate": 6.583761375862843e-07,
      "loss": 1.4047,
      "step": 753
    },
    {
      "epoch": 0.1657142857142857,
      "grad_norm": 0.24566766722942013,
      "learning_rate": 6.58268019341081e-07,
      "loss": 1.4837,
      "step": 754
    },
    {
      "epoch": 0.16593406593406593,
      "grad_norm": 0.2843984476327013,
      "learning_rate": 6.581597708194791e-07,
      "loss": 1.4328,
      "step": 755
    },
    {
      "epoch": 0.16615384615384615,
      "grad_norm": 0.24878333751936998,
      "learning_rate": 6.580513920730846e-07,
      "loss": 1.5184,
      "step": 756
    },
    {
      "epoch": 0.16637362637362638,
      "grad_norm": 0.43823398418556275,
      "learning_rate": 6.579428831535656e-07,
      "loss": 1.4559,
      "step": 757
    },
    {
      "epoch": 0.1665934065934066,
      "grad_norm": 0.2709095406799129,
      "learning_rate": 6.578342441126517e-07,
      "loss": 1.4877,
      "step": 758
    },
    {
      "epoch": 0.16681318681318683,
      "grad_norm": 0.27615866398918093,
      "learning_rate": 6.577254750021351e-07,
      "loss": 1.5182,
      "step": 759
    },
    {
      "epoch": 0.16703296703296702,
      "grad_norm": 0.24791442073327075,
      "learning_rate": 6.5761657587387e-07,
      "loss": 1.4094,
      "step": 760
    },
    {
      "epoch": 0.16725274725274725,
      "grad_norm": 0.2733200136570676,
      "learning_rate": 6.575075467797722e-07,
      "loss": 1.4744,
      "step": 761
    },
    {
      "epoch": 0.16747252747252747,
      "grad_norm": 0.4033652973072014,
      "learning_rate": 6.573983877718196e-07,
      "loss": 1.5362,
      "step": 762
    },
    {
      "epoch": 0.1676923076923077,
      "grad_norm": 0.26085757654455294,
      "learning_rate": 6.572890989020523e-07,
      "loss": 1.494,
      "step": 763
    },
    {
      "epoch": 0.16791208791208792,
      "grad_norm": 0.2522065149512857,
      "learning_rate": 6.571796802225721e-07,
      "loss": 1.5137,
      "step": 764
    },
    {
      "epoch": 0.16813186813186815,
      "grad_norm": 0.2514119880043795,
      "learning_rate": 6.570701317855428e-07,
      "loss": 1.4943,
      "step": 765
    },
    {
      "epoch": 0.16835164835164834,
      "grad_norm": 0.36309687601939433,
      "learning_rate": 6.569604536431896e-07,
      "loss": 1.4506,
      "step": 766
    },
    {
      "epoch": 0.16857142857142857,
      "grad_norm": 0.2357495570393573,
      "learning_rate": 6.568506458478003e-07,
      "loss": 1.4307,
      "step": 767
    },
    {
      "epoch": 0.1687912087912088,
      "grad_norm": 0.2528823695798724,
      "learning_rate": 6.567407084517239e-07,
      "loss": 1.4654,
      "step": 768
    },
    {
      "epoch": 0.16901098901098902,
      "grad_norm": 0.2551977277049756,
      "learning_rate": 6.566306415073713e-07,
      "loss": 1.4555,
      "step": 769
    },
    {
      "epoch": 0.16923076923076924,
      "grad_norm": 0.26677009195582646,
      "learning_rate": 6.565204450672155e-07,
      "loss": 1.4539,
      "step": 770
    },
    {
      "epoch": 0.16945054945054944,
      "grad_norm": 0.24832375944081736,
      "learning_rate": 6.564101191837908e-07,
      "loss": 1.4998,
      "step": 771
    },
    {
      "epoch": 0.16967032967032966,
      "grad_norm": 0.2513859529703173,
      "learning_rate": 6.562996639096935e-07,
      "loss": 1.4962,
      "step": 772
    },
    {
      "epoch": 0.16989010989010989,
      "grad_norm": 0.24532097751680812,
      "learning_rate": 6.561890792975814e-07,
      "loss": 1.4674,
      "step": 773
    },
    {
      "epoch": 0.1701098901098901,
      "grad_norm": 0.26123806133019406,
      "learning_rate": 6.560783654001741e-07,
      "loss": 1.5093,
      "step": 774
    },
    {
      "epoch": 0.17032967032967034,
      "grad_norm": 0.24447838413836245,
      "learning_rate": 6.559675222702527e-07,
      "loss": 1.437,
      "step": 775
    },
    {
      "epoch": 0.17054945054945056,
      "grad_norm": 0.25644416310526846,
      "learning_rate": 6.558565499606601e-07,
      "loss": 1.4511,
      "step": 776
    },
    {
      "epoch": 0.17076923076923076,
      "grad_norm": 0.2548628302557498,
      "learning_rate": 6.557454485243006e-07,
      "loss": 1.4297,
      "step": 777
    },
    {
      "epoch": 0.17098901098901098,
      "grad_norm": 0.26264082635266367,
      "learning_rate": 6.556342180141401e-07,
      "loss": 1.4437,
      "step": 778
    },
    {
      "epoch": 0.1712087912087912,
      "grad_norm": 0.2495974706112277,
      "learning_rate": 6.555228584832063e-07,
      "loss": 1.5243,
      "step": 779
    },
    {
      "epoch": 0.17142857142857143,
      "grad_norm": 0.2742468814162477,
      "learning_rate": 6.554113699845878e-07,
      "loss": 1.4201,
      "step": 780
    },
    {
      "epoch": 0.17164835164835165,
      "grad_norm": 0.25386008822047945,
      "learning_rate": 6.552997525714353e-07,
      "loss": 1.5163,
      "step": 781
    },
    {
      "epoch": 0.17186813186813188,
      "grad_norm": 0.23684642564064617,
      "learning_rate": 6.551880062969606e-07,
      "loss": 1.4626,
      "step": 782
    },
    {
      "epoch": 0.17208791208791208,
      "grad_norm": 0.26592975780850897,
      "learning_rate": 6.550761312144372e-07,
      "loss": 1.5183,
      "step": 783
    },
    {
      "epoch": 0.1723076923076923,
      "grad_norm": 0.25814240172040015,
      "learning_rate": 6.549641273771998e-07,
      "loss": 1.5035,
      "step": 784
    },
    {
      "epoch": 0.17252747252747253,
      "grad_norm": 0.24480519686156274,
      "learning_rate": 6.548519948386444e-07,
      "loss": 1.5086,
      "step": 785
    },
    {
      "epoch": 0.17274725274725275,
      "grad_norm": 0.25949934694857035,
      "learning_rate": 6.547397336522287e-07,
      "loss": 1.4474,
      "step": 786
    },
    {
      "epoch": 0.17296703296703297,
      "grad_norm": 0.2777324804802751,
      "learning_rate": 6.546273438714714e-07,
      "loss": 1.51,
      "step": 787
    },
    {
      "epoch": 0.1731868131868132,
      "grad_norm": 0.259494420828199,
      "learning_rate": 6.545148255499525e-07,
      "loss": 1.5005,
      "step": 788
    },
    {
      "epoch": 0.1734065934065934,
      "grad_norm": 0.256246190619224,
      "learning_rate": 6.544021787413136e-07,
      "loss": 1.4771,
      "step": 789
    },
    {
      "epoch": 0.17362637362637362,
      "grad_norm": 0.2586622385799712,
      "learning_rate": 6.542894034992573e-07,
      "loss": 1.551,
      "step": 790
    },
    {
      "epoch": 0.17384615384615384,
      "grad_norm": 0.2581020576686451,
      "learning_rate": 6.541764998775477e-07,
      "loss": 1.4716,
      "step": 791
    },
    {
      "epoch": 0.17406593406593407,
      "grad_norm": 0.2521819900830967,
      "learning_rate": 6.540634679300096e-07,
      "loss": 1.4476,
      "step": 792
    },
    {
      "epoch": 0.1742857142857143,
      "grad_norm": 0.25954292847728805,
      "learning_rate": 6.539503077105293e-07,
      "loss": 1.4297,
      "step": 793
    },
    {
      "epoch": 0.17450549450549452,
      "grad_norm": 0.5217522634747214,
      "learning_rate": 6.538370192730544e-07,
      "loss": 1.4547,
      "step": 794
    },
    {
      "epoch": 0.17472527472527472,
      "grad_norm": 0.24777676470476245,
      "learning_rate": 6.537236026715933e-07,
      "loss": 1.4339,
      "step": 795
    },
    {
      "epoch": 0.17494505494505494,
      "grad_norm": 0.2522563128550129,
      "learning_rate": 6.536100579602157e-07,
      "loss": 1.4643,
      "step": 796
    },
    {
      "epoch": 0.17516483516483516,
      "grad_norm": 0.24515684833735804,
      "learning_rate": 6.534963851930525e-07,
      "loss": 1.4544,
      "step": 797
    },
    {
      "epoch": 0.1753846153846154,
      "grad_norm": 0.2581061962491689,
      "learning_rate": 6.533825844242951e-07,
      "loss": 1.5236,
      "step": 798
    },
    {
      "epoch": 0.1756043956043956,
      "grad_norm": 0.2527242478069631,
      "learning_rate": 6.532686557081965e-07,
      "loss": 1.4979,
      "step": 799
    },
    {
      "epoch": 0.17582417582417584,
      "grad_norm": 0.2524452532048287,
      "learning_rate": 6.531545990990706e-07,
      "loss": 1.4929,
      "step": 800
    },
    {
      "epoch": 0.17604395604395603,
      "grad_norm": 0.2735738920989129,
      "learning_rate": 6.53040414651292e-07,
      "loss": 1.5277,
      "step": 801
    },
    {
      "epoch": 0.17626373626373626,
      "grad_norm": 0.24114702665372068,
      "learning_rate": 6.529261024192965e-07,
      "loss": 1.4867,
      "step": 802
    },
    {
      "epoch": 0.17648351648351648,
      "grad_norm": 0.25020352180657546,
      "learning_rate": 6.528116624575806e-07,
      "loss": 1.4446,
      "step": 803
    },
    {
      "epoch": 0.1767032967032967,
      "grad_norm": 0.2449838740049229,
      "learning_rate": 6.52697094820702e-07,
      "loss": 1.4695,
      "step": 804
    },
    {
      "epoch": 0.17692307692307693,
      "grad_norm": 0.2654891491654527,
      "learning_rate": 6.525823995632791e-07,
      "loss": 1.4923,
      "step": 805
    },
    {
      "epoch": 0.17714285714285713,
      "grad_norm": 0.2765560877843684,
      "learning_rate": 6.52467576739991e-07,
      "loss": 1.5173,
      "step": 806
    },
    {
      "epoch": 0.17736263736263735,
      "grad_norm": 0.24012827113899465,
      "learning_rate": 6.523526264055777e-07,
      "loss": 1.4966,
      "step": 807
    },
    {
      "epoch": 0.17758241758241758,
      "grad_norm": 0.2526242470250657,
      "learning_rate": 6.522375486148402e-07,
      "loss": 1.4905,
      "step": 808
    },
    {
      "epoch": 0.1778021978021978,
      "grad_norm": 0.2860121568632104,
      "learning_rate": 6.521223434226399e-07,
      "loss": 1.4563,
      "step": 809
    },
    {
      "epoch": 0.17802197802197803,
      "grad_norm": 0.259604810086271,
      "learning_rate": 6.520070108838993e-07,
      "loss": 1.4251,
      "step": 810
    },
    {
      "epoch": 0.17824175824175825,
      "grad_norm": 0.24765284014411357,
      "learning_rate": 6.518915510536015e-07,
      "loss": 1.451,
      "step": 811
    },
    {
      "epoch": 0.17846153846153845,
      "grad_norm": 0.4919211505847288,
      "learning_rate": 6.5177596398679e-07,
      "loss": 1.4414,
      "step": 812
    },
    {
      "epoch": 0.17868131868131867,
      "grad_norm": 0.2477839437882422,
      "learning_rate": 6.516602497385695e-07,
      "loss": 1.472,
      "step": 813
    },
    {
      "epoch": 0.1789010989010989,
      "grad_norm": 0.25991095796059227,
      "learning_rate": 6.515444083641046e-07,
      "loss": 1.4993,
      "step": 814
    },
    {
      "epoch": 0.17912087912087912,
      "grad_norm": 0.26451668595525285,
      "learning_rate": 6.514284399186213e-07,
      "loss": 1.4656,
      "step": 815
    },
    {
      "epoch": 0.17934065934065935,
      "grad_norm": 0.25624340531234613,
      "learning_rate": 6.513123444574055e-07,
      "loss": 1.4278,
      "step": 816
    },
    {
      "epoch": 0.17956043956043957,
      "grad_norm": 0.24061256258328917,
      "learning_rate": 6.511961220358043e-07,
      "loss": 1.4867,
      "step": 817
    },
    {
      "epoch": 0.17978021978021977,
      "grad_norm": 0.2448896146864455,
      "learning_rate": 6.510797727092248e-07,
      "loss": 1.4961,
      "step": 818
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.248983494471306,
      "learning_rate": 6.509632965331348e-07,
      "loss": 1.4646,
      "step": 819
    },
    {
      "epoch": 0.18021978021978022,
      "grad_norm": 0.2524894221390734,
      "learning_rate": 6.508466935630625e-07,
      "loss": 1.5152,
      "step": 820
    },
    {
      "epoch": 0.18043956043956044,
      "grad_norm": 0.27034739087854265,
      "learning_rate": 6.507299638545966e-07,
      "loss": 1.5272,
      "step": 821
    },
    {
      "epoch": 0.18065934065934067,
      "grad_norm": 0.24935185571053273,
      "learning_rate": 6.506131074633862e-07,
      "loss": 1.4247,
      "step": 822
    },
    {
      "epoch": 0.1808791208791209,
      "grad_norm": 0.25373264603118834,
      "learning_rate": 6.504961244451409e-07,
      "loss": 1.5418,
      "step": 823
    },
    {
      "epoch": 0.1810989010989011,
      "grad_norm": 0.23834875950733034,
      "learning_rate": 6.503790148556308e-07,
      "loss": 1.4224,
      "step": 824
    },
    {
      "epoch": 0.1813186813186813,
      "grad_norm": 0.25102572617026075,
      "learning_rate": 6.502617787506857e-07,
      "loss": 1.5222,
      "step": 825
    },
    {
      "epoch": 0.18153846153846154,
      "grad_norm": 0.27305593767207875,
      "learning_rate": 6.501444161861964e-07,
      "loss": 1.4816,
      "step": 826
    },
    {
      "epoch": 0.18175824175824176,
      "grad_norm": 0.24391140874490652,
      "learning_rate": 6.500269272181136e-07,
      "loss": 1.4516,
      "step": 827
    },
    {
      "epoch": 0.181978021978022,
      "grad_norm": 0.25132073954371353,
      "learning_rate": 6.499093119024486e-07,
      "loss": 1.4231,
      "step": 828
    },
    {
      "epoch": 0.1821978021978022,
      "grad_norm": 0.26264618591492184,
      "learning_rate": 6.497915702952724e-07,
      "loss": 1.4947,
      "step": 829
    },
    {
      "epoch": 0.1824175824175824,
      "grad_norm": 0.2532582107048698,
      "learning_rate": 6.496737024527169e-07,
      "loss": 1.4384,
      "step": 830
    },
    {
      "epoch": 0.18263736263736263,
      "grad_norm": 0.2555219574762248,
      "learning_rate": 6.495557084309736e-07,
      "loss": 1.4393,
      "step": 831
    },
    {
      "epoch": 0.18285714285714286,
      "grad_norm": 0.24369500766902816,
      "learning_rate": 6.494375882862943e-07,
      "loss": 1.4629,
      "step": 832
    },
    {
      "epoch": 0.18307692307692308,
      "grad_norm": 0.24796294917899964,
      "learning_rate": 6.493193420749912e-07,
      "loss": 1.4816,
      "step": 833
    },
    {
      "epoch": 0.1832967032967033,
      "grad_norm": 0.24614824664988436,
      "learning_rate": 6.492009698534362e-07,
      "loss": 1.5261,
      "step": 834
    },
    {
      "epoch": 0.1835164835164835,
      "grad_norm": 0.25264208108926695,
      "learning_rate": 6.490824716780615e-07,
      "loss": 1.4578,
      "step": 835
    },
    {
      "epoch": 0.18373626373626373,
      "grad_norm": 0.23119773648091366,
      "learning_rate": 6.489638476053595e-07,
      "loss": 1.4633,
      "step": 836
    },
    {
      "epoch": 0.18395604395604395,
      "grad_norm": 0.6424182090535926,
      "learning_rate": 6.488450976918823e-07,
      "loss": 1.4467,
      "step": 837
    },
    {
      "epoch": 0.18417582417582418,
      "grad_norm": 0.24645636620710157,
      "learning_rate": 6.487262219942419e-07,
      "loss": 1.5229,
      "step": 838
    },
    {
      "epoch": 0.1843956043956044,
      "grad_norm": 0.2502673940161327,
      "learning_rate": 6.48607220569111e-07,
      "loss": 1.4655,
      "step": 839
    },
    {
      "epoch": 0.18461538461538463,
      "grad_norm": 0.26559029081889357,
      "learning_rate": 6.484880934732213e-07,
      "loss": 1.4673,
      "step": 840
    },
    {
      "epoch": 0.18483516483516482,
      "grad_norm": 0.24400080120013196,
      "learning_rate": 6.483688407633651e-07,
      "loss": 1.4661,
      "step": 841
    },
    {
      "epoch": 0.18505494505494505,
      "grad_norm": 0.252351566985222,
      "learning_rate": 6.482494624963943e-07,
      "loss": 1.4289,
      "step": 842
    },
    {
      "epoch": 0.18527472527472527,
      "grad_norm": 0.24945902207623144,
      "learning_rate": 6.481299587292204e-07,
      "loss": 1.5083,
      "step": 843
    },
    {
      "epoch": 0.1854945054945055,
      "grad_norm": 0.2547686116647054,
      "learning_rate": 6.480103295188153e-07,
      "loss": 1.4025,
      "step": 844
    },
    {
      "epoch": 0.18571428571428572,
      "grad_norm": 0.24437959120424912,
      "learning_rate": 6.478905749222103e-07,
      "loss": 1.5313,
      "step": 845
    },
    {
      "epoch": 0.18593406593406595,
      "grad_norm": 0.2668805234876379,
      "learning_rate": 6.477706949964966e-07,
      "loss": 1.4854,
      "step": 846
    },
    {
      "epoch": 0.18615384615384614,
      "grad_norm": 0.24669978610275448,
      "learning_rate": 6.476506897988254e-07,
      "loss": 1.4374,
      "step": 847
    },
    {
      "epoch": 0.18637362637362637,
      "grad_norm": 0.2691489026603144,
      "learning_rate": 6.475305593864069e-07,
      "loss": 1.4749,
      "step": 848
    },
    {
      "epoch": 0.1865934065934066,
      "grad_norm": 0.24556468560377964,
      "learning_rate": 6.474103038165117e-07,
      "loss": 1.4873,
      "step": 849
    },
    {
      "epoch": 0.18681318681318682,
      "grad_norm": 0.27248587243588246,
      "learning_rate": 6.472899231464699e-07,
      "loss": 1.5296,
      "step": 850
    },
    {
      "epoch": 0.18703296703296704,
      "grad_norm": 0.24175210505081607,
      "learning_rate": 6.471694174336711e-07,
      "loss": 1.4805,
      "step": 851
    },
    {
      "epoch": 0.18725274725274726,
      "grad_norm": 0.2533379443925673,
      "learning_rate": 6.470487867355646e-07,
      "loss": 1.4816,
      "step": 852
    },
    {
      "epoch": 0.18747252747252746,
      "grad_norm": 0.2638665301079791,
      "learning_rate": 6.469280311096593e-07,
      "loss": 1.4679,
      "step": 853
    },
    {
      "epoch": 0.18769230769230769,
      "grad_norm": 0.2474694831768457,
      "learning_rate": 6.468071506135234e-07,
      "loss": 1.4558,
      "step": 854
    },
    {
      "epoch": 0.1879120879120879,
      "grad_norm": 0.26113569153035554,
      "learning_rate": 6.46686145304785e-07,
      "loss": 1.5425,
      "step": 855
    },
    {
      "epoch": 0.18813186813186814,
      "grad_norm": 0.2528536810109955,
      "learning_rate": 6.465650152411316e-07,
      "loss": 1.5692,
      "step": 856
    },
    {
      "epoch": 0.18835164835164836,
      "grad_norm": 0.24133013640021078,
      "learning_rate": 6.4644376048031e-07,
      "loss": 1.4281,
      "step": 857
    },
    {
      "epoch": 0.18857142857142858,
      "grad_norm": 0.6294298242866595,
      "learning_rate": 6.463223810801268e-07,
      "loss": 1.4473,
      "step": 858
    },
    {
      "epoch": 0.18879120879120878,
      "grad_norm": 0.25023604464808363,
      "learning_rate": 6.462008770984475e-07,
      "loss": 1.4098,
      "step": 859
    },
    {
      "epoch": 0.189010989010989,
      "grad_norm": 0.24797013473975096,
      "learning_rate": 6.460792485931976e-07,
      "loss": 1.4359,
      "step": 860
    },
    {
      "epoch": 0.18923076923076923,
      "grad_norm": 0.25206812258465444,
      "learning_rate": 6.459574956223613e-07,
      "loss": 1.4624,
      "step": 861
    },
    {
      "epoch": 0.18945054945054945,
      "grad_norm": 0.2489549376961613,
      "learning_rate": 6.458356182439828e-07,
      "loss": 1.4618,
      "step": 862
    },
    {
      "epoch": 0.18967032967032968,
      "grad_norm": 0.250557255187581,
      "learning_rate": 6.457136165161652e-07,
      "loss": 1.5019,
      "step": 863
    },
    {
      "epoch": 0.1898901098901099,
      "grad_norm": 0.2510639666261246,
      "learning_rate": 6.455914904970709e-07,
      "loss": 1.4827,
      "step": 864
    },
    {
      "epoch": 0.1901098901098901,
      "grad_norm": 0.41430231161919934,
      "learning_rate": 6.454692402449216e-07,
      "loss": 1.4659,
      "step": 865
    },
    {
      "epoch": 0.19032967032967033,
      "grad_norm": 0.2465675465651625,
      "learning_rate": 6.453468658179984e-07,
      "loss": 1.4819,
      "step": 866
    },
    {
      "epoch": 0.19054945054945055,
      "grad_norm": 0.24963496466847715,
      "learning_rate": 6.452243672746415e-07,
      "loss": 1.5047,
      "step": 867
    },
    {
      "epoch": 0.19076923076923077,
      "grad_norm": 0.2716313031010093,
      "learning_rate": 6.4510174467325e-07,
      "loss": 1.4016,
      "step": 868
    },
    {
      "epoch": 0.190989010989011,
      "grad_norm": 0.24718273880514013,
      "learning_rate": 6.449789980722827e-07,
      "loss": 1.3956,
      "step": 869
    },
    {
      "epoch": 0.1912087912087912,
      "grad_norm": 0.2482414244879754,
      "learning_rate": 6.448561275302569e-07,
      "loss": 1.5006,
      "step": 870
    },
    {
      "epoch": 0.19142857142857142,
      "grad_norm": 0.29032338053231516,
      "learning_rate": 6.447331331057494e-07,
      "loss": 1.4202,
      "step": 871
    },
    {
      "epoch": 0.19164835164835164,
      "grad_norm": 0.263484713757411,
      "learning_rate": 6.446100148573957e-07,
      "loss": 1.4702,
      "step": 872
    },
    {
      "epoch": 0.19186813186813187,
      "grad_norm": 0.318710014023105,
      "learning_rate": 6.44486772843891e-07,
      "loss": 1.4786,
      "step": 873
    },
    {
      "epoch": 0.1920879120879121,
      "grad_norm": 0.26178198978589434,
      "learning_rate": 6.443634071239887e-07,
      "loss": 1.4381,
      "step": 874
    },
    {
      "epoch": 0.19230769230769232,
      "grad_norm": 0.2616784646721311,
      "learning_rate": 6.442399177565017e-07,
      "loss": 1.4749,
      "step": 875
    },
    {
      "epoch": 0.19252747252747252,
      "grad_norm": 0.24694702837666085,
      "learning_rate": 6.441163048003017e-07,
      "loss": 1.4417,
      "step": 876
    },
    {
      "epoch": 0.19274725274725274,
      "grad_norm": 0.24286260970664117,
      "learning_rate": 6.439925683143192e-07,
      "loss": 1.5366,
      "step": 877
    },
    {
      "epoch": 0.19296703296703296,
      "grad_norm": 0.26143931461528386,
      "learning_rate": 6.43868708357544e-07,
      "loss": 1.4638,
      "step": 878
    },
    {
      "epoch": 0.1931868131868132,
      "grad_norm": 0.2549936055391091,
      "learning_rate": 6.43744724989024e-07,
      "loss": 1.488,
      "step": 879
    },
    {
      "epoch": 0.1934065934065934,
      "grad_norm": 0.2457925274541861,
      "learning_rate": 6.436206182678667e-07,
      "loss": 1.4757,
      "step": 880
    },
    {
      "epoch": 0.19362637362637364,
      "grad_norm": 0.25332057771158173,
      "learning_rate": 6.434963882532381e-07,
      "loss": 1.541,
      "step": 881
    },
    {
      "epoch": 0.19384615384615383,
      "grad_norm": 0.2527740783450117,
      "learning_rate": 6.433720350043628e-07,
      "loss": 1.5169,
      "step": 882
    },
    {
      "epoch": 0.19406593406593406,
      "grad_norm": 0.25737766445240007,
      "learning_rate": 6.432475585805246e-07,
      "loss": 1.5362,
      "step": 883
    },
    {
      "epoch": 0.19428571428571428,
      "grad_norm": 0.24722408218766065,
      "learning_rate": 6.431229590410655e-07,
      "loss": 1.4255,
      "step": 884
    },
    {
      "epoch": 0.1945054945054945,
      "grad_norm": 0.24874655782608956,
      "learning_rate": 6.429982364453866e-07,
      "loss": 1.445,
      "step": 885
    },
    {
      "epoch": 0.19472527472527473,
      "grad_norm": 0.9157681858416035,
      "learning_rate": 6.428733908529474e-07,
      "loss": 1.5339,
      "step": 886
    },
    {
      "epoch": 0.19494505494505496,
      "grad_norm": 0.24661783363809,
      "learning_rate": 6.427484223232663e-07,
      "loss": 1.4337,
      "step": 887
    },
    {
      "epoch": 0.19516483516483515,
      "grad_norm": 0.39788485611443236,
      "learning_rate": 6.4262333091592e-07,
      "loss": 1.4669,
      "step": 888
    },
    {
      "epoch": 0.19538461538461538,
      "grad_norm": 0.2547946401026291,
      "learning_rate": 6.424981166905441e-07,
      "loss": 1.4872,
      "step": 889
    },
    {
      "epoch": 0.1956043956043956,
      "grad_norm": 0.23213374586324453,
      "learning_rate": 6.423727797068325e-07,
      "loss": 1.4222,
      "step": 890
    },
    {
      "epoch": 0.19582417582417583,
      "grad_norm": 0.24623234969287922,
      "learning_rate": 6.422473200245377e-07,
      "loss": 1.5105,
      "step": 891
    },
    {
      "epoch": 0.19604395604395605,
      "grad_norm": 0.2353778095539658,
      "learning_rate": 6.421217377034706e-07,
      "loss": 1.452,
      "step": 892
    },
    {
      "epoch": 0.19626373626373628,
      "grad_norm": 0.2560236768429536,
      "learning_rate": 6.419960328035009e-07,
      "loss": 1.4666,
      "step": 893
    },
    {
      "epoch": 0.19648351648351647,
      "grad_norm": 0.5951869201313986,
      "learning_rate": 6.418702053845564e-07,
      "loss": 1.423,
      "step": 894
    },
    {
      "epoch": 0.1967032967032967,
      "grad_norm": 0.25662714208804754,
      "learning_rate": 6.417442555066234e-07,
      "loss": 1.4927,
      "step": 895
    },
    {
      "epoch": 0.19692307692307692,
      "grad_norm": 0.2530816905689608,
      "learning_rate": 6.416181832297467e-07,
      "loss": 1.4384,
      "step": 896
    },
    {
      "epoch": 0.19714285714285715,
      "grad_norm": 0.23498808230493776,
      "learning_rate": 6.414919886140292e-07,
      "loss": 1.4791,
      "step": 897
    },
    {
      "epoch": 0.19736263736263737,
      "grad_norm": 0.26344742398120496,
      "learning_rate": 6.413656717196324e-07,
      "loss": 1.4971,
      "step": 898
    },
    {
      "epoch": 0.19758241758241757,
      "grad_norm": 0.25143082783374615,
      "learning_rate": 6.412392326067759e-07,
      "loss": 1.4597,
      "step": 899
    },
    {
      "epoch": 0.1978021978021978,
      "grad_norm": 0.23990604547072275,
      "learning_rate": 6.411126713357377e-07,
      "loss": 1.4232,
      "step": 900
    },
    {
      "epoch": 0.19802197802197802,
      "grad_norm": 0.2692939081735677,
      "learning_rate": 6.409859879668539e-07,
      "loss": 1.4706,
      "step": 901
    },
    {
      "epoch": 0.19824175824175824,
      "grad_norm": 0.269185023377902,
      "learning_rate": 6.40859182560519e-07,
      "loss": 1.477,
      "step": 902
    },
    {
      "epoch": 0.19846153846153847,
      "grad_norm": 0.25792597757080377,
      "learning_rate": 6.407322551771853e-07,
      "loss": 1.414,
      "step": 903
    },
    {
      "epoch": 0.1986813186813187,
      "grad_norm": 0.2717337009081469,
      "learning_rate": 6.406052058773639e-07,
      "loss": 1.411,
      "step": 904
    },
    {
      "epoch": 0.1989010989010989,
      "grad_norm": 0.2479518988127083,
      "learning_rate": 6.404780347216234e-07,
      "loss": 1.5042,
      "step": 905
    },
    {
      "epoch": 0.1991208791208791,
      "grad_norm": 0.2459542294208431,
      "learning_rate": 6.403507417705905e-07,
      "loss": 1.445,
      "step": 906
    },
    {
      "epoch": 0.19934065934065934,
      "grad_norm": 0.26575861998920286,
      "learning_rate": 6.402233270849507e-07,
      "loss": 1.4916,
      "step": 907
    },
    {
      "epoch": 0.19956043956043956,
      "grad_norm": 0.27081833099425034,
      "learning_rate": 6.400957907254468e-07,
      "loss": 1.4499,
      "step": 908
    },
    {
      "epoch": 0.1997802197802198,
      "grad_norm": 0.252453291758668,
      "learning_rate": 6.399681327528796e-07,
      "loss": 1.4982,
      "step": 909
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.24537417092917735,
      "learning_rate": 6.398403532281084e-07,
      "loss": 1.4778,
      "step": 910
    },
    {
      "epoch": 0.2002197802197802,
      "grad_norm": 0.2506023943950271,
      "learning_rate": 6.397124522120501e-07,
      "loss": 1.5031,
      "step": 911
    },
    {
      "epoch": 0.20043956043956043,
      "grad_norm": 0.25036021304321665,
      "learning_rate": 6.395844297656795e-07,
      "loss": 1.4888,
      "step": 912
    },
    {
      "epoch": 0.20065934065934066,
      "grad_norm": 0.24254651002406466,
      "learning_rate": 6.394562859500295e-07,
      "loss": 1.4459,
      "step": 913
    },
    {
      "epoch": 0.20087912087912088,
      "grad_norm": 0.25867272996848184,
      "learning_rate": 6.393280208261904e-07,
      "loss": 1.545,
      "step": 914
    },
    {
      "epoch": 0.2010989010989011,
      "grad_norm": 0.24893331006562164,
      "learning_rate": 6.391996344553111e-07,
      "loss": 1.5024,
      "step": 915
    },
    {
      "epoch": 0.20131868131868133,
      "grad_norm": 0.24696705887897696,
      "learning_rate": 6.390711268985977e-07,
      "loss": 1.4732,
      "step": 916
    },
    {
      "epoch": 0.20153846153846153,
      "grad_norm": 0.240259720360503,
      "learning_rate": 6.389424982173141e-07,
      "loss": 1.4846,
      "step": 917
    },
    {
      "epoch": 0.20175824175824175,
      "grad_norm": 0.24233606092879775,
      "learning_rate": 6.388137484727822e-07,
      "loss": 1.4527,
      "step": 918
    },
    {
      "epoch": 0.20197802197802198,
      "grad_norm": 0.24239663106316878,
      "learning_rate": 6.386848777263814e-07,
      "loss": 1.4479,
      "step": 919
    },
    {
      "epoch": 0.2021978021978022,
      "grad_norm": 0.23606123410211843,
      "learning_rate": 6.38555886039549e-07,
      "loss": 1.4073,
      "step": 920
    },
    {
      "epoch": 0.20241758241758243,
      "grad_norm": 0.2501744387518463,
      "learning_rate": 6.384267734737798e-07,
      "loss": 1.4662,
      "step": 921
    },
    {
      "epoch": 0.20263736263736265,
      "grad_norm": 0.25849377503081267,
      "learning_rate": 6.382975400906261e-07,
      "loss": 1.4666,
      "step": 922
    },
    {
      "epoch": 0.20285714285714285,
      "grad_norm": 0.2502574149323045,
      "learning_rate": 6.381681859516985e-07,
      "loss": 1.4485,
      "step": 923
    },
    {
      "epoch": 0.20307692307692307,
      "grad_norm": 0.2508995533000133,
      "learning_rate": 6.380387111186639e-07,
      "loss": 1.4335,
      "step": 924
    },
    {
      "epoch": 0.2032967032967033,
      "grad_norm": 0.2559998901231916,
      "learning_rate": 6.37909115653248e-07,
      "loss": 1.5031,
      "step": 925
    },
    {
      "epoch": 0.20351648351648352,
      "grad_norm": 0.26059348234814095,
      "learning_rate": 6.377793996172331e-07,
      "loss": 1.4643,
      "step": 926
    },
    {
      "epoch": 0.20373626373626375,
      "grad_norm": 0.2595881455088972,
      "learning_rate": 6.376495630724598e-07,
      "loss": 1.5158,
      "step": 927
    },
    {
      "epoch": 0.20395604395604397,
      "grad_norm": 0.2439256330478379,
      "learning_rate": 6.375196060808253e-07,
      "loss": 1.4548,
      "step": 928
    },
    {
      "epoch": 0.20417582417582417,
      "grad_norm": 0.24263873371003117,
      "learning_rate": 6.373895287042848e-07,
      "loss": 1.4889,
      "step": 929
    },
    {
      "epoch": 0.2043956043956044,
      "grad_norm": 0.23460790293609043,
      "learning_rate": 6.372593310048507e-07,
      "loss": 1.4443,
      "step": 930
    },
    {
      "epoch": 0.20461538461538462,
      "grad_norm": 0.2451319771571135,
      "learning_rate": 6.371290130445929e-07,
      "loss": 1.5045,
      "step": 931
    },
    {
      "epoch": 0.20483516483516484,
      "grad_norm": 0.2529639158714934,
      "learning_rate": 6.369985748856385e-07,
      "loss": 1.5279,
      "step": 932
    },
    {
      "epoch": 0.20505494505494506,
      "grad_norm": 0.32201940620813707,
      "learning_rate": 6.368680165901716e-07,
      "loss": 1.5645,
      "step": 933
    },
    {
      "epoch": 0.20527472527472526,
      "grad_norm": 0.2875582546042058,
      "learning_rate": 6.367373382204343e-07,
      "loss": 1.4613,
      "step": 934
    },
    {
      "epoch": 0.20549450549450549,
      "grad_norm": 0.25410679968514366,
      "learning_rate": 6.366065398387253e-07,
      "loss": 1.488,
      "step": 935
    },
    {
      "epoch": 0.2057142857142857,
      "grad_norm": 0.26336986404865037,
      "learning_rate": 6.364756215074009e-07,
      "loss": 1.4869,
      "step": 936
    },
    {
      "epoch": 0.20593406593406594,
      "grad_norm": 0.2508092121036278,
      "learning_rate": 6.363445832888743e-07,
      "loss": 1.4721,
      "step": 937
    },
    {
      "epoch": 0.20615384615384616,
      "grad_norm": 0.25674556894140477,
      "learning_rate": 6.362134252456162e-07,
      "loss": 1.5337,
      "step": 938
    },
    {
      "epoch": 0.20637362637362638,
      "grad_norm": 0.24537133466589953,
      "learning_rate": 6.36082147440154e-07,
      "loss": 1.4449,
      "step": 939
    },
    {
      "epoch": 0.20659340659340658,
      "grad_norm": 0.24937842575749275,
      "learning_rate": 6.359507499350724e-07,
      "loss": 1.5097,
      "step": 940
    },
    {
      "epoch": 0.2068131868131868,
      "grad_norm": 0.2665009293024344,
      "learning_rate": 6.358192327930133e-07,
      "loss": 1.4204,
      "step": 941
    },
    {
      "epoch": 0.20703296703296703,
      "grad_norm": 0.24297154419198846,
      "learning_rate": 6.356875960766754e-07,
      "loss": 1.4208,
      "step": 942
    },
    {
      "epoch": 0.20725274725274725,
      "grad_norm": 0.2438850102539349,
      "learning_rate": 6.355558398488147e-07,
      "loss": 1.4701,
      "step": 943
    },
    {
      "epoch": 0.20747252747252748,
      "grad_norm": 0.2514148533451485,
      "learning_rate": 6.354239641722438e-07,
      "loss": 1.5029,
      "step": 944
    },
    {
      "epoch": 0.2076923076923077,
      "grad_norm": 0.2628877418338969,
      "learning_rate": 6.352919691098326e-07,
      "loss": 1.4808,
      "step": 945
    },
    {
      "epoch": 0.2079120879120879,
      "grad_norm": 0.25494646791834513,
      "learning_rate": 6.351598547245075e-07,
      "loss": 1.4681,
      "step": 946
    },
    {
      "epoch": 0.20813186813186813,
      "grad_norm": 0.25407267277377743,
      "learning_rate": 6.350276210792524e-07,
      "loss": 1.4641,
      "step": 947
    },
    {
      "epoch": 0.20835164835164835,
      "grad_norm": 0.26857491418989893,
      "learning_rate": 6.348952682371073e-07,
      "loss": 1.4759,
      "step": 948
    },
    {
      "epoch": 0.20857142857142857,
      "grad_norm": 0.23972152226069182,
      "learning_rate": 6.347627962611697e-07,
      "loss": 1.454,
      "step": 949
    },
    {
      "epoch": 0.2087912087912088,
      "grad_norm": 0.2565036917274771,
      "learning_rate": 6.346302052145935e-07,
      "loss": 1.4978,
      "step": 950
    },
    {
      "epoch": 0.20901098901098902,
      "grad_norm": 0.258139709502027,
      "learning_rate": 6.344974951605894e-07,
      "loss": 1.4461,
      "step": 951
    },
    {
      "epoch": 0.20923076923076922,
      "grad_norm": 0.25213149359953296,
      "learning_rate": 6.343646661624251e-07,
      "loss": 1.433,
      "step": 952
    },
    {
      "epoch": 0.20945054945054944,
      "grad_norm": 0.2517626189796978,
      "learning_rate": 6.342317182834248e-07,
      "loss": 1.4274,
      "step": 953
    },
    {
      "epoch": 0.20967032967032967,
      "grad_norm": 0.2474446088373573,
      "learning_rate": 6.340986515869691e-07,
      "loss": 1.4738,
      "step": 954
    },
    {
      "epoch": 0.2098901098901099,
      "grad_norm": 0.2583933717951051,
      "learning_rate": 6.339654661364959e-07,
      "loss": 1.4963,
      "step": 955
    },
    {
      "epoch": 0.21010989010989012,
      "grad_norm": 0.2629986391355057,
      "learning_rate": 6.338321619954993e-07,
      "loss": 1.4565,
      "step": 956
    },
    {
      "epoch": 0.21032967032967034,
      "grad_norm": 0.2457716206199386,
      "learning_rate": 6.336987392275298e-07,
      "loss": 1.5438,
      "step": 957
    },
    {
      "epoch": 0.21054945054945054,
      "grad_norm": 0.2772638128277807,
      "learning_rate": 6.335651978961948e-07,
      "loss": 1.3944,
      "step": 958
    },
    {
      "epoch": 0.21076923076923076,
      "grad_norm": 0.25403073870139725,
      "learning_rate": 6.334315380651581e-07,
      "loss": 1.5137,
      "step": 959
    },
    {
      "epoch": 0.210989010989011,
      "grad_norm": 0.24811373405946546,
      "learning_rate": 6.332977597981402e-07,
      "loss": 1.463,
      "step": 960
    },
    {
      "epoch": 0.2112087912087912,
      "grad_norm": 0.2615651322350011,
      "learning_rate": 6.331638631589177e-07,
      "loss": 1.4613,
      "step": 961
    },
    {
      "epoch": 0.21142857142857144,
      "grad_norm": 0.263440486156173,
      "learning_rate": 6.330298482113238e-07,
      "loss": 1.4732,
      "step": 962
    },
    {
      "epoch": 0.21164835164835163,
      "grad_norm": 0.25509452968099045,
      "learning_rate": 6.328957150192481e-07,
      "loss": 1.4419,
      "step": 963
    },
    {
      "epoch": 0.21186813186813186,
      "grad_norm": 0.2558217352425839,
      "learning_rate": 6.327614636466365e-07,
      "loss": 1.5063,
      "step": 964
    },
    {
      "epoch": 0.21208791208791208,
      "grad_norm": 0.24999222039824714,
      "learning_rate": 6.326270941574915e-07,
      "loss": 1.4379,
      "step": 965
    },
    {
      "epoch": 0.2123076923076923,
      "grad_norm": 0.25986839759589353,
      "learning_rate": 6.324926066158715e-07,
      "loss": 1.495,
      "step": 966
    },
    {
      "epoch": 0.21252747252747253,
      "grad_norm": 0.2501101206247595,
      "learning_rate": 6.323580010858915e-07,
      "loss": 1.4537,
      "step": 967
    },
    {
      "epoch": 0.21274725274725276,
      "grad_norm": 0.26059640962348585,
      "learning_rate": 6.322232776317227e-07,
      "loss": 1.4483,
      "step": 968
    },
    {
      "epoch": 0.21296703296703295,
      "grad_norm": 0.3253965606400039,
      "learning_rate": 6.320884363175926e-07,
      "loss": 1.5221,
      "step": 969
    },
    {
      "epoch": 0.21318681318681318,
      "grad_norm": 0.2738405972770376,
      "learning_rate": 6.319534772077844e-07,
      "loss": 1.4758,
      "step": 970
    },
    {
      "epoch": 0.2134065934065934,
      "grad_norm": 0.23998450538869,
      "learning_rate": 6.318184003666382e-07,
      "loss": 1.5019,
      "step": 971
    },
    {
      "epoch": 0.21362637362637363,
      "grad_norm": 0.2750740502569545,
      "learning_rate": 6.316832058585496e-07,
      "loss": 1.4747,
      "step": 972
    },
    {
      "epoch": 0.21384615384615385,
      "grad_norm": 0.2495073621097044,
      "learning_rate": 6.315478937479705e-07,
      "loss": 1.4663,
      "step": 973
    },
    {
      "epoch": 0.21406593406593408,
      "grad_norm": 0.25360133485533454,
      "learning_rate": 6.314124640994091e-07,
      "loss": 1.4852,
      "step": 974
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.25701838922538556,
      "learning_rate": 6.312769169774293e-07,
      "loss": 1.4635,
      "step": 975
    },
    {
      "epoch": 0.2145054945054945,
      "grad_norm": 0.245066880216118,
      "learning_rate": 6.311412524466512e-07,
      "loss": 1.4443,
      "step": 976
    },
    {
      "epoch": 0.21472527472527472,
      "grad_norm": 0.27326964827957495,
      "learning_rate": 6.310054705717509e-07,
      "loss": 1.4559,
      "step": 977
    },
    {
      "epoch": 0.21494505494505495,
      "grad_norm": 0.4643132469222394,
      "learning_rate": 6.308695714174602e-07,
      "loss": 1.4578,
      "step": 978
    },
    {
      "epoch": 0.21516483516483517,
      "grad_norm": 0.2553315407913207,
      "learning_rate": 6.30733555048567e-07,
      "loss": 1.4741,
      "step": 979
    },
    {
      "epoch": 0.2153846153846154,
      "grad_norm": 0.2866978785567234,
      "learning_rate": 6.30597421529915e-07,
      "loss": 1.4369,
      "step": 980
    },
    {
      "epoch": 0.2156043956043956,
      "grad_norm": 0.254151772984924,
      "learning_rate": 6.30461170926404e-07,
      "loss": 1.4763,
      "step": 981
    },
    {
      "epoch": 0.21582417582417582,
      "grad_norm": 0.2505294152945484,
      "learning_rate": 6.303248033029892e-07,
      "loss": 1.5003,
      "step": 982
    },
    {
      "epoch": 0.21604395604395604,
      "grad_norm": 0.24788177291163854,
      "learning_rate": 6.301883187246819e-07,
      "loss": 1.4565,
      "step": 983
    },
    {
      "epoch": 0.21626373626373627,
      "grad_norm": 0.25852553005445156,
      "learning_rate": 6.300517172565491e-07,
      "loss": 1.4868,
      "step": 984
    },
    {
      "epoch": 0.2164835164835165,
      "grad_norm": 0.27705295208335157,
      "learning_rate": 6.299149989637134e-07,
      "loss": 1.5281,
      "step": 985
    },
    {
      "epoch": 0.21670329670329672,
      "grad_norm": 0.2486468181676995,
      "learning_rate": 6.297781639113532e-07,
      "loss": 1.5065,
      "step": 986
    },
    {
      "epoch": 0.2169230769230769,
      "grad_norm": 0.25924678229751535,
      "learning_rate": 6.296412121647026e-07,
      "loss": 1.4719,
      "step": 987
    },
    {
      "epoch": 0.21714285714285714,
      "grad_norm": 0.2508896633465167,
      "learning_rate": 6.295041437890513e-07,
      "loss": 1.4852,
      "step": 988
    },
    {
      "epoch": 0.21736263736263736,
      "grad_norm": 0.2564252926625278,
      "learning_rate": 6.293669588497444e-07,
      "loss": 1.5041,
      "step": 989
    },
    {
      "epoch": 0.2175824175824176,
      "grad_norm": 0.24729873883885692,
      "learning_rate": 6.292296574121828e-07,
      "loss": 1.462,
      "step": 990
    },
    {
      "epoch": 0.2178021978021978,
      "grad_norm": 0.24882002931157068,
      "learning_rate": 6.29092239541823e-07,
      "loss": 1.4737,
      "step": 991
    },
    {
      "epoch": 0.21802197802197804,
      "grad_norm": 0.2477338170470924,
      "learning_rate": 6.289547053041769e-07,
      "loss": 1.4523,
      "step": 992
    },
    {
      "epoch": 0.21824175824175823,
      "grad_norm": 0.2524284804610147,
      "learning_rate": 6.288170547648117e-07,
      "loss": 1.4448,
      "step": 993
    },
    {
      "epoch": 0.21846153846153846,
      "grad_norm": 0.2618226168369328,
      "learning_rate": 6.286792879893504e-07,
      "loss": 1.4149,
      "step": 994
    },
    {
      "epoch": 0.21868131868131868,
      "grad_norm": 0.26073755610441146,
      "learning_rate": 6.285414050434712e-07,
      "loss": 1.4662,
      "step": 995
    },
    {
      "epoch": 0.2189010989010989,
      "grad_norm": 0.2509363125644294,
      "learning_rate": 6.284034059929075e-07,
      "loss": 1.4694,
      "step": 996
    },
    {
      "epoch": 0.21912087912087913,
      "grad_norm": 0.2604106550335233,
      "learning_rate": 6.282652909034486e-07,
      "loss": 1.465,
      "step": 997
    },
    {
      "epoch": 0.21934065934065933,
      "grad_norm": 0.2613176452159072,
      "learning_rate": 6.281270598409384e-07,
      "loss": 1.5201,
      "step": 998
    },
    {
      "epoch": 0.21956043956043955,
      "grad_norm": 0.25412235551601126,
      "learning_rate": 6.279887128712769e-07,
      "loss": 1.5038,
      "step": 999
    },
    {
      "epoch": 0.21978021978021978,
      "grad_norm": 0.24241093033928177,
      "learning_rate": 6.278502500604185e-07,
      "loss": 1.5119,
      "step": 1000
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2564936510368631,
      "learning_rate": 6.277116714743736e-07,
      "loss": 1.4621,
      "step": 1001
    },
    {
      "epoch": 0.22021978021978023,
      "grad_norm": 0.25132189781818387,
      "learning_rate": 6.275729771792071e-07,
      "loss": 1.4864,
      "step": 1002
    },
    {
      "epoch": 0.22043956043956045,
      "grad_norm": 0.262435909568915,
      "learning_rate": 6.274341672410399e-07,
      "loss": 1.4876,
      "step": 1003
    },
    {
      "epoch": 0.22065934065934065,
      "grad_norm": 0.2442607144061378,
      "learning_rate": 6.27295241726047e-07,
      "loss": 1.4955,
      "step": 1004
    },
    {
      "epoch": 0.22087912087912087,
      "grad_norm": 0.24918844109204902,
      "learning_rate": 6.271562007004594e-07,
      "loss": 1.4581,
      "step": 1005
    },
    {
      "epoch": 0.2210989010989011,
      "grad_norm": 0.3007102940008447,
      "learning_rate": 6.270170442305628e-07,
      "loss": 1.4649,
      "step": 1006
    },
    {
      "epoch": 0.22131868131868132,
      "grad_norm": 0.24458518612482621,
      "learning_rate": 6.268777723826977e-07,
      "loss": 1.5223,
      "step": 1007
    },
    {
      "epoch": 0.22153846153846155,
      "grad_norm": 0.24485989164173935,
      "learning_rate": 6.267383852232599e-07,
      "loss": 1.4227,
      "step": 1008
    },
    {
      "epoch": 0.22175824175824177,
      "grad_norm": 0.26994854817969943,
      "learning_rate": 6.265988828187004e-07,
      "loss": 1.4892,
      "step": 1009
    },
    {
      "epoch": 0.22197802197802197,
      "grad_norm": 0.25257673752539994,
      "learning_rate": 6.264592652355246e-07,
      "loss": 1.4901,
      "step": 1010
    },
    {
      "epoch": 0.2221978021978022,
      "grad_norm": 0.44345204761261503,
      "learning_rate": 6.26319532540293e-07,
      "loss": 1.4548,
      "step": 1011
    },
    {
      "epoch": 0.22241758241758242,
      "grad_norm": 0.26091916059194475,
      "learning_rate": 6.261796847996214e-07,
      "loss": 1.4834,
      "step": 1012
    },
    {
      "epoch": 0.22263736263736264,
      "grad_norm": 0.25779038287917105,
      "learning_rate": 6.260397220801798e-07,
      "loss": 1.4937,
      "step": 1013
    },
    {
      "epoch": 0.22285714285714286,
      "grad_norm": 0.24084632378403625,
      "learning_rate": 6.258996444486935e-07,
      "loss": 1.438,
      "step": 1014
    },
    {
      "epoch": 0.2230769230769231,
      "grad_norm": 0.24738965156649764,
      "learning_rate": 6.257594519719421e-07,
      "loss": 1.4311,
      "step": 1015
    },
    {
      "epoch": 0.22329670329670329,
      "grad_norm": 0.2677682094041224,
      "learning_rate": 6.256191447167607e-07,
      "loss": 1.4418,
      "step": 1016
    },
    {
      "epoch": 0.2235164835164835,
      "grad_norm": 0.24732574403308874,
      "learning_rate": 6.254787227500382e-07,
      "loss": 1.455,
      "step": 1017
    },
    {
      "epoch": 0.22373626373626374,
      "grad_norm": 0.2738078752234439,
      "learning_rate": 6.25338186138719e-07,
      "loss": 1.4768,
      "step": 1018
    },
    {
      "epoch": 0.22395604395604396,
      "grad_norm": 0.2530054621084579,
      "learning_rate": 6.251975349498016e-07,
      "loss": 1.4965,
      "step": 1019
    },
    {
      "epoch": 0.22417582417582418,
      "grad_norm": 0.25608936529023674,
      "learning_rate": 6.250567692503394e-07,
      "loss": 1.479,
      "step": 1020
    },
    {
      "epoch": 0.2243956043956044,
      "grad_norm": 0.2530095649283402,
      "learning_rate": 6.249158891074402e-07,
      "loss": 1.5014,
      "step": 1021
    },
    {
      "epoch": 0.2246153846153846,
      "grad_norm": 0.28426157573050004,
      "learning_rate": 6.247748945882668e-07,
      "loss": 1.4383,
      "step": 1022
    },
    {
      "epoch": 0.22483516483516483,
      "grad_norm": 0.2571595455354892,
      "learning_rate": 6.246337857600359e-07,
      "loss": 1.4475,
      "step": 1023
    },
    {
      "epoch": 0.22505494505494505,
      "grad_norm": 0.2431164945851439,
      "learning_rate": 6.24492562690019e-07,
      "loss": 1.461,
      "step": 1024
    },
    {
      "epoch": 0.22527472527472528,
      "grad_norm": 0.2965910957444324,
      "learning_rate": 6.243512254455423e-07,
      "loss": 1.4434,
      "step": 1025
    },
    {
      "epoch": 0.2254945054945055,
      "grad_norm": 0.24318863816189085,
      "learning_rate": 6.24209774093986e-07,
      "loss": 1.4125,
      "step": 1026
    },
    {
      "epoch": 0.2257142857142857,
      "grad_norm": 0.2699868320603675,
      "learning_rate": 6.24068208702785e-07,
      "loss": 1.4273,
      "step": 1027
    },
    {
      "epoch": 0.22593406593406593,
      "grad_norm": 0.2573116416423263,
      "learning_rate": 6.239265293394283e-07,
      "loss": 1.4804,
      "step": 1028
    },
    {
      "epoch": 0.22615384615384615,
      "grad_norm": 0.2537042712591057,
      "learning_rate": 6.237847360714597e-07,
      "loss": 1.4497,
      "step": 1029
    },
    {
      "epoch": 0.22637362637362637,
      "grad_norm": 0.56996514124249,
      "learning_rate": 6.236428289664767e-07,
      "loss": 1.453,
      "step": 1030
    },
    {
      "epoch": 0.2265934065934066,
      "grad_norm": 0.2556228682124189,
      "learning_rate": 6.235008080921314e-07,
      "loss": 1.4659,
      "step": 1031
    },
    {
      "epoch": 0.22681318681318682,
      "grad_norm": 0.2718315386107101,
      "learning_rate": 6.233586735161303e-07,
      "loss": 1.4188,
      "step": 1032
    },
    {
      "epoch": 0.22703296703296702,
      "grad_norm": 0.2462699534901121,
      "learning_rate": 6.232164253062336e-07,
      "loss": 1.4673,
      "step": 1033
    },
    {
      "epoch": 0.22725274725274724,
      "grad_norm": 0.25640174813341954,
      "learning_rate": 6.230740635302563e-07,
      "loss": 1.4689,
      "step": 1034
    },
    {
      "epoch": 0.22747252747252747,
      "grad_norm": 0.2754401724093636,
      "learning_rate": 6.229315882560671e-07,
      "loss": 1.4283,
      "step": 1035
    },
    {
      "epoch": 0.2276923076923077,
      "grad_norm": 0.2898185200896864,
      "learning_rate": 6.22788999551589e-07,
      "loss": 1.5236,
      "step": 1036
    },
    {
      "epoch": 0.22791208791208792,
      "grad_norm": 0.2501607998148379,
      "learning_rate": 6.226462974847989e-07,
      "loss": 1.4728,
      "step": 1037
    },
    {
      "epoch": 0.22813186813186814,
      "grad_norm": 0.26636250011797713,
      "learning_rate": 6.22503482123728e-07,
      "loss": 1.5376,
      "step": 1038
    },
    {
      "epoch": 0.22835164835164834,
      "grad_norm": 0.2502974862297266,
      "learning_rate": 6.223605535364611e-07,
      "loss": 1.4333,
      "step": 1039
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 0.2564062296316927,
      "learning_rate": 6.222175117911375e-07,
      "loss": 1.4807,
      "step": 1040
    },
    {
      "epoch": 0.2287912087912088,
      "grad_norm": 0.25275825535718294,
      "learning_rate": 6.220743569559499e-07,
      "loss": 1.4511,
      "step": 1041
    },
    {
      "epoch": 0.229010989010989,
      "grad_norm": 0.3662792300369227,
      "learning_rate": 6.219310890991456e-07,
      "loss": 1.4546,
      "step": 1042
    },
    {
      "epoch": 0.22923076923076924,
      "grad_norm": 0.24743135427408056,
      "learning_rate": 6.217877082890251e-07,
      "loss": 1.4985,
      "step": 1043
    },
    {
      "epoch": 0.22945054945054946,
      "grad_norm": 0.2444138562542184,
      "learning_rate": 6.216442145939431e-07,
      "loss": 1.4754,
      "step": 1044
    },
    {
      "epoch": 0.22967032967032966,
      "grad_norm": 0.24883832216323296,
      "learning_rate": 6.215006080823083e-07,
      "loss": 1.4691,
      "step": 1045
    },
    {
      "epoch": 0.22989010989010988,
      "grad_norm": 0.2585144888066422,
      "learning_rate": 6.213568888225825e-07,
      "loss": 1.5124,
      "step": 1046
    },
    {
      "epoch": 0.2301098901098901,
      "grad_norm": 0.2519698762792617,
      "learning_rate": 6.212130568832819e-07,
      "loss": 1.4571,
      "step": 1047
    },
    {
      "epoch": 0.23032967032967033,
      "grad_norm": 0.24806043565398622,
      "learning_rate": 6.210691123329763e-07,
      "loss": 1.4892,
      "step": 1048
    },
    {
      "epoch": 0.23054945054945056,
      "grad_norm": 0.2414973297774547,
      "learning_rate": 6.209250552402888e-07,
      "loss": 1.4565,
      "step": 1049
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 0.27596519798931884,
      "learning_rate": 6.207808856738968e-07,
      "loss": 1.4836,
      "step": 1050
    },
    {
      "epoch": 0.23098901098901098,
      "grad_norm": 0.24408299883564805,
      "learning_rate": 6.206366037025308e-07,
      "loss": 1.366,
      "step": 1051
    },
    {
      "epoch": 0.2312087912087912,
      "grad_norm": 0.2568232578421352,
      "learning_rate": 6.204922093949749e-07,
      "loss": 1.4863,
      "step": 1052
    },
    {
      "epoch": 0.23142857142857143,
      "grad_norm": 0.2593019193101645,
      "learning_rate": 6.203477028200672e-07,
      "loss": 1.4893,
      "step": 1053
    },
    {
      "epoch": 0.23164835164835165,
      "grad_norm": 0.27805061915979157,
      "learning_rate": 6.202030840466987e-07,
      "loss": 1.4951,
      "step": 1054
    },
    {
      "epoch": 0.23186813186813188,
      "grad_norm": 0.28336235205970894,
      "learning_rate": 6.200583531438145e-07,
      "loss": 1.475,
      "step": 1055
    },
    {
      "epoch": 0.2320879120879121,
      "grad_norm": 0.2572369675479279,
      "learning_rate": 6.199135101804126e-07,
      "loss": 1.5397,
      "step": 1056
    },
    {
      "epoch": 0.2323076923076923,
      "grad_norm": 0.28325937224786646,
      "learning_rate": 6.19768555225545e-07,
      "loss": 1.4759,
      "step": 1057
    },
    {
      "epoch": 0.23252747252747252,
      "grad_norm": 0.2572233336831093,
      "learning_rate": 6.196234883483166e-07,
      "loss": 1.4766,
      "step": 1058
    },
    {
      "epoch": 0.23274725274725275,
      "grad_norm": 0.24063422185551667,
      "learning_rate": 6.194783096178858e-07,
      "loss": 1.4624,
      "step": 1059
    },
    {
      "epoch": 0.23296703296703297,
      "grad_norm": 0.25066074513711945,
      "learning_rate": 6.193330191034645e-07,
      "loss": 1.4517,
      "step": 1060
    },
    {
      "epoch": 0.2331868131868132,
      "grad_norm": 0.25145707077417156,
      "learning_rate": 6.191876168743177e-07,
      "loss": 1.5347,
      "step": 1061
    },
    {
      "epoch": 0.2334065934065934,
      "grad_norm": 0.25596965013778983,
      "learning_rate": 6.190421029997638e-07,
      "loss": 1.4145,
      "step": 1062
    },
    {
      "epoch": 0.23362637362637362,
      "grad_norm": 0.24480346083152169,
      "learning_rate": 6.18896477549174e-07,
      "loss": 1.4719,
      "step": 1063
    },
    {
      "epoch": 0.23384615384615384,
      "grad_norm": 0.2647756689489498,
      "learning_rate": 6.187507405919735e-07,
      "loss": 1.4484,
      "step": 1064
    },
    {
      "epoch": 0.23406593406593407,
      "grad_norm": 0.24807787727321778,
      "learning_rate": 6.186048921976398e-07,
      "loss": 1.4709,
      "step": 1065
    },
    {
      "epoch": 0.2342857142857143,
      "grad_norm": 0.26266770594278166,
      "learning_rate": 6.184589324357042e-07,
      "loss": 1.4805,
      "step": 1066
    },
    {
      "epoch": 0.23450549450549452,
      "grad_norm": 0.244979882115583,
      "learning_rate": 6.183128613757507e-07,
      "loss": 1.4319,
      "step": 1067
    },
    {
      "epoch": 0.2347252747252747,
      "grad_norm": 0.25745588471583153,
      "learning_rate": 6.181666790874166e-07,
      "loss": 1.4383,
      "step": 1068
    },
    {
      "epoch": 0.23494505494505494,
      "grad_norm": 0.2610718353414187,
      "learning_rate": 6.180203856403917e-07,
      "loss": 1.3963,
      "step": 1069
    },
    {
      "epoch": 0.23516483516483516,
      "grad_norm": 0.2476652541181489,
      "learning_rate": 6.178739811044196e-07,
      "loss": 1.5077,
      "step": 1070
    },
    {
      "epoch": 0.2353846153846154,
      "grad_norm": 0.2507386054259677,
      "learning_rate": 6.177274655492963e-07,
      "loss": 1.3914,
      "step": 1071
    },
    {
      "epoch": 0.2356043956043956,
      "grad_norm": 0.2635488437850268,
      "learning_rate": 6.175808390448709e-07,
      "loss": 1.4864,
      "step": 1072
    },
    {
      "epoch": 0.23582417582417584,
      "grad_norm": 0.3189120592904306,
      "learning_rate": 6.174341016610453e-07,
      "loss": 1.4856,
      "step": 1073
    },
    {
      "epoch": 0.23604395604395603,
      "grad_norm": 0.2514042402305899,
      "learning_rate": 6.172872534677745e-07,
      "loss": 1.4953,
      "step": 1074
    },
    {
      "epoch": 0.23626373626373626,
      "grad_norm": 0.2509635934820503,
      "learning_rate": 6.171402945350661e-07,
      "loss": 1.5035,
      "step": 1075
    },
    {
      "epoch": 0.23648351648351648,
      "grad_norm": 0.2518564268565751,
      "learning_rate": 6.169932249329803e-07,
      "loss": 1.4302,
      "step": 1076
    },
    {
      "epoch": 0.2367032967032967,
      "grad_norm": 0.31047075081107933,
      "learning_rate": 6.168460447316305e-07,
      "loss": 1.4227,
      "step": 1077
    },
    {
      "epoch": 0.23692307692307693,
      "grad_norm": 0.2550929504416796,
      "learning_rate": 6.166987540011829e-07,
      "loss": 1.495,
      "step": 1078
    },
    {
      "epoch": 0.23714285714285716,
      "grad_norm": 0.250243931654738,
      "learning_rate": 6.165513528118556e-07,
      "loss": 1.505,
      "step": 1079
    },
    {
      "epoch": 0.23736263736263735,
      "grad_norm": 0.24949386266821572,
      "learning_rate": 6.164038412339203e-07,
      "loss": 1.4828,
      "step": 1080
    },
    {
      "epoch": 0.23758241758241758,
      "grad_norm": 0.25401637665261445,
      "learning_rate": 6.162562193377008e-07,
      "loss": 1.4658,
      "step": 1081
    },
    {
      "epoch": 0.2378021978021978,
      "grad_norm": 0.2602092323257622,
      "learning_rate": 6.161084871935735e-07,
      "loss": 1.4638,
      "step": 1082
    },
    {
      "epoch": 0.23802197802197803,
      "grad_norm": 0.23971141027923804,
      "learning_rate": 6.159606448719676e-07,
      "loss": 1.4253,
      "step": 1083
    },
    {
      "epoch": 0.23824175824175825,
      "grad_norm": 0.24009592180226755,
      "learning_rate": 6.158126924433646e-07,
      "loss": 1.3871,
      "step": 1084
    },
    {
      "epoch": 0.23846153846153847,
      "grad_norm": 0.26622764617315986,
      "learning_rate": 6.156646299782986e-07,
      "loss": 1.5069,
      "step": 1085
    },
    {
      "epoch": 0.23868131868131867,
      "grad_norm": 0.28647737693092973,
      "learning_rate": 6.155164575473562e-07,
      "loss": 1.4419,
      "step": 1086
    },
    {
      "epoch": 0.2389010989010989,
      "grad_norm": 0.25808251652871966,
      "learning_rate": 6.153681752211764e-07,
      "loss": 1.4971,
      "step": 1087
    },
    {
      "epoch": 0.23912087912087912,
      "grad_norm": 0.25735275747936,
      "learning_rate": 6.152197830704505e-07,
      "loss": 1.4938,
      "step": 1088
    },
    {
      "epoch": 0.23934065934065935,
      "grad_norm": 0.24904505230549445,
      "learning_rate": 6.150712811659221e-07,
      "loss": 1.4535,
      "step": 1089
    },
    {
      "epoch": 0.23956043956043957,
      "grad_norm": 0.253228066051037,
      "learning_rate": 6.149226695783873e-07,
      "loss": 1.5039,
      "step": 1090
    },
    {
      "epoch": 0.23978021978021977,
      "grad_norm": 0.2419367774032695,
      "learning_rate": 6.147739483786947e-07,
      "loss": 1.5014,
      "step": 1091
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24568896065683954,
      "learning_rate": 6.146251176377446e-07,
      "loss": 1.4932,
      "step": 1092
    },
    {
      "epoch": 0.24021978021978022,
      "grad_norm": 0.24254086343904321,
      "learning_rate": 6.144761774264899e-07,
      "loss": 1.4741,
      "step": 1093
    },
    {
      "epoch": 0.24043956043956044,
      "grad_norm": 0.2503162565465864,
      "learning_rate": 6.143271278159354e-07,
      "loss": 1.5192,
      "step": 1094
    },
    {
      "epoch": 0.24065934065934066,
      "grad_norm": 0.2633170556423953,
      "learning_rate": 6.141779688771385e-07,
      "loss": 1.4848,
      "step": 1095
    },
    {
      "epoch": 0.2408791208791209,
      "grad_norm": 0.2666038663461278,
      "learning_rate": 6.140287006812085e-07,
      "loss": 1.4404,
      "step": 1096
    },
    {
      "epoch": 0.24109890109890109,
      "grad_norm": 0.26634068694512053,
      "learning_rate": 6.138793232993064e-07,
      "loss": 1.449,
      "step": 1097
    },
    {
      "epoch": 0.2413186813186813,
      "grad_norm": 0.24613572287675384,
      "learning_rate": 6.137298368026461e-07,
      "loss": 1.4759,
      "step": 1098
    },
    {
      "epoch": 0.24153846153846154,
      "grad_norm": 0.2522587364888656,
      "learning_rate": 6.135802412624926e-07,
      "loss": 1.4832,
      "step": 1099
    },
    {
      "epoch": 0.24175824175824176,
      "grad_norm": 0.24552154772760235,
      "learning_rate": 6.134305367501636e-07,
      "loss": 1.4248,
      "step": 1100
    },
    {
      "epoch": 0.24197802197802198,
      "grad_norm": 0.2533079304799372,
      "learning_rate": 6.132807233370282e-07,
      "loss": 1.4299,
      "step": 1101
    },
    {
      "epoch": 0.2421978021978022,
      "grad_norm": 0.24640104021676112,
      "learning_rate": 6.13130801094508e-07,
      "loss": 1.4303,
      "step": 1102
    },
    {
      "epoch": 0.2424175824175824,
      "grad_norm": 0.28495899690235366,
      "learning_rate": 6.129807700940759e-07,
      "loss": 1.4791,
      "step": 1103
    },
    {
      "epoch": 0.24263736263736263,
      "grad_norm": 0.24643612019427724,
      "learning_rate": 6.128306304072571e-07,
      "loss": 1.4784,
      "step": 1104
    },
    {
      "epoch": 0.24285714285714285,
      "grad_norm": 0.25865255982401164,
      "learning_rate": 6.126803821056284e-07,
      "loss": 1.5042,
      "step": 1105
    },
    {
      "epoch": 0.24307692307692308,
      "grad_norm": 0.23727486138472725,
      "learning_rate": 6.125300252608182e-07,
      "loss": 1.4437,
      "step": 1106
    },
    {
      "epoch": 0.2432967032967033,
      "grad_norm": 0.30202384726110537,
      "learning_rate": 6.12379559944507e-07,
      "loss": 1.4983,
      "step": 1107
    },
    {
      "epoch": 0.24351648351648353,
      "grad_norm": 0.2805173630023854,
      "learning_rate": 6.122289862284268e-07,
      "loss": 1.5131,
      "step": 1108
    },
    {
      "epoch": 0.24373626373626373,
      "grad_norm": 0.26879845342320885,
      "learning_rate": 6.120783041843616e-07,
      "loss": 1.5035,
      "step": 1109
    },
    {
      "epoch": 0.24395604395604395,
      "grad_norm": 0.25402096552527337,
      "learning_rate": 6.119275138841464e-07,
      "loss": 1.4861,
      "step": 1110
    },
    {
      "epoch": 0.24417582417582417,
      "grad_norm": 0.31932832501073694,
      "learning_rate": 6.117766153996686e-07,
      "loss": 1.4874,
      "step": 1111
    },
    {
      "epoch": 0.2443956043956044,
      "grad_norm": 0.2532955815762932,
      "learning_rate": 6.116256088028665e-07,
      "loss": 1.5421,
      "step": 1112
    },
    {
      "epoch": 0.24461538461538462,
      "grad_norm": 0.2571008324647733,
      "learning_rate": 6.114744941657302e-07,
      "loss": 1.5376,
      "step": 1113
    },
    {
      "epoch": 0.24483516483516485,
      "grad_norm": 0.26672802162222276,
      "learning_rate": 6.113232715603014e-07,
      "loss": 1.5261,
      "step": 1114
    },
    {
      "epoch": 0.24505494505494504,
      "grad_norm": 0.27630820615775087,
      "learning_rate": 6.111719410586733e-07,
      "loss": 1.4139,
      "step": 1115
    },
    {
      "epoch": 0.24527472527472527,
      "grad_norm": 0.25158388736303877,
      "learning_rate": 6.110205027329901e-07,
      "loss": 1.4808,
      "step": 1116
    },
    {
      "epoch": 0.2454945054945055,
      "grad_norm": 0.2536800550823491,
      "learning_rate": 6.108689566554482e-07,
      "loss": 1.4362,
      "step": 1117
    },
    {
      "epoch": 0.24571428571428572,
      "grad_norm": 0.2522730808904053,
      "learning_rate": 6.107173028982945e-07,
      "loss": 1.4658,
      "step": 1118
    },
    {
      "epoch": 0.24593406593406594,
      "grad_norm": 0.25592103393523086,
      "learning_rate": 6.105655415338279e-07,
      "loss": 1.5232,
      "step": 1119
    },
    {
      "epoch": 0.24615384615384617,
      "grad_norm": 0.26320712384616624,
      "learning_rate": 6.104136726343981e-07,
      "loss": 1.4594,
      "step": 1120
    },
    {
      "epoch": 0.24637362637362636,
      "grad_norm": 0.25206434885669243,
      "learning_rate": 6.102616962724065e-07,
      "loss": 1.5077,
      "step": 1121
    },
    {
      "epoch": 0.2465934065934066,
      "grad_norm": 0.2506092463649332,
      "learning_rate": 6.101096125203055e-07,
      "loss": 1.4598,
      "step": 1122
    },
    {
      "epoch": 0.2468131868131868,
      "grad_norm": 0.2535839157735764,
      "learning_rate": 6.099574214505987e-07,
      "loss": 1.5111,
      "step": 1123
    },
    {
      "epoch": 0.24703296703296704,
      "grad_norm": 0.2706976752314015,
      "learning_rate": 6.098051231358407e-07,
      "loss": 1.5531,
      "step": 1124
    },
    {
      "epoch": 0.24725274725274726,
      "grad_norm": 0.261814361613236,
      "learning_rate": 6.096527176486378e-07,
      "loss": 1.4746,
      "step": 1125
    },
    {
      "epoch": 0.24747252747252746,
      "grad_norm": 0.25563828066269206,
      "learning_rate": 6.095002050616468e-07,
      "loss": 1.4351,
      "step": 1126
    },
    {
      "epoch": 0.24769230769230768,
      "grad_norm": 0.24262746175929104,
      "learning_rate": 6.093475854475757e-07,
      "loss": 1.4482,
      "step": 1127
    },
    {
      "epoch": 0.2479120879120879,
      "grad_norm": 0.24905697752627842,
      "learning_rate": 6.091948588791837e-07,
      "loss": 1.4243,
      "step": 1128
    },
    {
      "epoch": 0.24813186813186813,
      "grad_norm": 0.27964225780152474,
      "learning_rate": 6.09042025429281e-07,
      "loss": 1.4902,
      "step": 1129
    },
    {
      "epoch": 0.24835164835164836,
      "grad_norm": 0.25082380315304037,
      "learning_rate": 6.088890851707283e-07,
      "loss": 1.4631,
      "step": 1130
    },
    {
      "epoch": 0.24857142857142858,
      "grad_norm": 0.24560773808871394,
      "learning_rate": 6.08736038176438e-07,
      "loss": 1.4844,
      "step": 1131
    },
    {
      "epoch": 0.24879120879120878,
      "grad_norm": 0.24299561284907775,
      "learning_rate": 6.085828845193726e-07,
      "loss": 1.4254,
      "step": 1132
    },
    {
      "epoch": 0.249010989010989,
      "grad_norm": 0.2549622820375491,
      "learning_rate": 6.084296242725458e-07,
      "loss": 1.5371,
      "step": 1133
    },
    {
      "epoch": 0.24923076923076923,
      "grad_norm": 0.2610460521985358,
      "learning_rate": 6.082762575090224e-07,
      "loss": 1.5026,
      "step": 1134
    },
    {
      "epoch": 0.24945054945054945,
      "grad_norm": 0.25629212629555614,
      "learning_rate": 6.081227843019173e-07,
      "loss": 1.4537,
      "step": 1135
    },
    {
      "epoch": 0.24967032967032968,
      "grad_norm": 0.24685471940099946,
      "learning_rate": 6.079692047243967e-07,
      "loss": 1.4827,
      "step": 1136
    },
    {
      "epoch": 0.2498901098901099,
      "grad_norm": 0.2453785314020694,
      "learning_rate": 6.078155188496775e-07,
      "loss": 1.4802,
      "step": 1137
    },
    {
      "epoch": 0.2501098901098901,
      "grad_norm": 0.24864026929057834,
      "learning_rate": 6.076617267510269e-07,
      "loss": 1.4681,
      "step": 1138
    },
    {
      "epoch": 0.25032967032967035,
      "grad_norm": 0.257752415113131,
      "learning_rate": 6.075078285017632e-07,
      "loss": 1.5005,
      "step": 1139
    },
    {
      "epoch": 0.25054945054945055,
      "grad_norm": 0.3517741283967574,
      "learning_rate": 6.073538241752547e-07,
      "loss": 1.4816,
      "step": 1140
    },
    {
      "epoch": 0.25076923076923074,
      "grad_norm": 0.26452520257594947,
      "learning_rate": 6.07199713844921e-07,
      "loss": 1.4731,
      "step": 1141
    },
    {
      "epoch": 0.250989010989011,
      "grad_norm": 0.25064377542417143,
      "learning_rate": 6.070454975842316e-07,
      "loss": 1.4347,
      "step": 1142
    },
    {
      "epoch": 0.2512087912087912,
      "grad_norm": 0.2542937350562004,
      "learning_rate": 6.068911754667068e-07,
      "loss": 1.4986,
      "step": 1143
    },
    {
      "epoch": 0.25142857142857145,
      "grad_norm": 0.2519101012965945,
      "learning_rate": 6.067367475659176e-07,
      "loss": 1.4823,
      "step": 1144
    },
    {
      "epoch": 0.25164835164835164,
      "grad_norm": 0.25828895174954947,
      "learning_rate": 6.065822139554849e-07,
      "loss": 1.4537,
      "step": 1145
    },
    {
      "epoch": 0.2518681318681319,
      "grad_norm": 0.24608984776859877,
      "learning_rate": 6.064275747090804e-07,
      "loss": 1.4811,
      "step": 1146
    },
    {
      "epoch": 0.2520879120879121,
      "grad_norm": 0.3663766605797682,
      "learning_rate": 6.06272829900426e-07,
      "loss": 1.4824,
      "step": 1147
    },
    {
      "epoch": 0.2523076923076923,
      "grad_norm": 0.24301132196881353,
      "learning_rate": 6.061179796032939e-07,
      "loss": 1.5269,
      "step": 1148
    },
    {
      "epoch": 0.25252747252747254,
      "grad_norm": 0.24987778657281365,
      "learning_rate": 6.059630238915066e-07,
      "loss": 1.4716,
      "step": 1149
    },
    {
      "epoch": 0.25274725274725274,
      "grad_norm": 0.25596706986657414,
      "learning_rate": 6.058079628389371e-07,
      "loss": 1.4299,
      "step": 1150
    },
    {
      "epoch": 0.252967032967033,
      "grad_norm": 0.24944191907968008,
      "learning_rate": 6.056527965195081e-07,
      "loss": 1.4441,
      "step": 1151
    },
    {
      "epoch": 0.2531868131868132,
      "grad_norm": 0.25309000759960226,
      "learning_rate": 6.05497525007193e-07,
      "loss": 1.5258,
      "step": 1152
    },
    {
      "epoch": 0.2534065934065934,
      "grad_norm": 0.24681891625672514,
      "learning_rate": 6.053421483760152e-07,
      "loss": 1.4661,
      "step": 1153
    },
    {
      "epoch": 0.25362637362637364,
      "grad_norm": 0.25797776885383356,
      "learning_rate": 6.05186666700048e-07,
      "loss": 1.4855,
      "step": 1154
    },
    {
      "epoch": 0.25384615384615383,
      "grad_norm": 0.251472266688807,
      "learning_rate": 6.050310800534152e-07,
      "loss": 1.4079,
      "step": 1155
    },
    {
      "epoch": 0.2540659340659341,
      "grad_norm": 0.25504410785351683,
      "learning_rate": 6.048753885102899e-07,
      "loss": 1.5241,
      "step": 1156
    },
    {
      "epoch": 0.2542857142857143,
      "grad_norm": 0.2659572613583344,
      "learning_rate": 6.047195921448962e-07,
      "loss": 1.5127,
      "step": 1157
    },
    {
      "epoch": 0.2545054945054945,
      "grad_norm": 0.24619056708730233,
      "learning_rate": 6.045636910315074e-07,
      "loss": 1.5726,
      "step": 1158
    },
    {
      "epoch": 0.25472527472527473,
      "grad_norm": 0.24478826490368982,
      "learning_rate": 6.044076852444471e-07,
      "loss": 1.4612,
      "step": 1159
    },
    {
      "epoch": 0.2549450549450549,
      "grad_norm": 0.29208441870154067,
      "learning_rate": 6.042515748580885e-07,
      "loss": 1.5348,
      "step": 1160
    },
    {
      "epoch": 0.2551648351648352,
      "grad_norm": 0.24538584809024377,
      "learning_rate": 6.040953599468552e-07,
      "loss": 1.5186,
      "step": 1161
    },
    {
      "epoch": 0.2553846153846154,
      "grad_norm": 0.25643685300585894,
      "learning_rate": 6.039390405852199e-07,
      "loss": 1.4723,
      "step": 1162
    },
    {
      "epoch": 0.25560439560439563,
      "grad_norm": 0.2547927574158795,
      "learning_rate": 6.03782616847706e-07,
      "loss": 1.3975,
      "step": 1163
    },
    {
      "epoch": 0.2558241758241758,
      "grad_norm": 0.25904888762465,
      "learning_rate": 6.036260888088857e-07,
      "loss": 1.486,
      "step": 1164
    },
    {
      "epoch": 0.256043956043956,
      "grad_norm": 0.2590186582179336,
      "learning_rate": 6.034694565433816e-07,
      "loss": 1.4909,
      "step": 1165
    },
    {
      "epoch": 0.2562637362637363,
      "grad_norm": 0.2514821084980522,
      "learning_rate": 6.033127201258656e-07,
      "loss": 1.4762,
      "step": 1166
    },
    {
      "epoch": 0.25648351648351647,
      "grad_norm": 0.25495314239700223,
      "learning_rate": 6.031558796310597e-07,
      "loss": 1.4726,
      "step": 1167
    },
    {
      "epoch": 0.2567032967032967,
      "grad_norm": 0.24749835721393953,
      "learning_rate": 6.029989351337348e-07,
      "loss": 1.4428,
      "step": 1168
    },
    {
      "epoch": 0.2569230769230769,
      "grad_norm": 0.2539325966425873,
      "learning_rate": 6.028418867087122e-07,
      "loss": 1.5169,
      "step": 1169
    },
    {
      "epoch": 0.2571428571428571,
      "grad_norm": 0.24306333874593364,
      "learning_rate": 6.026847344308623e-07,
      "loss": 1.5018,
      "step": 1170
    },
    {
      "epoch": 0.25736263736263737,
      "grad_norm": 0.24297781880095035,
      "learning_rate": 6.025274783751049e-07,
      "loss": 1.4832,
      "step": 1171
    },
    {
      "epoch": 0.25758241758241757,
      "grad_norm": 0.25627900845877255,
      "learning_rate": 6.023701186164098e-07,
      "loss": 1.4379,
      "step": 1172
    },
    {
      "epoch": 0.2578021978021978,
      "grad_norm": 0.24866316311975892,
      "learning_rate": 6.022126552297955e-07,
      "loss": 1.4023,
      "step": 1173
    },
    {
      "epoch": 0.258021978021978,
      "grad_norm": 0.24874008895024652,
      "learning_rate": 6.020550882903304e-07,
      "loss": 1.4933,
      "step": 1174
    },
    {
      "epoch": 0.25824175824175827,
      "grad_norm": 0.25464878977353284,
      "learning_rate": 6.018974178731322e-07,
      "loss": 1.4762,
      "step": 1175
    },
    {
      "epoch": 0.25846153846153846,
      "grad_norm": 0.24715180656412153,
      "learning_rate": 6.01739644053368e-07,
      "loss": 1.4166,
      "step": 1176
    },
    {
      "epoch": 0.25868131868131866,
      "grad_norm": 0.24382487413214948,
      "learning_rate": 6.015817669062539e-07,
      "loss": 1.4211,
      "step": 1177
    },
    {
      "epoch": 0.2589010989010989,
      "grad_norm": 0.26210602072240047,
      "learning_rate": 6.014237865070555e-07,
      "loss": 1.5183,
      "step": 1178
    },
    {
      "epoch": 0.2591208791208791,
      "grad_norm": 0.24780245259611958,
      "learning_rate": 6.012657029310876e-07,
      "loss": 1.4403,
      "step": 1179
    },
    {
      "epoch": 0.25934065934065936,
      "grad_norm": 0.23982738635279338,
      "learning_rate": 6.01107516253714e-07,
      "loss": 1.4952,
      "step": 1180
    },
    {
      "epoch": 0.25956043956043956,
      "grad_norm": 0.25313364511518943,
      "learning_rate": 6.009492265503482e-07,
      "loss": 1.472,
      "step": 1181
    },
    {
      "epoch": 0.25978021978021976,
      "grad_norm": 0.2523675011130071,
      "learning_rate": 6.00790833896452e-07,
      "loss": 1.5084,
      "step": 1182
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2423015383390281,
      "learning_rate": 6.006323383675369e-07,
      "loss": 1.4853,
      "step": 1183
    },
    {
      "epoch": 0.2602197802197802,
      "grad_norm": 0.26489790333011937,
      "learning_rate": 6.004737400391632e-07,
      "loss": 1.5596,
      "step": 1184
    },
    {
      "epoch": 0.26043956043956046,
      "grad_norm": 0.2418984060776012,
      "learning_rate": 6.003150389869404e-07,
      "loss": 1.4923,
      "step": 1185
    },
    {
      "epoch": 0.26065934065934065,
      "grad_norm": 0.2678745582741978,
      "learning_rate": 6.001562352865266e-07,
      "loss": 1.5146,
      "step": 1186
    },
    {
      "epoch": 0.26087912087912085,
      "grad_norm": 0.2520122665809618,
      "learning_rate": 5.999973290136292e-07,
      "loss": 1.4754,
      "step": 1187
    },
    {
      "epoch": 0.2610989010989011,
      "grad_norm": 0.248200953070769,
      "learning_rate": 5.998383202440043e-07,
      "loss": 1.5028,
      "step": 1188
    },
    {
      "epoch": 0.2613186813186813,
      "grad_norm": 0.24778067156462122,
      "learning_rate": 5.99679209053457e-07,
      "loss": 1.4667,
      "step": 1189
    },
    {
      "epoch": 0.26153846153846155,
      "grad_norm": 0.2722630605507682,
      "learning_rate": 5.99519995517841e-07,
      "loss": 1.5367,
      "step": 1190
    },
    {
      "epoch": 0.26175824175824175,
      "grad_norm": 0.2662416031120894,
      "learning_rate": 5.993606797130592e-07,
      "loss": 1.4845,
      "step": 1191
    },
    {
      "epoch": 0.261978021978022,
      "grad_norm": 0.2671806445602299,
      "learning_rate": 5.992012617150628e-07,
      "loss": 1.4437,
      "step": 1192
    },
    {
      "epoch": 0.2621978021978022,
      "grad_norm": 0.2524153192817131,
      "learning_rate": 5.99041741599852e-07,
      "loss": 1.4888,
      "step": 1193
    },
    {
      "epoch": 0.2624175824175824,
      "grad_norm": 0.24092041229443598,
      "learning_rate": 5.988821194434758e-07,
      "loss": 1.5152,
      "step": 1194
    },
    {
      "epoch": 0.26263736263736265,
      "grad_norm": 0.2568200801688485,
      "learning_rate": 5.987223953220311e-07,
      "loss": 1.4709,
      "step": 1195
    },
    {
      "epoch": 0.26285714285714284,
      "grad_norm": 0.2588652850938266,
      "learning_rate": 5.985625693116645e-07,
      "loss": 1.4862,
      "step": 1196
    },
    {
      "epoch": 0.2630769230769231,
      "grad_norm": 0.24963390531839405,
      "learning_rate": 5.984026414885704e-07,
      "loss": 1.4743,
      "step": 1197
    },
    {
      "epoch": 0.2632967032967033,
      "grad_norm": 0.2448874702757865,
      "learning_rate": 5.982426119289921e-07,
      "loss": 1.4444,
      "step": 1198
    },
    {
      "epoch": 0.2635164835164835,
      "grad_norm": 0.25669812321946955,
      "learning_rate": 5.980824807092211e-07,
      "loss": 1.47,
      "step": 1199
    },
    {
      "epoch": 0.26373626373626374,
      "grad_norm": 0.2421719234121139,
      "learning_rate": 5.979222479055976e-07,
      "loss": 1.4505,
      "step": 1200
    },
    {
      "epoch": 0.26395604395604394,
      "grad_norm": 0.27151723758011004,
      "learning_rate": 5.977619135945102e-07,
      "loss": 1.5311,
      "step": 1201
    },
    {
      "epoch": 0.2641758241758242,
      "grad_norm": 0.2542599355346271,
      "learning_rate": 5.976014778523959e-07,
      "loss": 1.5004,
      "step": 1202
    },
    {
      "epoch": 0.2643956043956044,
      "grad_norm": 0.2511556108189597,
      "learning_rate": 5.974409407557399e-07,
      "loss": 1.4541,
      "step": 1203
    },
    {
      "epoch": 0.26461538461538464,
      "grad_norm": 0.25551017572279605,
      "learning_rate": 5.97280302381076e-07,
      "loss": 1.44,
      "step": 1204
    },
    {
      "epoch": 0.26483516483516484,
      "grad_norm": 0.2553798629811734,
      "learning_rate": 5.971195628049859e-07,
      "loss": 1.4905,
      "step": 1205
    },
    {
      "epoch": 0.26505494505494503,
      "grad_norm": 0.2642029756555833,
      "learning_rate": 5.969587221041e-07,
      "loss": 1.4666,
      "step": 1206
    },
    {
      "epoch": 0.2652747252747253,
      "grad_norm": 0.2853320795960418,
      "learning_rate": 5.967977803550965e-07,
      "loss": 1.4327,
      "step": 1207
    },
    {
      "epoch": 0.2654945054945055,
      "grad_norm": 0.2405682250669,
      "learning_rate": 5.96636737634702e-07,
      "loss": 1.499,
      "step": 1208
    },
    {
      "epoch": 0.26571428571428574,
      "grad_norm": 0.24965478664594806,
      "learning_rate": 5.964755940196912e-07,
      "loss": 1.4309,
      "step": 1209
    },
    {
      "epoch": 0.26593406593406593,
      "grad_norm": 0.2477960693702363,
      "learning_rate": 5.96314349586887e-07,
      "loss": 1.4645,
      "step": 1210
    },
    {
      "epoch": 0.26615384615384613,
      "grad_norm": 0.25929811027231725,
      "learning_rate": 5.961530044131599e-07,
      "loss": 1.5014,
      "step": 1211
    },
    {
      "epoch": 0.2663736263736264,
      "grad_norm": 0.2696274977295813,
      "learning_rate": 5.95991558575429e-07,
      "loss": 1.5063,
      "step": 1212
    },
    {
      "epoch": 0.2665934065934066,
      "grad_norm": 0.24561139748105792,
      "learning_rate": 5.958300121506612e-07,
      "loss": 1.4739,
      "step": 1213
    },
    {
      "epoch": 0.26681318681318683,
      "grad_norm": 0.2549922263907234,
      "learning_rate": 5.956683652158714e-07,
      "loss": 1.4736,
      "step": 1214
    },
    {
      "epoch": 0.26703296703296703,
      "grad_norm": 0.27698993736333094,
      "learning_rate": 5.955066178481222e-07,
      "loss": 1.4859,
      "step": 1215
    },
    {
      "epoch": 0.2672527472527473,
      "grad_norm": 0.24492865588973883,
      "learning_rate": 5.95344770124524e-07,
      "loss": 1.4527,
      "step": 1216
    },
    {
      "epoch": 0.2674725274725275,
      "grad_norm": 0.24360866308103699,
      "learning_rate": 5.951828221222356e-07,
      "loss": 1.4194,
      "step": 1217
    },
    {
      "epoch": 0.2676923076923077,
      "grad_norm": 0.2422559900610527,
      "learning_rate": 5.950207739184632e-07,
      "loss": 1.4682,
      "step": 1218
    },
    {
      "epoch": 0.2679120879120879,
      "grad_norm": 0.2526136818543357,
      "learning_rate": 5.948586255904606e-07,
      "loss": 1.3901,
      "step": 1219
    },
    {
      "epoch": 0.2681318681318681,
      "grad_norm": 0.2453029494258157,
      "learning_rate": 5.9469637721553e-07,
      "loss": 1.45,
      "step": 1220
    },
    {
      "epoch": 0.2683516483516484,
      "grad_norm": 0.2448937162362778,
      "learning_rate": 5.945340288710202e-07,
      "loss": 1.4972,
      "step": 1221
    },
    {
      "epoch": 0.26857142857142857,
      "grad_norm": 0.248643964311628,
      "learning_rate": 5.943715806343291e-07,
      "loss": 1.4648,
      "step": 1222
    },
    {
      "epoch": 0.26879120879120877,
      "grad_norm": 0.2427660684241653,
      "learning_rate": 5.942090325829009e-07,
      "loss": 1.5058,
      "step": 1223
    },
    {
      "epoch": 0.269010989010989,
      "grad_norm": 0.3000472982461334,
      "learning_rate": 5.94046384794228e-07,
      "loss": 1.4523,
      "step": 1224
    },
    {
      "epoch": 0.2692307692307692,
      "grad_norm": 0.2510219939968679,
      "learning_rate": 5.938836373458505e-07,
      "loss": 1.5323,
      "step": 1225
    },
    {
      "epoch": 0.26945054945054947,
      "grad_norm": 0.25057047923701725,
      "learning_rate": 5.937207903153555e-07,
      "loss": 1.4894,
      "step": 1226
    },
    {
      "epoch": 0.26967032967032967,
      "grad_norm": 0.25916602966015984,
      "learning_rate": 5.935578437803782e-07,
      "loss": 1.4736,
      "step": 1227
    },
    {
      "epoch": 0.26989010989010986,
      "grad_norm": 0.2528863643291575,
      "learning_rate": 5.933947978186005e-07,
      "loss": 1.4519,
      "step": 1228
    },
    {
      "epoch": 0.2701098901098901,
      "grad_norm": 0.24951097085657922,
      "learning_rate": 5.932316525077523e-07,
      "loss": 1.5032,
      "step": 1229
    },
    {
      "epoch": 0.2703296703296703,
      "grad_norm": 0.2553958538809859,
      "learning_rate": 5.930684079256109e-07,
      "loss": 1.5227,
      "step": 1230
    },
    {
      "epoch": 0.27054945054945057,
      "grad_norm": 0.24481741189448933,
      "learning_rate": 5.929050641500002e-07,
      "loss": 1.4637,
      "step": 1231
    },
    {
      "epoch": 0.27076923076923076,
      "grad_norm": 0.33419297581145896,
      "learning_rate": 5.927416212587922e-07,
      "loss": 1.4128,
      "step": 1232
    },
    {
      "epoch": 0.270989010989011,
      "grad_norm": 0.2449685437448095,
      "learning_rate": 5.925780793299058e-07,
      "loss": 1.4575,
      "step": 1233
    },
    {
      "epoch": 0.2712087912087912,
      "grad_norm": 0.24577871034173174,
      "learning_rate": 5.924144384413071e-07,
      "loss": 1.4361,
      "step": 1234
    },
    {
      "epoch": 0.2714285714285714,
      "grad_norm": 0.26212461832473327,
      "learning_rate": 5.922506986710095e-07,
      "loss": 1.4193,
      "step": 1235
    },
    {
      "epoch": 0.27164835164835166,
      "grad_norm": 0.24295507349598566,
      "learning_rate": 5.920868600970731e-07,
      "loss": 1.4989,
      "step": 1236
    },
    {
      "epoch": 0.27186813186813186,
      "grad_norm": 0.24841010846308056,
      "learning_rate": 5.919229227976061e-07,
      "loss": 1.4311,
      "step": 1237
    },
    {
      "epoch": 0.2720879120879121,
      "grad_norm": 0.2728685246760535,
      "learning_rate": 5.917588868507625e-07,
      "loss": 1.4732,
      "step": 1238
    },
    {
      "epoch": 0.2723076923076923,
      "grad_norm": 0.241692791019892,
      "learning_rate": 5.915947523347445e-07,
      "loss": 1.5371,
      "step": 1239
    },
    {
      "epoch": 0.2725274725274725,
      "grad_norm": 0.26786092040486226,
      "learning_rate": 5.914305193278002e-07,
      "loss": 1.4705,
      "step": 1240
    },
    {
      "epoch": 0.27274725274725276,
      "grad_norm": 0.25055350348334465,
      "learning_rate": 5.912661879082256e-07,
      "loss": 1.556,
      "step": 1241
    },
    {
      "epoch": 0.27296703296703295,
      "grad_norm": 0.2559995740907689,
      "learning_rate": 5.91101758154363e-07,
      "loss": 1.4899,
      "step": 1242
    },
    {
      "epoch": 0.2731868131868132,
      "grad_norm": 0.247667585677873,
      "learning_rate": 5.909372301446021e-07,
      "loss": 1.4979,
      "step": 1243
    },
    {
      "epoch": 0.2734065934065934,
      "grad_norm": 0.24606259239820927,
      "learning_rate": 5.907726039573788e-07,
      "loss": 1.5447,
      "step": 1244
    },
    {
      "epoch": 0.27362637362637365,
      "grad_norm": 0.2574004117847577,
      "learning_rate": 5.906078796711764e-07,
      "loss": 1.4683,
      "step": 1245
    },
    {
      "epoch": 0.27384615384615385,
      "grad_norm": 0.24736881033961086,
      "learning_rate": 5.904430573645246e-07,
      "loss": 1.4195,
      "step": 1246
    },
    {
      "epoch": 0.27406593406593405,
      "grad_norm": 0.24530580572464508,
      "learning_rate": 5.902781371159999e-07,
      "loss": 1.4444,
      "step": 1247
    },
    {
      "epoch": 0.2742857142857143,
      "grad_norm": 0.34801615245113166,
      "learning_rate": 5.901131190042257e-07,
      "loss": 1.4543,
      "step": 1248
    },
    {
      "epoch": 0.2745054945054945,
      "grad_norm": 0.27425721649670726,
      "learning_rate": 5.899480031078717e-07,
      "loss": 1.4905,
      "step": 1249
    },
    {
      "epoch": 0.27472527472527475,
      "grad_norm": 0.24909979543101832,
      "learning_rate": 5.897827895056544e-07,
      "loss": 1.4948,
      "step": 1250
    },
    {
      "epoch": 0.27494505494505495,
      "grad_norm": 0.2453928589008479,
      "learning_rate": 5.896174782763371e-07,
      "loss": 1.4192,
      "step": 1251
    },
    {
      "epoch": 0.27516483516483514,
      "grad_norm": 0.24784211470781856,
      "learning_rate": 5.894520694987292e-07,
      "loss": 1.5016,
      "step": 1252
    },
    {
      "epoch": 0.2753846153846154,
      "grad_norm": 0.25385193474328976,
      "learning_rate": 5.89286563251687e-07,
      "loss": 1.4749,
      "step": 1253
    },
    {
      "epoch": 0.2756043956043956,
      "grad_norm": 0.2500375390716796,
      "learning_rate": 5.891209596141132e-07,
      "loss": 1.4689,
      "step": 1254
    },
    {
      "epoch": 0.27582417582417584,
      "grad_norm": 0.25183689524162134,
      "learning_rate": 5.889552586649564e-07,
      "loss": 1.4391,
      "step": 1255
    },
    {
      "epoch": 0.27604395604395604,
      "grad_norm": 0.25229187969337635,
      "learning_rate": 5.887894604832125e-07,
      "loss": 1.4815,
      "step": 1256
    },
    {
      "epoch": 0.27626373626373624,
      "grad_norm": 0.24928781380181728,
      "learning_rate": 5.88623565147923e-07,
      "loss": 1.4695,
      "step": 1257
    },
    {
      "epoch": 0.2764835164835165,
      "grad_norm": 0.2562942501214982,
      "learning_rate": 5.884575727381761e-07,
      "loss": 1.4697,
      "step": 1258
    },
    {
      "epoch": 0.2767032967032967,
      "grad_norm": 0.24559651092598303,
      "learning_rate": 5.882914833331062e-07,
      "loss": 1.4488,
      "step": 1259
    },
    {
      "epoch": 0.27692307692307694,
      "grad_norm": 0.35499717285209875,
      "learning_rate": 5.881252970118937e-07,
      "loss": 1.4326,
      "step": 1260
    },
    {
      "epoch": 0.27714285714285714,
      "grad_norm": 0.278990004685268,
      "learning_rate": 5.879590138537657e-07,
      "loss": 1.4719,
      "step": 1261
    },
    {
      "epoch": 0.2773626373626374,
      "grad_norm": 0.40140864415897365,
      "learning_rate": 5.877926339379949e-07,
      "loss": 1.3861,
      "step": 1262
    },
    {
      "epoch": 0.2775824175824176,
      "grad_norm": 0.29577977795576144,
      "learning_rate": 5.876261573439008e-07,
      "loss": 1.4429,
      "step": 1263
    },
    {
      "epoch": 0.2778021978021978,
      "grad_norm": 0.25780139038569383,
      "learning_rate": 5.874595841508483e-07,
      "loss": 1.492,
      "step": 1264
    },
    {
      "epoch": 0.27802197802197803,
      "grad_norm": 0.25949464208174866,
      "learning_rate": 5.872929144382486e-07,
      "loss": 1.4194,
      "step": 1265
    },
    {
      "epoch": 0.27824175824175823,
      "grad_norm": 0.23896647886805938,
      "learning_rate": 5.871261482855592e-07,
      "loss": 1.4672,
      "step": 1266
    },
    {
      "epoch": 0.2784615384615385,
      "grad_norm": 0.29873658112030305,
      "learning_rate": 5.869592857722832e-07,
      "loss": 1.5212,
      "step": 1267
    },
    {
      "epoch": 0.2786813186813187,
      "grad_norm": 0.24995422539987305,
      "learning_rate": 5.867923269779698e-07,
      "loss": 1.5121,
      "step": 1268
    },
    {
      "epoch": 0.2789010989010989,
      "grad_norm": 0.24988830906918102,
      "learning_rate": 5.866252719822142e-07,
      "loss": 1.5253,
      "step": 1269
    },
    {
      "epoch": 0.27912087912087913,
      "grad_norm": 0.2822070269451855,
      "learning_rate": 5.864581208646571e-07,
      "loss": 1.4793,
      "step": 1270
    },
    {
      "epoch": 0.2793406593406593,
      "grad_norm": 0.24384360636722566,
      "learning_rate": 5.862908737049855e-07,
      "loss": 1.409,
      "step": 1271
    },
    {
      "epoch": 0.2795604395604396,
      "grad_norm": 0.3010815653589461,
      "learning_rate": 5.861235305829318e-07,
      "loss": 1.4737,
      "step": 1272
    },
    {
      "epoch": 0.2797802197802198,
      "grad_norm": 0.2426473360780779,
      "learning_rate": 5.859560915782743e-07,
      "loss": 1.4597,
      "step": 1273
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.24031709265797813,
      "learning_rate": 5.857885567708372e-07,
      "loss": 1.4473,
      "step": 1274
    },
    {
      "epoch": 0.2802197802197802,
      "grad_norm": 0.24887282661998375,
      "learning_rate": 5.856209262404901e-07,
      "loss": 1.5392,
      "step": 1275
    },
    {
      "epoch": 0.2804395604395604,
      "grad_norm": 0.33045530720916494,
      "learning_rate": 5.854532000671482e-07,
      "loss": 1.4809,
      "step": 1276
    },
    {
      "epoch": 0.2806593406593407,
      "grad_norm": 0.2598270734406072,
      "learning_rate": 5.852853783307725e-07,
      "loss": 1.5037,
      "step": 1277
    },
    {
      "epoch": 0.28087912087912087,
      "grad_norm": 0.2599588112358863,
      "learning_rate": 5.851174611113696e-07,
      "loss": 1.5084,
      "step": 1278
    },
    {
      "epoch": 0.2810989010989011,
      "grad_norm": 0.2603727300609368,
      "learning_rate": 5.849494484889913e-07,
      "loss": 1.4728,
      "step": 1279
    },
    {
      "epoch": 0.2813186813186813,
      "grad_norm": 0.25999787393649765,
      "learning_rate": 5.847813405437353e-07,
      "loss": 1.4216,
      "step": 1280
    },
    {
      "epoch": 0.2815384615384615,
      "grad_norm": 0.24769042216447004,
      "learning_rate": 5.846131373557442e-07,
      "loss": 1.4689,
      "step": 1281
    },
    {
      "epoch": 0.28175824175824177,
      "grad_norm": 0.2540554170799448,
      "learning_rate": 5.844448390052066e-07,
      "loss": 1.4834,
      "step": 1282
    },
    {
      "epoch": 0.28197802197802196,
      "grad_norm": 0.26568707123092805,
      "learning_rate": 5.84276445572356e-07,
      "loss": 1.4566,
      "step": 1283
    },
    {
      "epoch": 0.2821978021978022,
      "grad_norm": 0.26055271950319026,
      "learning_rate": 5.841079571374717e-07,
      "loss": 1.5031,
      "step": 1284
    },
    {
      "epoch": 0.2824175824175824,
      "grad_norm": 0.2452788727718544,
      "learning_rate": 5.839393737808778e-07,
      "loss": 1.474,
      "step": 1285
    },
    {
      "epoch": 0.2826373626373626,
      "grad_norm": 0.24755361748064456,
      "learning_rate": 5.83770695582944e-07,
      "loss": 1.4287,
      "step": 1286
    },
    {
      "epoch": 0.28285714285714286,
      "grad_norm": 0.408446084926422,
      "learning_rate": 5.836019226240848e-07,
      "loss": 1.4857,
      "step": 1287
    },
    {
      "epoch": 0.28307692307692306,
      "grad_norm": 0.24802297923693442,
      "learning_rate": 5.834330549847604e-07,
      "loss": 1.4467,
      "step": 1288
    },
    {
      "epoch": 0.2832967032967033,
      "grad_norm": 0.2456449266928603,
      "learning_rate": 5.832640927454759e-07,
      "loss": 1.4625,
      "step": 1289
    },
    {
      "epoch": 0.2835164835164835,
      "grad_norm": 0.24884916776155724,
      "learning_rate": 5.830950359867815e-07,
      "loss": 1.4963,
      "step": 1290
    },
    {
      "epoch": 0.28373626373626376,
      "grad_norm": 0.2525680171064067,
      "learning_rate": 5.829258847892723e-07,
      "loss": 1.4822,
      "step": 1291
    },
    {
      "epoch": 0.28395604395604396,
      "grad_norm": 0.24423421079466054,
      "learning_rate": 5.827566392335887e-07,
      "loss": 1.447,
      "step": 1292
    },
    {
      "epoch": 0.28417582417582415,
      "grad_norm": 0.269983032180082,
      "learning_rate": 5.82587299400416e-07,
      "loss": 1.5456,
      "step": 1293
    },
    {
      "epoch": 0.2843956043956044,
      "grad_norm": 0.2559594861722576,
      "learning_rate": 5.824178653704844e-07,
      "loss": 1.4456,
      "step": 1294
    },
    {
      "epoch": 0.2846153846153846,
      "grad_norm": 0.2436519307438797,
      "learning_rate": 5.822483372245688e-07,
      "loss": 1.5017,
      "step": 1295
    },
    {
      "epoch": 0.28483516483516486,
      "grad_norm": 0.2520568394304604,
      "learning_rate": 5.820787150434893e-07,
      "loss": 1.4413,
      "step": 1296
    },
    {
      "epoch": 0.28505494505494505,
      "grad_norm": 0.24142164071110692,
      "learning_rate": 5.819089989081109e-07,
      "loss": 1.481,
      "step": 1297
    },
    {
      "epoch": 0.28527472527472525,
      "grad_norm": 0.25596033369892546,
      "learning_rate": 5.817391888993431e-07,
      "loss": 1.4318,
      "step": 1298
    },
    {
      "epoch": 0.2854945054945055,
      "grad_norm": 0.2544386478012293,
      "learning_rate": 5.8156928509814e-07,
      "loss": 1.4276,
      "step": 1299
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.25212373912829383,
      "learning_rate": 5.813992875855011e-07,
      "loss": 1.4799,
      "step": 1300
    },
    {
      "epoch": 0.28593406593406595,
      "grad_norm": 0.25407121739201727,
      "learning_rate": 5.812291964424698e-07,
      "loss": 1.4665,
      "step": 1301
    },
    {
      "epoch": 0.28615384615384615,
      "grad_norm": 0.2511055574424577,
      "learning_rate": 5.810590117501346e-07,
      "loss": 1.4749,
      "step": 1302
    },
    {
      "epoch": 0.2863736263736264,
      "grad_norm": 0.2772291435018925,
      "learning_rate": 5.808887335896284e-07,
      "loss": 1.534,
      "step": 1303
    },
    {
      "epoch": 0.2865934065934066,
      "grad_norm": 0.248031946299549,
      "learning_rate": 5.80718362042129e-07,
      "loss": 1.4937,
      "step": 1304
    },
    {
      "epoch": 0.2868131868131868,
      "grad_norm": 0.25396776910396807,
      "learning_rate": 5.805478971888583e-07,
      "loss": 1.4854,
      "step": 1305
    },
    {
      "epoch": 0.28703296703296705,
      "grad_norm": 0.24410515270199715,
      "learning_rate": 5.803773391110826e-07,
      "loss": 1.4456,
      "step": 1306
    },
    {
      "epoch": 0.28725274725274724,
      "grad_norm": 0.2770756880635025,
      "learning_rate": 5.802066878901133e-07,
      "loss": 1.4769,
      "step": 1307
    },
    {
      "epoch": 0.2874725274725275,
      "grad_norm": 0.25348000723592246,
      "learning_rate": 5.800359436073057e-07,
      "loss": 1.4376,
      "step": 1308
    },
    {
      "epoch": 0.2876923076923077,
      "grad_norm": 0.2484354440295395,
      "learning_rate": 5.798651063440592e-07,
      "loss": 1.4513,
      "step": 1309
    },
    {
      "epoch": 0.2879120879120879,
      "grad_norm": 0.2517249240776544,
      "learning_rate": 5.796941761818182e-07,
      "loss": 1.4351,
      "step": 1310
    },
    {
      "epoch": 0.28813186813186814,
      "grad_norm": 0.2504832272346342,
      "learning_rate": 5.795231532020711e-07,
      "loss": 1.4538,
      "step": 1311
    },
    {
      "epoch": 0.28835164835164834,
      "grad_norm": 0.3019003841272291,
      "learning_rate": 5.793520374863504e-07,
      "loss": 1.5249,
      "step": 1312
    },
    {
      "epoch": 0.2885714285714286,
      "grad_norm": 0.2630909033801328,
      "learning_rate": 5.791808291162331e-07,
      "loss": 1.5838,
      "step": 1313
    },
    {
      "epoch": 0.2887912087912088,
      "grad_norm": 0.24143116292557143,
      "learning_rate": 5.790095281733399e-07,
      "loss": 1.4717,
      "step": 1314
    },
    {
      "epoch": 0.289010989010989,
      "grad_norm": 0.25895879947502787,
      "learning_rate": 5.788381347393361e-07,
      "loss": 1.5802,
      "step": 1315
    },
    {
      "epoch": 0.28923076923076924,
      "grad_norm": 0.24263969650420297,
      "learning_rate": 5.78666648895931e-07,
      "loss": 1.5295,
      "step": 1316
    },
    {
      "epoch": 0.28945054945054943,
      "grad_norm": 0.2418927429342586,
      "learning_rate": 5.784950707248779e-07,
      "loss": 1.4659,
      "step": 1317
    },
    {
      "epoch": 0.2896703296703297,
      "grad_norm": 0.3974036191577161,
      "learning_rate": 5.783234003079739e-07,
      "loss": 1.4877,
      "step": 1318
    },
    {
      "epoch": 0.2898901098901099,
      "grad_norm": 0.24524436489741314,
      "learning_rate": 5.781516377270604e-07,
      "loss": 1.4382,
      "step": 1319
    },
    {
      "epoch": 0.29010989010989013,
      "grad_norm": 0.2851478272657233,
      "learning_rate": 5.779797830640227e-07,
      "loss": 1.5369,
      "step": 1320
    },
    {
      "epoch": 0.29032967032967033,
      "grad_norm": 0.2503039752750346,
      "learning_rate": 5.778078364007899e-07,
      "loss": 1.4435,
      "step": 1321
    },
    {
      "epoch": 0.2905494505494505,
      "grad_norm": 0.24884356273385935,
      "learning_rate": 5.776357978193347e-07,
      "loss": 1.4543,
      "step": 1322
    },
    {
      "epoch": 0.2907692307692308,
      "grad_norm": 0.25888769163782055,
      "learning_rate": 5.774636674016741e-07,
      "loss": 1.4053,
      "step": 1323
    },
    {
      "epoch": 0.290989010989011,
      "grad_norm": 0.2462852035495459,
      "learning_rate": 5.772914452298688e-07,
      "loss": 1.4834,
      "step": 1324
    },
    {
      "epoch": 0.29120879120879123,
      "grad_norm": 0.25411395270956194,
      "learning_rate": 5.771191313860229e-07,
      "loss": 1.5166,
      "step": 1325
    },
    {
      "epoch": 0.2914285714285714,
      "grad_norm": 0.2513998510012807,
      "learning_rate": 5.769467259522846e-07,
      "loss": 1.4944,
      "step": 1326
    },
    {
      "epoch": 0.2916483516483516,
      "grad_norm": 0.43988168100265335,
      "learning_rate": 5.767742290108453e-07,
      "loss": 1.4793,
      "step": 1327
    },
    {
      "epoch": 0.2918681318681319,
      "grad_norm": 0.2534106539853273,
      "learning_rate": 5.766016406439406e-07,
      "loss": 1.4667,
      "step": 1328
    },
    {
      "epoch": 0.29208791208791207,
      "grad_norm": 0.2394330470229594,
      "learning_rate": 5.764289609338495e-07,
      "loss": 1.4811,
      "step": 1329
    },
    {
      "epoch": 0.2923076923076923,
      "grad_norm": 0.2585764308429949,
      "learning_rate": 5.76256189962894e-07,
      "loss": 1.4908,
      "step": 1330
    },
    {
      "epoch": 0.2925274725274725,
      "grad_norm": 0.2416542791979125,
      "learning_rate": 5.760833278134404e-07,
      "loss": 1.5639,
      "step": 1331
    },
    {
      "epoch": 0.2927472527472528,
      "grad_norm": 0.24692480043414214,
      "learning_rate": 5.759103745678979e-07,
      "loss": 1.4709,
      "step": 1332
    },
    {
      "epoch": 0.29296703296703297,
      "grad_norm": 0.3458281045720652,
      "learning_rate": 5.757373303087195e-07,
      "loss": 1.4075,
      "step": 1333
    },
    {
      "epoch": 0.29318681318681317,
      "grad_norm": 0.25017970446322996,
      "learning_rate": 5.755641951184015e-07,
      "loss": 1.4471,
      "step": 1334
    },
    {
      "epoch": 0.2934065934065934,
      "grad_norm": 0.2573881075194735,
      "learning_rate": 5.753909690794832e-07,
      "loss": 1.4943,
      "step": 1335
    },
    {
      "epoch": 0.2936263736263736,
      "grad_norm": 0.24695784811193833,
      "learning_rate": 5.752176522745478e-07,
      "loss": 1.4421,
      "step": 1336
    },
    {
      "epoch": 0.29384615384615387,
      "grad_norm": 0.24586543625747756,
      "learning_rate": 5.750442447862213e-07,
      "loss": 1.4503,
      "step": 1337
    },
    {
      "epoch": 0.29406593406593406,
      "grad_norm": 0.24646862660755073,
      "learning_rate": 5.748707466971731e-07,
      "loss": 1.4443,
      "step": 1338
    },
    {
      "epoch": 0.29428571428571426,
      "grad_norm": 0.26282677786531006,
      "learning_rate": 5.746971580901159e-07,
      "loss": 1.4756,
      "step": 1339
    },
    {
      "epoch": 0.2945054945054945,
      "grad_norm": 0.261353131292857,
      "learning_rate": 5.745234790478053e-07,
      "loss": 1.4198,
      "step": 1340
    },
    {
      "epoch": 0.2947252747252747,
      "grad_norm": 0.24357210900308515,
      "learning_rate": 5.743497096530402e-07,
      "loss": 1.5065,
      "step": 1341
    },
    {
      "epoch": 0.29494505494505496,
      "grad_norm": 0.2517316952398751,
      "learning_rate": 5.741758499886627e-07,
      "loss": 1.5051,
      "step": 1342
    },
    {
      "epoch": 0.29516483516483516,
      "grad_norm": 0.2510808672786306,
      "learning_rate": 5.740019001375576e-07,
      "loss": 1.4981,
      "step": 1343
    },
    {
      "epoch": 0.2953846153846154,
      "grad_norm": 0.25127079169990896,
      "learning_rate": 5.738278601826528e-07,
      "loss": 1.5269,
      "step": 1344
    },
    {
      "epoch": 0.2956043956043956,
      "grad_norm": 0.2584965663841474,
      "learning_rate": 5.736537302069193e-07,
      "loss": 1.5491,
      "step": 1345
    },
    {
      "epoch": 0.2958241758241758,
      "grad_norm": 0.259841902528515,
      "learning_rate": 5.734795102933711e-07,
      "loss": 1.4774,
      "step": 1346
    },
    {
      "epoch": 0.29604395604395606,
      "grad_norm": 0.26010438725480944,
      "learning_rate": 5.733052005250646e-07,
      "loss": 1.4402,
      "step": 1347
    },
    {
      "epoch": 0.29626373626373625,
      "grad_norm": 0.2504879713131675,
      "learning_rate": 5.731308009850997e-07,
      "loss": 1.4833,
      "step": 1348
    },
    {
      "epoch": 0.2964835164835165,
      "grad_norm": 0.2509677515197842,
      "learning_rate": 5.729563117566183e-07,
      "loss": 1.5006,
      "step": 1349
    },
    {
      "epoch": 0.2967032967032967,
      "grad_norm": 0.24838804541143591,
      "learning_rate": 5.727817329228059e-07,
      "loss": 1.464,
      "step": 1350
    },
    {
      "epoch": 0.2969230769230769,
      "grad_norm": 0.24959313046978843,
      "learning_rate": 5.726070645668903e-07,
      "loss": 1.4541,
      "step": 1351
    },
    {
      "epoch": 0.29714285714285715,
      "grad_norm": 0.26537308150910344,
      "learning_rate": 5.724323067721417e-07,
      "loss": 1.4479,
      "step": 1352
    },
    {
      "epoch": 0.29736263736263735,
      "grad_norm": 0.2509782256447158,
      "learning_rate": 5.722574596218733e-07,
      "loss": 1.4574,
      "step": 1353
    },
    {
      "epoch": 0.2975824175824176,
      "grad_norm": 0.2659576892006521,
      "learning_rate": 5.720825231994411e-07,
      "loss": 1.4676,
      "step": 1354
    },
    {
      "epoch": 0.2978021978021978,
      "grad_norm": 0.25523298060847843,
      "learning_rate": 5.719074975882435e-07,
      "loss": 1.439,
      "step": 1355
    },
    {
      "epoch": 0.298021978021978,
      "grad_norm": 0.2775273273573927,
      "learning_rate": 5.717323828717207e-07,
      "loss": 1.4567,
      "step": 1356
    },
    {
      "epoch": 0.29824175824175825,
      "grad_norm": 0.25879672991156294,
      "learning_rate": 5.715571791333565e-07,
      "loss": 1.5381,
      "step": 1357
    },
    {
      "epoch": 0.29846153846153844,
      "grad_norm": 0.24054171393461082,
      "learning_rate": 5.713818864566764e-07,
      "loss": 1.4311,
      "step": 1358
    },
    {
      "epoch": 0.2986813186813187,
      "grad_norm": 0.25140462321818235,
      "learning_rate": 5.712065049252486e-07,
      "loss": 1.5125,
      "step": 1359
    },
    {
      "epoch": 0.2989010989010989,
      "grad_norm": 0.2626765262508115,
      "learning_rate": 5.710310346226838e-07,
      "loss": 1.4802,
      "step": 1360
    },
    {
      "epoch": 0.29912087912087915,
      "grad_norm": 0.25202646435155157,
      "learning_rate": 5.708554756326345e-07,
      "loss": 1.4718,
      "step": 1361
    },
    {
      "epoch": 0.29934065934065934,
      "grad_norm": 0.24388182954928223,
      "learning_rate": 5.70679828038796e-07,
      "loss": 1.4558,
      "step": 1362
    },
    {
      "epoch": 0.29956043956043954,
      "grad_norm": 0.2414756659678707,
      "learning_rate": 5.705040919249056e-07,
      "loss": 1.4651,
      "step": 1363
    },
    {
      "epoch": 0.2997802197802198,
      "grad_norm": 0.24735696929962378,
      "learning_rate": 5.703282673747426e-07,
      "loss": 1.5122,
      "step": 1364
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2792398861015185,
      "learning_rate": 5.70152354472129e-07,
      "loss": 1.4243,
      "step": 1365
    },
    {
      "epoch": 0.30021978021978024,
      "grad_norm": 0.2581459142568791,
      "learning_rate": 5.699763533009285e-07,
      "loss": 1.4987,
      "step": 1366
    },
    {
      "epoch": 0.30043956043956044,
      "grad_norm": 0.3093871599270535,
      "learning_rate": 5.698002639450471e-07,
      "loss": 1.4171,
      "step": 1367
    },
    {
      "epoch": 0.30065934065934063,
      "grad_norm": 0.2572564961720901,
      "learning_rate": 5.696240864884327e-07,
      "loss": 1.4722,
      "step": 1368
    },
    {
      "epoch": 0.3008791208791209,
      "grad_norm": 0.262863970873206,
      "learning_rate": 5.69447821015075e-07,
      "loss": 1.5212,
      "step": 1369
    },
    {
      "epoch": 0.3010989010989011,
      "grad_norm": 0.26222057362657936,
      "learning_rate": 5.692714676090062e-07,
      "loss": 1.5192,
      "step": 1370
    },
    {
      "epoch": 0.30131868131868134,
      "grad_norm": 0.3071417614804383,
      "learning_rate": 5.690950263543e-07,
      "loss": 1.5249,
      "step": 1371
    },
    {
      "epoch": 0.30153846153846153,
      "grad_norm": 0.25010627254354717,
      "learning_rate": 5.689184973350722e-07,
      "loss": 1.4781,
      "step": 1372
    },
    {
      "epoch": 0.3017582417582418,
      "grad_norm": 0.2553041392298195,
      "learning_rate": 5.6874188063548e-07,
      "loss": 1.4998,
      "step": 1373
    },
    {
      "epoch": 0.301978021978022,
      "grad_norm": 0.2639407112810674,
      "learning_rate": 5.685651763397232e-07,
      "loss": 1.4387,
      "step": 1374
    },
    {
      "epoch": 0.3021978021978022,
      "grad_norm": 0.249452213087093,
      "learning_rate": 5.683883845320428e-07,
      "loss": 1.4466,
      "step": 1375
    },
    {
      "epoch": 0.30241758241758243,
      "grad_norm": 0.2545854400749403,
      "learning_rate": 5.682115052967213e-07,
      "loss": 1.5069,
      "step": 1376
    },
    {
      "epoch": 0.30263736263736263,
      "grad_norm": 0.2721401982511904,
      "learning_rate": 5.680345387180835e-07,
      "loss": 1.4467,
      "step": 1377
    },
    {
      "epoch": 0.3028571428571429,
      "grad_norm": 0.26793985557990907,
      "learning_rate": 5.678574848804957e-07,
      "loss": 1.4538,
      "step": 1378
    },
    {
      "epoch": 0.3030769230769231,
      "grad_norm": 0.2822194318720025,
      "learning_rate": 5.676803438683652e-07,
      "loss": 1.5214,
      "step": 1379
    },
    {
      "epoch": 0.3032967032967033,
      "grad_norm": 0.23572713357279437,
      "learning_rate": 5.675031157661414e-07,
      "loss": 1.4715,
      "step": 1380
    },
    {
      "epoch": 0.3035164835164835,
      "grad_norm": 0.26595924002700233,
      "learning_rate": 5.673258006583154e-07,
      "loss": 1.4978,
      "step": 1381
    },
    {
      "epoch": 0.3037362637362637,
      "grad_norm": 0.23755624844769707,
      "learning_rate": 5.671483986294193e-07,
      "loss": 1.4383,
      "step": 1382
    },
    {
      "epoch": 0.303956043956044,
      "grad_norm": 0.30137774105205695,
      "learning_rate": 5.669709097640269e-07,
      "loss": 1.5048,
      "step": 1383
    },
    {
      "epoch": 0.30417582417582417,
      "grad_norm": 0.3534728549437255,
      "learning_rate": 5.667933341467533e-07,
      "loss": 1.4428,
      "step": 1384
    },
    {
      "epoch": 0.30439560439560437,
      "grad_norm": 0.2560030593834548,
      "learning_rate": 5.66615671862255e-07,
      "loss": 1.4715,
      "step": 1385
    },
    {
      "epoch": 0.3046153846153846,
      "grad_norm": 0.2506556454171206,
      "learning_rate": 5.664379229952297e-07,
      "loss": 1.4632,
      "step": 1386
    },
    {
      "epoch": 0.3048351648351648,
      "grad_norm": 0.2472721096073287,
      "learning_rate": 5.662600876304167e-07,
      "loss": 1.4685,
      "step": 1387
    },
    {
      "epoch": 0.30505494505494507,
      "grad_norm": 0.42497518640623017,
      "learning_rate": 5.66082165852596e-07,
      "loss": 1.4645,
      "step": 1388
    },
    {
      "epoch": 0.30527472527472527,
      "grad_norm": 0.2449598067042146,
      "learning_rate": 5.659041577465897e-07,
      "loss": 1.4671,
      "step": 1389
    },
    {
      "epoch": 0.3054945054945055,
      "grad_norm": 0.2628152734817873,
      "learning_rate": 5.657260633972598e-07,
      "loss": 1.4671,
      "step": 1390
    },
    {
      "epoch": 0.3057142857142857,
      "grad_norm": 0.2502257772986754,
      "learning_rate": 5.655478828895106e-07,
      "loss": 1.4535,
      "step": 1391
    },
    {
      "epoch": 0.3059340659340659,
      "grad_norm": 0.25471252931559857,
      "learning_rate": 5.653696163082866e-07,
      "loss": 1.4988,
      "step": 1392
    },
    {
      "epoch": 0.30615384615384617,
      "grad_norm": 0.24385797600271722,
      "learning_rate": 5.651912637385738e-07,
      "loss": 1.4096,
      "step": 1393
    },
    {
      "epoch": 0.30637362637362636,
      "grad_norm": 0.25676546116826904,
      "learning_rate": 5.650128252653993e-07,
      "loss": 1.4715,
      "step": 1394
    },
    {
      "epoch": 0.3065934065934066,
      "grad_norm": 0.24784394586038405,
      "learning_rate": 5.648343009738307e-07,
      "loss": 1.48,
      "step": 1395
    },
    {
      "epoch": 0.3068131868131868,
      "grad_norm": 0.2457472609882082,
      "learning_rate": 5.646556909489769e-07,
      "loss": 1.5201,
      "step": 1396
    },
    {
      "epoch": 0.307032967032967,
      "grad_norm": 0.2458655155769294,
      "learning_rate": 5.644769952759875e-07,
      "loss": 1.4737,
      "step": 1397
    },
    {
      "epoch": 0.30725274725274726,
      "grad_norm": 0.25223849576167445,
      "learning_rate": 5.642982140400529e-07,
      "loss": 1.5003,
      "step": 1398
    },
    {
      "epoch": 0.30747252747252746,
      "grad_norm": 0.2552847391712624,
      "learning_rate": 5.641193473264045e-07,
      "loss": 1.4565,
      "step": 1399
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 0.2459471016037614,
      "learning_rate": 5.63940395220314e-07,
      "loss": 1.4169,
      "step": 1400
    },
    {
      "epoch": 0.3079120879120879,
      "grad_norm": 0.2483389831046875,
      "learning_rate": 5.637613578070945e-07,
      "loss": 1.4946,
      "step": 1401
    },
    {
      "epoch": 0.30813186813186816,
      "grad_norm": 0.2633400739347006,
      "learning_rate": 5.635822351720991e-07,
      "loss": 1.4627,
      "step": 1402
    },
    {
      "epoch": 0.30835164835164836,
      "grad_norm": 0.258757752980863,
      "learning_rate": 5.634030274007219e-07,
      "loss": 1.4718,
      "step": 1403
    },
    {
      "epoch": 0.30857142857142855,
      "grad_norm": 0.5070997448693416,
      "learning_rate": 5.632237345783975e-07,
      "loss": 1.4994,
      "step": 1404
    },
    {
      "epoch": 0.3087912087912088,
      "grad_norm": 0.30790057285886036,
      "learning_rate": 5.630443567906011e-07,
      "loss": 1.4706,
      "step": 1405
    },
    {
      "epoch": 0.309010989010989,
      "grad_norm": 0.2631032173876283,
      "learning_rate": 5.628648941228482e-07,
      "loss": 1.4788,
      "step": 1406
    },
    {
      "epoch": 0.30923076923076925,
      "grad_norm": 0.5276387218909935,
      "learning_rate": 5.626853466606951e-07,
      "loss": 1.458,
      "step": 1407
    },
    {
      "epoch": 0.30945054945054945,
      "grad_norm": 0.24859726884660674,
      "learning_rate": 5.625057144897381e-07,
      "loss": 1.4666,
      "step": 1408
    },
    {
      "epoch": 0.30967032967032965,
      "grad_norm": 0.25281773514583455,
      "learning_rate": 5.623259976956144e-07,
      "loss": 1.4884,
      "step": 1409
    },
    {
      "epoch": 0.3098901098901099,
      "grad_norm": 0.24749271519094207,
      "learning_rate": 5.621461963640012e-07,
      "loss": 1.5311,
      "step": 1410
    },
    {
      "epoch": 0.3101098901098901,
      "grad_norm": 0.25232121505427474,
      "learning_rate": 5.619663105806159e-07,
      "loss": 1.4901,
      "step": 1411
    },
    {
      "epoch": 0.31032967032967035,
      "grad_norm": 0.2434137284776949,
      "learning_rate": 5.617863404312165e-07,
      "loss": 1.5244,
      "step": 1412
    },
    {
      "epoch": 0.31054945054945055,
      "grad_norm": 0.26079053162019655,
      "learning_rate": 5.616062860016009e-07,
      "loss": 1.4994,
      "step": 1413
    },
    {
      "epoch": 0.31076923076923074,
      "grad_norm": 0.2504660744024973,
      "learning_rate": 5.614261473776074e-07,
      "loss": 1.4593,
      "step": 1414
    },
    {
      "epoch": 0.310989010989011,
      "grad_norm": 0.2449194007940803,
      "learning_rate": 5.612459246451143e-07,
      "loss": 1.4826,
      "step": 1415
    },
    {
      "epoch": 0.3112087912087912,
      "grad_norm": 0.24300282511180893,
      "learning_rate": 5.610656178900403e-07,
      "loss": 1.4438,
      "step": 1416
    },
    {
      "epoch": 0.31142857142857144,
      "grad_norm": 0.24648580582833773,
      "learning_rate": 5.608852271983437e-07,
      "loss": 1.4502,
      "step": 1417
    },
    {
      "epoch": 0.31164835164835164,
      "grad_norm": 0.2523865024641673,
      "learning_rate": 5.607047526560231e-07,
      "loss": 1.4736,
      "step": 1418
    },
    {
      "epoch": 0.3118681318681319,
      "grad_norm": 0.25679243551564807,
      "learning_rate": 5.605241943491169e-07,
      "loss": 1.4476,
      "step": 1419
    },
    {
      "epoch": 0.3120879120879121,
      "grad_norm": 0.263785746954739,
      "learning_rate": 5.603435523637038e-07,
      "loss": 1.4684,
      "step": 1420
    },
    {
      "epoch": 0.3123076923076923,
      "grad_norm": 0.24918462298015576,
      "learning_rate": 5.601628267859018e-07,
      "loss": 1.4207,
      "step": 1421
    },
    {
      "epoch": 0.31252747252747254,
      "grad_norm": 0.23788222017014576,
      "learning_rate": 5.599820177018695e-07,
      "loss": 1.4753,
      "step": 1422
    },
    {
      "epoch": 0.31274725274725274,
      "grad_norm": 0.24615322583325078,
      "learning_rate": 5.598011251978045e-07,
      "loss": 1.4569,
      "step": 1423
    },
    {
      "epoch": 0.312967032967033,
      "grad_norm": 0.28121091251243596,
      "learning_rate": 5.596201493599449e-07,
      "loss": 1.4872,
      "step": 1424
    },
    {
      "epoch": 0.3131868131868132,
      "grad_norm": 0.2470932136083059,
      "learning_rate": 5.594390902745679e-07,
      "loss": 1.4697,
      "step": 1425
    },
    {
      "epoch": 0.3134065934065934,
      "grad_norm": 0.25181414935091184,
      "learning_rate": 5.592579480279909e-07,
      "loss": 1.4594,
      "step": 1426
    },
    {
      "epoch": 0.31362637362637363,
      "grad_norm": 0.24000622688772374,
      "learning_rate": 5.590767227065706e-07,
      "loss": 1.4319,
      "step": 1427
    },
    {
      "epoch": 0.31384615384615383,
      "grad_norm": 0.27023095618555476,
      "learning_rate": 5.588954143967035e-07,
      "loss": 1.5004,
      "step": 1428
    },
    {
      "epoch": 0.3140659340659341,
      "grad_norm": 0.25649542568927564,
      "learning_rate": 5.587140231848255e-07,
      "loss": 1.4952,
      "step": 1429
    },
    {
      "epoch": 0.3142857142857143,
      "grad_norm": 0.25434607909533713,
      "learning_rate": 5.585325491574123e-07,
      "loss": 1.4152,
      "step": 1430
    },
    {
      "epoch": 0.31450549450549453,
      "grad_norm": 0.29623888881495286,
      "learning_rate": 5.583509924009786e-07,
      "loss": 1.4615,
      "step": 1431
    },
    {
      "epoch": 0.31472527472527473,
      "grad_norm": 0.24808496218296028,
      "learning_rate": 5.581693530020793e-07,
      "loss": 1.4673,
      "step": 1432
    },
    {
      "epoch": 0.3149450549450549,
      "grad_norm": 0.2431664108867739,
      "learning_rate": 5.579876310473079e-07,
      "loss": 1.4419,
      "step": 1433
    },
    {
      "epoch": 0.3151648351648352,
      "grad_norm": 0.25625680181131755,
      "learning_rate": 5.578058266232975e-07,
      "loss": 1.5086,
      "step": 1434
    },
    {
      "epoch": 0.3153846153846154,
      "grad_norm": 0.25239128493632057,
      "learning_rate": 5.576239398167211e-07,
      "loss": 1.3956,
      "step": 1435
    },
    {
      "epoch": 0.3156043956043956,
      "grad_norm": 0.3177987113062903,
      "learning_rate": 5.574419707142899e-07,
      "loss": 1.5333,
      "step": 1436
    },
    {
      "epoch": 0.3158241758241758,
      "grad_norm": 0.24921101440713225,
      "learning_rate": 5.572599194027553e-07,
      "loss": 1.4625,
      "step": 1437
    },
    {
      "epoch": 0.316043956043956,
      "grad_norm": 0.2527575768482531,
      "learning_rate": 5.570777859689074e-07,
      "loss": 1.4016,
      "step": 1438
    },
    {
      "epoch": 0.3162637362637363,
      "grad_norm": 0.25944300953619764,
      "learning_rate": 5.568955704995754e-07,
      "loss": 1.5011,
      "step": 1439
    },
    {
      "epoch": 0.31648351648351647,
      "grad_norm": 0.3000210968151791,
      "learning_rate": 5.567132730816281e-07,
      "loss": 1.43,
      "step": 1440
    },
    {
      "epoch": 0.3167032967032967,
      "grad_norm": 0.2436095655406767,
      "learning_rate": 5.565308938019729e-07,
      "loss": 1.4973,
      "step": 1441
    },
    {
      "epoch": 0.3169230769230769,
      "grad_norm": 0.2513959320546392,
      "learning_rate": 5.563484327475562e-07,
      "loss": 1.4844,
      "step": 1442
    },
    {
      "epoch": 0.3171428571428571,
      "grad_norm": 0.24366656363672098,
      "learning_rate": 5.561658900053638e-07,
      "loss": 1.5005,
      "step": 1443
    },
    {
      "epoch": 0.31736263736263737,
      "grad_norm": 0.24275773764369232,
      "learning_rate": 5.5598326566242e-07,
      "loss": 1.5112,
      "step": 1444
    },
    {
      "epoch": 0.31758241758241756,
      "grad_norm": 0.27869479147088283,
      "learning_rate": 5.558005598057883e-07,
      "loss": 1.4517,
      "step": 1445
    },
    {
      "epoch": 0.3178021978021978,
      "grad_norm": 0.44836788997017724,
      "learning_rate": 5.55617772522571e-07,
      "loss": 1.4828,
      "step": 1446
    },
    {
      "epoch": 0.318021978021978,
      "grad_norm": 0.2524172023696438,
      "learning_rate": 5.55434903899909e-07,
      "loss": 1.4908,
      "step": 1447
    },
    {
      "epoch": 0.31824175824175827,
      "grad_norm": 0.25560366190871997,
      "learning_rate": 5.552519540249823e-07,
      "loss": 1.4537,
      "step": 1448
    },
    {
      "epoch": 0.31846153846153846,
      "grad_norm": 0.24590257703234258,
      "learning_rate": 5.550689229850094e-07,
      "loss": 1.4444,
      "step": 1449
    },
    {
      "epoch": 0.31868131868131866,
      "grad_norm": 0.23869349912438817,
      "learning_rate": 5.548858108672478e-07,
      "loss": 1.3799,
      "step": 1450
    },
    {
      "epoch": 0.3189010989010989,
      "grad_norm": 0.25425221516964575,
      "learning_rate": 5.54702617758993e-07,
      "loss": 1.4499,
      "step": 1451
    },
    {
      "epoch": 0.3191208791208791,
      "grad_norm": 0.24652127780233524,
      "learning_rate": 5.5451934374758e-07,
      "loss": 1.4834,
      "step": 1452
    },
    {
      "epoch": 0.31934065934065936,
      "grad_norm": 0.25748748540312894,
      "learning_rate": 5.543359889203816e-07,
      "loss": 1.5082,
      "step": 1453
    },
    {
      "epoch": 0.31956043956043956,
      "grad_norm": 0.24735774192630483,
      "learning_rate": 5.541525533648098e-07,
      "loss": 1.4448,
      "step": 1454
    },
    {
      "epoch": 0.31978021978021975,
      "grad_norm": 0.25548677018599053,
      "learning_rate": 5.539690371683143e-07,
      "loss": 1.5432,
      "step": 1455
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.24170529129716967,
      "learning_rate": 5.537854404183839e-07,
      "loss": 1.4369,
      "step": 1456
    },
    {
      "epoch": 0.3202197802197802,
      "grad_norm": 0.2499469498321262,
      "learning_rate": 5.536017632025456e-07,
      "loss": 1.4728,
      "step": 1457
    },
    {
      "epoch": 0.32043956043956046,
      "grad_norm": 0.25214039364600493,
      "learning_rate": 5.534180056083646e-07,
      "loss": 1.4431,
      "step": 1458
    },
    {
      "epoch": 0.32065934065934065,
      "grad_norm": 0.25028272901560455,
      "learning_rate": 5.532341677234446e-07,
      "loss": 1.4518,
      "step": 1459
    },
    {
      "epoch": 0.3208791208791209,
      "grad_norm": 0.2563486330300104,
      "learning_rate": 5.530502496354277e-07,
      "loss": 1.4964,
      "step": 1460
    },
    {
      "epoch": 0.3210989010989011,
      "grad_norm": 0.24113311014075664,
      "learning_rate": 5.528662514319939e-07,
      "loss": 1.4403,
      "step": 1461
    },
    {
      "epoch": 0.3213186813186813,
      "grad_norm": 0.24337623711648143,
      "learning_rate": 5.526821732008617e-07,
      "loss": 1.3784,
      "step": 1462
    },
    {
      "epoch": 0.32153846153846155,
      "grad_norm": 0.25167443291990105,
      "learning_rate": 5.524980150297874e-07,
      "loss": 1.5231,
      "step": 1463
    },
    {
      "epoch": 0.32175824175824175,
      "grad_norm": 0.24809838716659088,
      "learning_rate": 5.523137770065658e-07,
      "loss": 1.4718,
      "step": 1464
    },
    {
      "epoch": 0.321978021978022,
      "grad_norm": 0.2511115922667999,
      "learning_rate": 5.521294592190295e-07,
      "loss": 1.4306,
      "step": 1465
    },
    {
      "epoch": 0.3221978021978022,
      "grad_norm": 0.23752395664987544,
      "learning_rate": 5.519450617550492e-07,
      "loss": 1.3631,
      "step": 1466
    },
    {
      "epoch": 0.3224175824175824,
      "grad_norm": 0.2384592070302975,
      "learning_rate": 5.517605847025338e-07,
      "loss": 1.4627,
      "step": 1467
    },
    {
      "epoch": 0.32263736263736265,
      "grad_norm": 0.2592006138141254,
      "learning_rate": 5.515760281494296e-07,
      "loss": 1.4825,
      "step": 1468
    },
    {
      "epoch": 0.32285714285714284,
      "grad_norm": 0.25791435149494757,
      "learning_rate": 5.513913921837216e-07,
      "loss": 1.5405,
      "step": 1469
    },
    {
      "epoch": 0.3230769230769231,
      "grad_norm": 0.25262378690212317,
      "learning_rate": 5.512066768934318e-07,
      "loss": 1.4445,
      "step": 1470
    },
    {
      "epoch": 0.3232967032967033,
      "grad_norm": 0.2618464186806256,
      "learning_rate": 5.510218823666206e-07,
      "loss": 1.4505,
      "step": 1471
    },
    {
      "epoch": 0.32351648351648354,
      "grad_norm": 0.2516345721409484,
      "learning_rate": 5.50837008691386e-07,
      "loss": 1.4902,
      "step": 1472
    },
    {
      "epoch": 0.32373626373626374,
      "grad_norm": 0.2462564886830425,
      "learning_rate": 5.506520559558638e-07,
      "loss": 1.4555,
      "step": 1473
    },
    {
      "epoch": 0.32395604395604394,
      "grad_norm": 0.24897616895697583,
      "learning_rate": 5.504670242482272e-07,
      "loss": 1.4917,
      "step": 1474
    },
    {
      "epoch": 0.3241758241758242,
      "grad_norm": 0.26007246117108856,
      "learning_rate": 5.502819136566875e-07,
      "loss": 1.499,
      "step": 1475
    },
    {
      "epoch": 0.3243956043956044,
      "grad_norm": 0.24427868150389598,
      "learning_rate": 5.500967242694933e-07,
      "loss": 1.4759,
      "step": 1476
    },
    {
      "epoch": 0.32461538461538464,
      "grad_norm": 0.26643060912685457,
      "learning_rate": 5.499114561749307e-07,
      "loss": 1.5278,
      "step": 1477
    },
    {
      "epoch": 0.32483516483516484,
      "grad_norm": 0.25911203533432375,
      "learning_rate": 5.497261094613236e-07,
      "loss": 1.4835,
      "step": 1478
    },
    {
      "epoch": 0.32505494505494503,
      "grad_norm": 0.24657749336443374,
      "learning_rate": 5.495406842170334e-07,
      "loss": 1.4352,
      "step": 1479
    },
    {
      "epoch": 0.3252747252747253,
      "grad_norm": 0.29035923784862794,
      "learning_rate": 5.493551805304585e-07,
      "loss": 1.5027,
      "step": 1480
    },
    {
      "epoch": 0.3254945054945055,
      "grad_norm": 0.24385608355519042,
      "learning_rate": 5.491695984900351e-07,
      "loss": 1.4728,
      "step": 1481
    },
    {
      "epoch": 0.32571428571428573,
      "grad_norm": 0.280604665862483,
      "learning_rate": 5.489839381842367e-07,
      "loss": 1.4319,
      "step": 1482
    },
    {
      "epoch": 0.32593406593406593,
      "grad_norm": 0.24708258531599805,
      "learning_rate": 5.487981997015739e-07,
      "loss": 1.4393,
      "step": 1483
    },
    {
      "epoch": 0.3261538461538461,
      "grad_norm": 0.26137202452625863,
      "learning_rate": 5.486123831305949e-07,
      "loss": 1.4754,
      "step": 1484
    },
    {
      "epoch": 0.3263736263736264,
      "grad_norm": 0.2470221269140314,
      "learning_rate": 5.484264885598847e-07,
      "loss": 1.4455,
      "step": 1485
    },
    {
      "epoch": 0.3265934065934066,
      "grad_norm": 0.24638699029361294,
      "learning_rate": 5.482405160780657e-07,
      "loss": 1.4493,
      "step": 1486
    },
    {
      "epoch": 0.32681318681318683,
      "grad_norm": 0.2522195644263799,
      "learning_rate": 5.480544657737979e-07,
      "loss": 1.4263,
      "step": 1487
    },
    {
      "epoch": 0.327032967032967,
      "grad_norm": 0.24994236548654739,
      "learning_rate": 5.478683377357777e-07,
      "loss": 1.4574,
      "step": 1488
    },
    {
      "epoch": 0.3272527472527473,
      "grad_norm": 0.2542060320773594,
      "learning_rate": 5.476821320527388e-07,
      "loss": 1.5226,
      "step": 1489
    },
    {
      "epoch": 0.3274725274725275,
      "grad_norm": 0.2912830493896432,
      "learning_rate": 5.474958488134518e-07,
      "loss": 1.4136,
      "step": 1490
    },
    {
      "epoch": 0.32769230769230767,
      "grad_norm": 0.2586908381805732,
      "learning_rate": 5.473094881067247e-07,
      "loss": 1.446,
      "step": 1491
    },
    {
      "epoch": 0.3279120879120879,
      "grad_norm": 0.36209478940224404,
      "learning_rate": 5.471230500214021e-07,
      "loss": 1.5041,
      "step": 1492
    },
    {
      "epoch": 0.3281318681318681,
      "grad_norm": 0.25839569965035974,
      "learning_rate": 5.469365346463655e-07,
      "loss": 1.5342,
      "step": 1493
    },
    {
      "epoch": 0.3283516483516484,
      "grad_norm": 0.24023299334045561,
      "learning_rate": 5.467499420705332e-07,
      "loss": 1.466,
      "step": 1494
    },
    {
      "epoch": 0.32857142857142857,
      "grad_norm": 0.3501365700384398,
      "learning_rate": 5.465632723828604e-07,
      "loss": 1.4431,
      "step": 1495
    },
    {
      "epoch": 0.32879120879120877,
      "grad_norm": 0.24418008706073047,
      "learning_rate": 5.463765256723392e-07,
      "loss": 1.523,
      "step": 1496
    },
    {
      "epoch": 0.329010989010989,
      "grad_norm": 0.2559257648559177,
      "learning_rate": 5.461897020279981e-07,
      "loss": 1.4329,
      "step": 1497
    },
    {
      "epoch": 0.3292307692307692,
      "grad_norm": 0.2500029574376721,
      "learning_rate": 5.460028015389027e-07,
      "loss": 1.4854,
      "step": 1498
    },
    {
      "epoch": 0.32945054945054947,
      "grad_norm": 0.30864129210705915,
      "learning_rate": 5.458158242941547e-07,
      "loss": 1.5429,
      "step": 1499
    },
    {
      "epoch": 0.32967032967032966,
      "grad_norm": 0.24016461750834422,
      "learning_rate": 5.456287703828928e-07,
      "loss": 1.5318,
      "step": 1500
    },
    {
      "epoch": 0.3298901098901099,
      "grad_norm": 0.2538073310748625,
      "learning_rate": 5.454416398942922e-07,
      "loss": 1.4958,
      "step": 1501
    },
    {
      "epoch": 0.3301098901098901,
      "grad_norm": 0.24784466552667925,
      "learning_rate": 5.452544329175643e-07,
      "loss": 1.4459,
      "step": 1502
    },
    {
      "epoch": 0.3303296703296703,
      "grad_norm": 0.24349725851625248,
      "learning_rate": 5.450671495419574e-07,
      "loss": 1.5067,
      "step": 1503
    },
    {
      "epoch": 0.33054945054945056,
      "grad_norm": 0.24921253640470473,
      "learning_rate": 5.44879789856756e-07,
      "loss": 1.4844,
      "step": 1504
    },
    {
      "epoch": 0.33076923076923076,
      "grad_norm": 0.2565216850423278,
      "learning_rate": 5.446923539512809e-07,
      "loss": 1.4879,
      "step": 1505
    },
    {
      "epoch": 0.330989010989011,
      "grad_norm": 0.2538660241757986,
      "learning_rate": 5.445048419148894e-07,
      "loss": 1.4429,
      "step": 1506
    },
    {
      "epoch": 0.3312087912087912,
      "grad_norm": 0.2501390660018792,
      "learning_rate": 5.443172538369752e-07,
      "loss": 1.4767,
      "step": 1507
    },
    {
      "epoch": 0.3314285714285714,
      "grad_norm": 0.27314855713848507,
      "learning_rate": 5.441295898069678e-07,
      "loss": 1.4765,
      "step": 1508
    },
    {
      "epoch": 0.33164835164835166,
      "grad_norm": 0.24685113051896446,
      "learning_rate": 5.439418499143333e-07,
      "loss": 1.4548,
      "step": 1509
    },
    {
      "epoch": 0.33186813186813185,
      "grad_norm": 0.24626827370484014,
      "learning_rate": 5.437540342485739e-07,
      "loss": 1.4816,
      "step": 1510
    },
    {
      "epoch": 0.3320879120879121,
      "grad_norm": 0.24065483927609677,
      "learning_rate": 5.435661428992277e-07,
      "loss": 1.4732,
      "step": 1511
    },
    {
      "epoch": 0.3323076923076923,
      "grad_norm": 0.25897101893577673,
      "learning_rate": 5.433781759558694e-07,
      "loss": 1.463,
      "step": 1512
    },
    {
      "epoch": 0.3325274725274725,
      "grad_norm": 0.2535357142888248,
      "learning_rate": 5.431901335081091e-07,
      "loss": 1.4513,
      "step": 1513
    },
    {
      "epoch": 0.33274725274725275,
      "grad_norm": 0.23725764627316678,
      "learning_rate": 5.430020156455932e-07,
      "loss": 1.4642,
      "step": 1514
    },
    {
      "epoch": 0.33296703296703295,
      "grad_norm": 0.2528769432262968,
      "learning_rate": 5.428138224580043e-07,
      "loss": 1.4734,
      "step": 1515
    },
    {
      "epoch": 0.3331868131868132,
      "grad_norm": 0.243999952788361,
      "learning_rate": 5.426255540350604e-07,
      "loss": 1.4582,
      "step": 1516
    },
    {
      "epoch": 0.3334065934065934,
      "grad_norm": 0.2492316620333035,
      "learning_rate": 5.424372104665158e-07,
      "loss": 1.4326,
      "step": 1517
    },
    {
      "epoch": 0.33362637362637365,
      "grad_norm": 0.2353644179244723,
      "learning_rate": 5.422487918421602e-07,
      "loss": 1.4451,
      "step": 1518
    },
    {
      "epoch": 0.33384615384615385,
      "grad_norm": 0.2444805165573388,
      "learning_rate": 5.420602982518198e-07,
      "loss": 1.4774,
      "step": 1519
    },
    {
      "epoch": 0.33406593406593404,
      "grad_norm": 0.2534291025257919,
      "learning_rate": 5.418717297853555e-07,
      "loss": 1.4196,
      "step": 1520
    },
    {
      "epoch": 0.3342857142857143,
      "grad_norm": 0.2509051106833752,
      "learning_rate": 5.416830865326649e-07,
      "loss": 1.457,
      "step": 1521
    },
    {
      "epoch": 0.3345054945054945,
      "grad_norm": 0.24734925028850036,
      "learning_rate": 5.414943685836806e-07,
      "loss": 1.4824,
      "step": 1522
    },
    {
      "epoch": 0.33472527472527475,
      "grad_norm": 0.27155469459810494,
      "learning_rate": 5.413055760283712e-07,
      "loss": 1.4774,
      "step": 1523
    },
    {
      "epoch": 0.33494505494505494,
      "grad_norm": 0.24800887928970217,
      "learning_rate": 5.411167089567405e-07,
      "loss": 1.4984,
      "step": 1524
    },
    {
      "epoch": 0.33516483516483514,
      "grad_norm": 0.25307644663373857,
      "learning_rate": 5.409277674588281e-07,
      "loss": 1.433,
      "step": 1525
    },
    {
      "epoch": 0.3353846153846154,
      "grad_norm": 0.24470622096744218,
      "learning_rate": 5.407387516247091e-07,
      "loss": 1.4192,
      "step": 1526
    },
    {
      "epoch": 0.3356043956043956,
      "grad_norm": 0.25432007620145153,
      "learning_rate": 5.405496615444938e-07,
      "loss": 1.4579,
      "step": 1527
    },
    {
      "epoch": 0.33582417582417584,
      "grad_norm": 0.2554028206333398,
      "learning_rate": 5.40360497308328e-07,
      "loss": 1.4741,
      "step": 1528
    },
    {
      "epoch": 0.33604395604395604,
      "grad_norm": 0.2547609382189186,
      "learning_rate": 5.401712590063932e-07,
      "loss": 1.4119,
      "step": 1529
    },
    {
      "epoch": 0.3362637362637363,
      "grad_norm": 0.2532464316349404,
      "learning_rate": 5.399819467289056e-07,
      "loss": 1.4755,
      "step": 1530
    },
    {
      "epoch": 0.3364835164835165,
      "grad_norm": 0.24845126535571804,
      "learning_rate": 5.397925605661171e-07,
      "loss": 1.4979,
      "step": 1531
    },
    {
      "epoch": 0.3367032967032967,
      "grad_norm": 0.25299051344855855,
      "learning_rate": 5.396031006083145e-07,
      "loss": 1.4916,
      "step": 1532
    },
    {
      "epoch": 0.33692307692307694,
      "grad_norm": 0.24362075788128457,
      "learning_rate": 5.394135669458202e-07,
      "loss": 1.4479,
      "step": 1533
    },
    {
      "epoch": 0.33714285714285713,
      "grad_norm": 0.2456546265907831,
      "learning_rate": 5.392239596689915e-07,
      "loss": 1.4508,
      "step": 1534
    },
    {
      "epoch": 0.3373626373626374,
      "grad_norm": 0.25775963168034405,
      "learning_rate": 5.390342788682207e-07,
      "loss": 1.4284,
      "step": 1535
    },
    {
      "epoch": 0.3375824175824176,
      "grad_norm": 0.26861833752807923,
      "learning_rate": 5.388445246339352e-07,
      "loss": 1.5127,
      "step": 1536
    },
    {
      "epoch": 0.3378021978021978,
      "grad_norm": 0.25490000077366065,
      "learning_rate": 5.386546970565977e-07,
      "loss": 1.486,
      "step": 1537
    },
    {
      "epoch": 0.33802197802197803,
      "grad_norm": 0.23792265683948896,
      "learning_rate": 5.384647962267054e-07,
      "loss": 1.4162,
      "step": 1538
    },
    {
      "epoch": 0.33824175824175823,
      "grad_norm": 0.29004751691553654,
      "learning_rate": 5.382748222347907e-07,
      "loss": 1.5103,
      "step": 1539
    },
    {
      "epoch": 0.3384615384615385,
      "grad_norm": 0.2754869834655155,
      "learning_rate": 5.380847751714209e-07,
      "loss": 1.4724,
      "step": 1540
    },
    {
      "epoch": 0.3386813186813187,
      "grad_norm": 0.23558145026194702,
      "learning_rate": 5.37894655127198e-07,
      "loss": 1.4317,
      "step": 1541
    },
    {
      "epoch": 0.3389010989010989,
      "grad_norm": 0.27225269837790794,
      "learning_rate": 5.37704462192759e-07,
      "loss": 1.4369,
      "step": 1542
    },
    {
      "epoch": 0.3391208791208791,
      "grad_norm": 0.2554558886374575,
      "learning_rate": 5.375141964587753e-07,
      "loss": 1.4098,
      "step": 1543
    },
    {
      "epoch": 0.3393406593406593,
      "grad_norm": 0.23347401876322993,
      "learning_rate": 5.373238580159532e-07,
      "loss": 1.4225,
      "step": 1544
    },
    {
      "epoch": 0.3395604395604396,
      "grad_norm": 0.26102492919944953,
      "learning_rate": 5.371334469550339e-07,
      "loss": 1.4749,
      "step": 1545
    },
    {
      "epoch": 0.33978021978021977,
      "grad_norm": 0.3129338614884883,
      "learning_rate": 5.369429633667929e-07,
      "loss": 1.4221,
      "step": 1546
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2675264151861766,
      "learning_rate": 5.367524073420402e-07,
      "loss": 1.452,
      "step": 1547
    },
    {
      "epoch": 0.3402197802197802,
      "grad_norm": 0.3121086486667206,
      "learning_rate": 5.365617789716208e-07,
      "loss": 1.4714,
      "step": 1548
    },
    {
      "epoch": 0.3404395604395604,
      "grad_norm": 0.2529556901038112,
      "learning_rate": 5.363710783464138e-07,
      "loss": 1.3853,
      "step": 1549
    },
    {
      "epoch": 0.34065934065934067,
      "grad_norm": 0.24490213684593154,
      "learning_rate": 5.361803055573327e-07,
      "loss": 1.4663,
      "step": 1550
    },
    {
      "epoch": 0.34087912087912087,
      "grad_norm": 0.25835859916848847,
      "learning_rate": 5.359894606953256e-07,
      "loss": 1.4538,
      "step": 1551
    },
    {
      "epoch": 0.3410989010989011,
      "grad_norm": 0.2542381749964743,
      "learning_rate": 5.35798543851375e-07,
      "loss": 1.4712,
      "step": 1552
    },
    {
      "epoch": 0.3413186813186813,
      "grad_norm": 0.2547685273733034,
      "learning_rate": 5.356075551164975e-07,
      "loss": 1.5018,
      "step": 1553
    },
    {
      "epoch": 0.3415384615384615,
      "grad_norm": 0.25012508464326705,
      "learning_rate": 5.354164945817442e-07,
      "loss": 1.4672,
      "step": 1554
    },
    {
      "epoch": 0.34175824175824177,
      "grad_norm": 0.23845597850221073,
      "learning_rate": 5.352253623382002e-07,
      "loss": 1.5004,
      "step": 1555
    },
    {
      "epoch": 0.34197802197802196,
      "grad_norm": 0.2830684758224161,
      "learning_rate": 5.350341584769852e-07,
      "loss": 1.4355,
      "step": 1556
    },
    {
      "epoch": 0.3421978021978022,
      "grad_norm": 0.26053788268240063,
      "learning_rate": 5.348428830892524e-07,
      "loss": 1.5193,
      "step": 1557
    },
    {
      "epoch": 0.3424175824175824,
      "grad_norm": 0.2619656313595603,
      "learning_rate": 5.346515362661894e-07,
      "loss": 1.4591,
      "step": 1558
    },
    {
      "epoch": 0.34263736263736266,
      "grad_norm": 0.24384775270031842,
      "learning_rate": 5.344601180990184e-07,
      "loss": 1.473,
      "step": 1559
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 0.24529546851627845,
      "learning_rate": 5.342686286789946e-07,
      "loss": 1.4294,
      "step": 1560
    },
    {
      "epoch": 0.34307692307692306,
      "grad_norm": 0.255705642784143,
      "learning_rate": 5.34077068097408e-07,
      "loss": 1.5235,
      "step": 1561
    },
    {
      "epoch": 0.3432967032967033,
      "grad_norm": 0.2402994273303898,
      "learning_rate": 5.338854364455819e-07,
      "loss": 1.436,
      "step": 1562
    },
    {
      "epoch": 0.3435164835164835,
      "grad_norm": 0.2628094041683089,
      "learning_rate": 5.336937338148743e-07,
      "loss": 1.5406,
      "step": 1563
    },
    {
      "epoch": 0.34373626373626376,
      "grad_norm": 0.25726952374456596,
      "learning_rate": 5.33501960296676e-07,
      "loss": 1.4792,
      "step": 1564
    },
    {
      "epoch": 0.34395604395604396,
      "grad_norm": 0.2489143983098038,
      "learning_rate": 5.333101159824125e-07,
      "loss": 1.4696,
      "step": 1565
    },
    {
      "epoch": 0.34417582417582415,
      "grad_norm": 0.25393076226792394,
      "learning_rate": 5.331182009635425e-07,
      "loss": 1.4839,
      "step": 1566
    },
    {
      "epoch": 0.3443956043956044,
      "grad_norm": 0.24833238790122336,
      "learning_rate": 5.329262153315586e-07,
      "loss": 1.4813,
      "step": 1567
    },
    {
      "epoch": 0.3446153846153846,
      "grad_norm": 0.24877647344306816,
      "learning_rate": 5.327341591779871e-07,
      "loss": 1.4242,
      "step": 1568
    },
    {
      "epoch": 0.34483516483516485,
      "grad_norm": 0.24242360833877535,
      "learning_rate": 5.325420325943879e-07,
      "loss": 1.3694,
      "step": 1569
    },
    {
      "epoch": 0.34505494505494505,
      "grad_norm": 0.2511563121796833,
      "learning_rate": 5.323498356723542e-07,
      "loss": 1.3864,
      "step": 1570
    },
    {
      "epoch": 0.34527472527472525,
      "grad_norm": 0.24910280728725698,
      "learning_rate": 5.321575685035133e-07,
      "loss": 1.4293,
      "step": 1571
    },
    {
      "epoch": 0.3454945054945055,
      "grad_norm": 0.24148226911307036,
      "learning_rate": 5.319652311795255e-07,
      "loss": 1.47,
      "step": 1572
    },
    {
      "epoch": 0.3457142857142857,
      "grad_norm": 0.26042792170550655,
      "learning_rate": 5.317728237920848e-07,
      "loss": 1.5616,
      "step": 1573
    },
    {
      "epoch": 0.34593406593406595,
      "grad_norm": 0.25812273513162637,
      "learning_rate": 5.315803464329184e-07,
      "loss": 1.4829,
      "step": 1574
    },
    {
      "epoch": 0.34615384615384615,
      "grad_norm": 0.24473162545303032,
      "learning_rate": 5.313877991937871e-07,
      "loss": 1.4404,
      "step": 1575
    },
    {
      "epoch": 0.3463736263736264,
      "grad_norm": 0.25852660374979847,
      "learning_rate": 5.311951821664847e-07,
      "loss": 1.4336,
      "step": 1576
    },
    {
      "epoch": 0.3465934065934066,
      "grad_norm": 0.24409302924608858,
      "learning_rate": 5.310024954428387e-07,
      "loss": 1.476,
      "step": 1577
    },
    {
      "epoch": 0.3468131868131868,
      "grad_norm": 0.24349521727617168,
      "learning_rate": 5.308097391147093e-07,
      "loss": 1.5268,
      "step": 1578
    },
    {
      "epoch": 0.34703296703296704,
      "grad_norm": 0.3046686062942096,
      "learning_rate": 5.306169132739905e-07,
      "loss": 1.4584,
      "step": 1579
    },
    {
      "epoch": 0.34725274725274724,
      "grad_norm": 0.2547874869155784,
      "learning_rate": 5.304240180126089e-07,
      "loss": 1.4195,
      "step": 1580
    },
    {
      "epoch": 0.3474725274725275,
      "grad_norm": 0.26855475616874347,
      "learning_rate": 5.302310534225243e-07,
      "loss": 1.4895,
      "step": 1581
    },
    {
      "epoch": 0.3476923076923077,
      "grad_norm": 0.24765425237779218,
      "learning_rate": 5.300380195957299e-07,
      "loss": 1.4945,
      "step": 1582
    },
    {
      "epoch": 0.3479120879120879,
      "grad_norm": 0.2679634206275912,
      "learning_rate": 5.298449166242515e-07,
      "loss": 1.4574,
      "step": 1583
    },
    {
      "epoch": 0.34813186813186814,
      "grad_norm": 0.23905809227530828,
      "learning_rate": 5.29651744600148e-07,
      "loss": 1.4459,
      "step": 1584
    },
    {
      "epoch": 0.34835164835164834,
      "grad_norm": 0.2632656267120424,
      "learning_rate": 5.294585036155113e-07,
      "loss": 1.3949,
      "step": 1585
    },
    {
      "epoch": 0.3485714285714286,
      "grad_norm": 0.25489983481984224,
      "learning_rate": 5.292651937624662e-07,
      "loss": 1.4168,
      "step": 1586
    },
    {
      "epoch": 0.3487912087912088,
      "grad_norm": 0.24517758466897013,
      "learning_rate": 5.290718151331703e-07,
      "loss": 1.4415,
      "step": 1587
    },
    {
      "epoch": 0.34901098901098904,
      "grad_norm": 0.25513692791169457,
      "learning_rate": 5.288783678198137e-07,
      "loss": 1.4955,
      "step": 1588
    },
    {
      "epoch": 0.34923076923076923,
      "grad_norm": 0.2567853058561654,
      "learning_rate": 5.286848519146195e-07,
      "loss": 1.4635,
      "step": 1589
    },
    {
      "epoch": 0.34945054945054943,
      "grad_norm": 0.2536613534859628,
      "learning_rate": 5.284912675098436e-07,
      "loss": 1.4606,
      "step": 1590
    },
    {
      "epoch": 0.3496703296703297,
      "grad_norm": 0.24887742764255438,
      "learning_rate": 5.282976146977745e-07,
      "loss": 1.4145,
      "step": 1591
    },
    {
      "epoch": 0.3498901098901099,
      "grad_norm": 0.2652381331710101,
      "learning_rate": 5.281038935707332e-07,
      "loss": 1.529,
      "step": 1592
    },
    {
      "epoch": 0.35010989010989013,
      "grad_norm": 0.27378876042200684,
      "learning_rate": 5.279101042210732e-07,
      "loss": 1.5016,
      "step": 1593
    },
    {
      "epoch": 0.35032967032967033,
      "grad_norm": 0.2624307840502721,
      "learning_rate": 5.277162467411808e-07,
      "loss": 1.466,
      "step": 1594
    },
    {
      "epoch": 0.3505494505494505,
      "grad_norm": 0.2665690066369142,
      "learning_rate": 5.275223212234747e-07,
      "loss": 1.5186,
      "step": 1595
    },
    {
      "epoch": 0.3507692307692308,
      "grad_norm": 0.24693046472316255,
      "learning_rate": 5.273283277604057e-07,
      "loss": 1.5041,
      "step": 1596
    },
    {
      "epoch": 0.350989010989011,
      "grad_norm": 0.26296307979312056,
      "learning_rate": 5.271342664444574e-07,
      "loss": 1.4657,
      "step": 1597
    },
    {
      "epoch": 0.3512087912087912,
      "grad_norm": 0.24454598187445337,
      "learning_rate": 5.269401373681455e-07,
      "loss": 1.489,
      "step": 1598
    },
    {
      "epoch": 0.3514285714285714,
      "grad_norm": 0.24554422049185623,
      "learning_rate": 5.267459406240181e-07,
      "loss": 1.4633,
      "step": 1599
    },
    {
      "epoch": 0.3516483516483517,
      "grad_norm": 0.25108540014016983,
      "learning_rate": 5.265516763046558e-07,
      "loss": 1.4525,
      "step": 1600
    },
    {
      "epoch": 0.3518681318681319,
      "grad_norm": 0.26604553347994064,
      "learning_rate": 5.26357344502671e-07,
      "loss": 1.4961,
      "step": 1601
    },
    {
      "epoch": 0.35208791208791207,
      "grad_norm": 0.25438004120268287,
      "learning_rate": 5.261629453107084e-07,
      "loss": 1.5024,
      "step": 1602
    },
    {
      "epoch": 0.3523076923076923,
      "grad_norm": 0.2485671186455379,
      "learning_rate": 5.25968478821445e-07,
      "loss": 1.4366,
      "step": 1603
    },
    {
      "epoch": 0.3525274725274725,
      "grad_norm": 0.2576362245656226,
      "learning_rate": 5.257739451275897e-07,
      "loss": 1.5253,
      "step": 1604
    },
    {
      "epoch": 0.35274725274725277,
      "grad_norm": 0.24371850011055635,
      "learning_rate": 5.255793443218833e-07,
      "loss": 1.4072,
      "step": 1605
    },
    {
      "epoch": 0.35296703296703297,
      "grad_norm": 0.247289572853217,
      "learning_rate": 5.253846764970993e-07,
      "loss": 1.4588,
      "step": 1606
    },
    {
      "epoch": 0.35318681318681316,
      "grad_norm": 0.2530801955847355,
      "learning_rate": 5.251899417460424e-07,
      "loss": 1.4445,
      "step": 1607
    },
    {
      "epoch": 0.3534065934065934,
      "grad_norm": 0.2505612898934513,
      "learning_rate": 5.249951401615491e-07,
      "loss": 1.4632,
      "step": 1608
    },
    {
      "epoch": 0.3536263736263736,
      "grad_norm": 0.24772309343233298,
      "learning_rate": 5.248002718364885e-07,
      "loss": 1.4974,
      "step": 1609
    },
    {
      "epoch": 0.35384615384615387,
      "grad_norm": 0.266177888918143,
      "learning_rate": 5.24605336863761e-07,
      "loss": 1.545,
      "step": 1610
    },
    {
      "epoch": 0.35406593406593406,
      "grad_norm": 0.2533717626020377,
      "learning_rate": 5.244103353362991e-07,
      "loss": 1.4836,
      "step": 1611
    },
    {
      "epoch": 0.35428571428571426,
      "grad_norm": 0.2592502180167789,
      "learning_rate": 5.242152673470665e-07,
      "loss": 1.4776,
      "step": 1612
    },
    {
      "epoch": 0.3545054945054945,
      "grad_norm": 0.2969363779571865,
      "learning_rate": 5.24020132989059e-07,
      "loss": 1.5055,
      "step": 1613
    },
    {
      "epoch": 0.3547252747252747,
      "grad_norm": 0.25396152465741845,
      "learning_rate": 5.238249323553043e-07,
      "loss": 1.5134,
      "step": 1614
    },
    {
      "epoch": 0.35494505494505496,
      "grad_norm": 0.2652262284261434,
      "learning_rate": 5.236296655388607e-07,
      "loss": 1.4805,
      "step": 1615
    },
    {
      "epoch": 0.35516483516483516,
      "grad_norm": 0.24838897986825226,
      "learning_rate": 5.234343326328194e-07,
      "loss": 1.5331,
      "step": 1616
    },
    {
      "epoch": 0.3553846153846154,
      "grad_norm": 0.24000930493173722,
      "learning_rate": 5.232389337303021e-07,
      "loss": 1.4756,
      "step": 1617
    },
    {
      "epoch": 0.3556043956043956,
      "grad_norm": 0.25131057253330996,
      "learning_rate": 5.230434689244622e-07,
      "loss": 1.461,
      "step": 1618
    },
    {
      "epoch": 0.3558241758241758,
      "grad_norm": 0.2502103746681586,
      "learning_rate": 5.228479383084847e-07,
      "loss": 1.5073,
      "step": 1619
    },
    {
      "epoch": 0.35604395604395606,
      "grad_norm": 0.30898660360946234,
      "learning_rate": 5.226523419755858e-07,
      "loss": 1.4893,
      "step": 1620
    },
    {
      "epoch": 0.35626373626373625,
      "grad_norm": 0.24204538896547698,
      "learning_rate": 5.224566800190131e-07,
      "loss": 1.4025,
      "step": 1621
    },
    {
      "epoch": 0.3564835164835165,
      "grad_norm": 0.26778749718029793,
      "learning_rate": 5.222609525320456e-07,
      "loss": 1.4515,
      "step": 1622
    },
    {
      "epoch": 0.3567032967032967,
      "grad_norm": 0.2543818679452358,
      "learning_rate": 5.220651596079932e-07,
      "loss": 1.448,
      "step": 1623
    },
    {
      "epoch": 0.3569230769230769,
      "grad_norm": 0.23790757421239175,
      "learning_rate": 5.218693013401976e-07,
      "loss": 1.4724,
      "step": 1624
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 0.2524267519855943,
      "learning_rate": 5.216733778220308e-07,
      "loss": 1.4919,
      "step": 1625
    },
    {
      "epoch": 0.35736263736263735,
      "grad_norm": 0.25258120831921677,
      "learning_rate": 5.214773891468966e-07,
      "loss": 1.4999,
      "step": 1626
    },
    {
      "epoch": 0.3575824175824176,
      "grad_norm": 0.25253733179722276,
      "learning_rate": 5.212813354082297e-07,
      "loss": 1.5095,
      "step": 1627
    },
    {
      "epoch": 0.3578021978021978,
      "grad_norm": 0.25846443644643013,
      "learning_rate": 5.210852166994957e-07,
      "loss": 1.4919,
      "step": 1628
    },
    {
      "epoch": 0.35802197802197805,
      "grad_norm": 0.2547355568107916,
      "learning_rate": 5.208890331141913e-07,
      "loss": 1.4914,
      "step": 1629
    },
    {
      "epoch": 0.35824175824175825,
      "grad_norm": 0.2550941584430843,
      "learning_rate": 5.206927847458439e-07,
      "loss": 1.4769,
      "step": 1630
    },
    {
      "epoch": 0.35846153846153844,
      "grad_norm": 0.2525393773988044,
      "learning_rate": 5.20496471688012e-07,
      "loss": 1.4887,
      "step": 1631
    },
    {
      "epoch": 0.3586813186813187,
      "grad_norm": 0.2512907518837691,
      "learning_rate": 5.203000940342849e-07,
      "loss": 1.5132,
      "step": 1632
    },
    {
      "epoch": 0.3589010989010989,
      "grad_norm": 0.24874642160707364,
      "learning_rate": 5.201036518782827e-07,
      "loss": 1.4425,
      "step": 1633
    },
    {
      "epoch": 0.35912087912087914,
      "grad_norm": 0.24342043499629476,
      "learning_rate": 5.199071453136563e-07,
      "loss": 1.4835,
      "step": 1634
    },
    {
      "epoch": 0.35934065934065934,
      "grad_norm": 0.24306007021206974,
      "learning_rate": 5.19710574434087e-07,
      "loss": 1.4775,
      "step": 1635
    },
    {
      "epoch": 0.35956043956043954,
      "grad_norm": 0.26260245456911935,
      "learning_rate": 5.195139393332873e-07,
      "loss": 1.4983,
      "step": 1636
    },
    {
      "epoch": 0.3597802197802198,
      "grad_norm": 0.250414658674308,
      "learning_rate": 5.193172401049997e-07,
      "loss": 1.4913,
      "step": 1637
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.26081144251159283,
      "learning_rate": 5.191204768429979e-07,
      "loss": 1.4465,
      "step": 1638
    },
    {
      "epoch": 0.36021978021978024,
      "grad_norm": 0.3376410662457479,
      "learning_rate": 5.189236496410856e-07,
      "loss": 1.4447,
      "step": 1639
    },
    {
      "epoch": 0.36043956043956044,
      "grad_norm": 0.24503990004583534,
      "learning_rate": 5.187267585930972e-07,
      "loss": 1.5,
      "step": 1640
    },
    {
      "epoch": 0.36065934065934063,
      "grad_norm": 0.2617886668056767,
      "learning_rate": 5.185298037928977e-07,
      "loss": 1.4788,
      "step": 1641
    },
    {
      "epoch": 0.3608791208791209,
      "grad_norm": 0.2624974611371695,
      "learning_rate": 5.18332785334382e-07,
      "loss": 1.4284,
      "step": 1642
    },
    {
      "epoch": 0.3610989010989011,
      "grad_norm": 0.25429901880516215,
      "learning_rate": 5.181357033114761e-07,
      "loss": 1.498,
      "step": 1643
    },
    {
      "epoch": 0.36131868131868133,
      "grad_norm": 0.24410133896726638,
      "learning_rate": 5.179385578181356e-07,
      "loss": 1.469,
      "step": 1644
    },
    {
      "epoch": 0.36153846153846153,
      "grad_norm": 0.23753990605159025,
      "learning_rate": 5.177413489483467e-07,
      "loss": 1.4206,
      "step": 1645
    },
    {
      "epoch": 0.3617582417582418,
      "grad_norm": 0.24188579178071082,
      "learning_rate": 5.175440767961257e-07,
      "loss": 1.4974,
      "step": 1646
    },
    {
      "epoch": 0.361978021978022,
      "grad_norm": 0.24624653475079045,
      "learning_rate": 5.173467414555191e-07,
      "loss": 1.4652,
      "step": 1647
    },
    {
      "epoch": 0.3621978021978022,
      "grad_norm": 0.289364946335418,
      "learning_rate": 5.171493430206037e-07,
      "loss": 1.4478,
      "step": 1648
    },
    {
      "epoch": 0.36241758241758243,
      "grad_norm": 0.26847709016957916,
      "learning_rate": 5.169518815854861e-07,
      "loss": 1.5022,
      "step": 1649
    },
    {
      "epoch": 0.3626373626373626,
      "grad_norm": 0.2570575009886609,
      "learning_rate": 5.167543572443029e-07,
      "loss": 1.4544,
      "step": 1650
    },
    {
      "epoch": 0.3628571428571429,
      "grad_norm": 0.257286953135937,
      "learning_rate": 5.165567700912213e-07,
      "loss": 1.4779,
      "step": 1651
    },
    {
      "epoch": 0.3630769230769231,
      "grad_norm": 0.25155373502651135,
      "learning_rate": 5.163591202204376e-07,
      "loss": 1.4467,
      "step": 1652
    },
    {
      "epoch": 0.36329670329670327,
      "grad_norm": 0.2435159741570224,
      "learning_rate": 5.161614077261784e-07,
      "loss": 1.4807,
      "step": 1653
    },
    {
      "epoch": 0.3635164835164835,
      "grad_norm": 0.24590872121935767,
      "learning_rate": 5.159636327027003e-07,
      "loss": 1.5235,
      "step": 1654
    },
    {
      "epoch": 0.3637362637362637,
      "grad_norm": 0.2625138461633969,
      "learning_rate": 5.157657952442896e-07,
      "loss": 1.4511,
      "step": 1655
    },
    {
      "epoch": 0.363956043956044,
      "grad_norm": 0.24395697878331868,
      "learning_rate": 5.155678954452621e-07,
      "loss": 1.464,
      "step": 1656
    },
    {
      "epoch": 0.36417582417582417,
      "grad_norm": 0.24486474113797835,
      "learning_rate": 5.153699333999635e-07,
      "loss": 1.4713,
      "step": 1657
    },
    {
      "epoch": 0.3643956043956044,
      "grad_norm": 0.27316487002629064,
      "learning_rate": 5.151719092027696e-07,
      "loss": 1.4838,
      "step": 1658
    },
    {
      "epoch": 0.3646153846153846,
      "grad_norm": 0.2444490217388486,
      "learning_rate": 5.149738229480853e-07,
      "loss": 1.4095,
      "step": 1659
    },
    {
      "epoch": 0.3648351648351648,
      "grad_norm": 0.2607606967463366,
      "learning_rate": 5.14775674730345e-07,
      "loss": 1.4843,
      "step": 1660
    },
    {
      "epoch": 0.36505494505494507,
      "grad_norm": 0.26827223696239494,
      "learning_rate": 5.145774646440129e-07,
      "loss": 1.5493,
      "step": 1661
    },
    {
      "epoch": 0.36527472527472526,
      "grad_norm": 0.2648935836241543,
      "learning_rate": 5.143791927835829e-07,
      "loss": 1.4539,
      "step": 1662
    },
    {
      "epoch": 0.3654945054945055,
      "grad_norm": 0.2367344002413746,
      "learning_rate": 5.141808592435779e-07,
      "loss": 1.388,
      "step": 1663
    },
    {
      "epoch": 0.3657142857142857,
      "grad_norm": 0.27471419900946775,
      "learning_rate": 5.139824641185505e-07,
      "loss": 1.4936,
      "step": 1664
    },
    {
      "epoch": 0.3659340659340659,
      "grad_norm": 0.24806728311125092,
      "learning_rate": 5.137840075030826e-07,
      "loss": 1.4664,
      "step": 1665
    },
    {
      "epoch": 0.36615384615384616,
      "grad_norm": 0.26246681906181907,
      "learning_rate": 5.135854894917853e-07,
      "loss": 1.4827,
      "step": 1666
    },
    {
      "epoch": 0.36637362637362636,
      "grad_norm": 0.2675634521745081,
      "learning_rate": 5.133869101792993e-07,
      "loss": 1.5009,
      "step": 1667
    },
    {
      "epoch": 0.3665934065934066,
      "grad_norm": 0.2433570578602176,
      "learning_rate": 5.13188269660294e-07,
      "loss": 1.4872,
      "step": 1668
    },
    {
      "epoch": 0.3668131868131868,
      "grad_norm": 0.24948785671905935,
      "learning_rate": 5.129895680294683e-07,
      "loss": 1.4696,
      "step": 1669
    },
    {
      "epoch": 0.367032967032967,
      "grad_norm": 0.2554563221150702,
      "learning_rate": 5.127908053815505e-07,
      "loss": 1.5169,
      "step": 1670
    },
    {
      "epoch": 0.36725274725274726,
      "grad_norm": 0.258765505529143,
      "learning_rate": 5.125919818112974e-07,
      "loss": 1.4698,
      "step": 1671
    },
    {
      "epoch": 0.36747252747252745,
      "grad_norm": 0.24230009065253344,
      "learning_rate": 5.12393097413495e-07,
      "loss": 1.3884,
      "step": 1672
    },
    {
      "epoch": 0.3676923076923077,
      "grad_norm": 0.2548695647565594,
      "learning_rate": 5.121941522829588e-07,
      "loss": 1.4249,
      "step": 1673
    },
    {
      "epoch": 0.3679120879120879,
      "grad_norm": 0.25626680131633506,
      "learning_rate": 5.119951465145327e-07,
      "loss": 1.4579,
      "step": 1674
    },
    {
      "epoch": 0.36813186813186816,
      "grad_norm": 0.24812312088905314,
      "learning_rate": 5.117960802030897e-07,
      "loss": 1.5091,
      "step": 1675
    },
    {
      "epoch": 0.36835164835164835,
      "grad_norm": 0.24947351298920709,
      "learning_rate": 5.115969534435317e-07,
      "loss": 1.473,
      "step": 1676
    },
    {
      "epoch": 0.36857142857142855,
      "grad_norm": 0.25146348516505296,
      "learning_rate": 5.113977663307894e-07,
      "loss": 1.4645,
      "step": 1677
    },
    {
      "epoch": 0.3687912087912088,
      "grad_norm": 0.2633496405142336,
      "learning_rate": 5.11198518959822e-07,
      "loss": 1.503,
      "step": 1678
    },
    {
      "epoch": 0.369010989010989,
      "grad_norm": 0.25333672830246573,
      "learning_rate": 5.109992114256181e-07,
      "loss": 1.5201,
      "step": 1679
    },
    {
      "epoch": 0.36923076923076925,
      "grad_norm": 0.257925734816316,
      "learning_rate": 5.107998438231941e-07,
      "loss": 1.4488,
      "step": 1680
    },
    {
      "epoch": 0.36945054945054945,
      "grad_norm": 0.33885595813935115,
      "learning_rate": 5.106004162475958e-07,
      "loss": 1.4063,
      "step": 1681
    },
    {
      "epoch": 0.36967032967032964,
      "grad_norm": 0.25336712226363317,
      "learning_rate": 5.104009287938972e-07,
      "loss": 1.525,
      "step": 1682
    },
    {
      "epoch": 0.3698901098901099,
      "grad_norm": 0.25680866951052767,
      "learning_rate": 5.102013815572008e-07,
      "loss": 1.475,
      "step": 1683
    },
    {
      "epoch": 0.3701098901098901,
      "grad_norm": 0.24826865203303666,
      "learning_rate": 5.100017746326379e-07,
      "loss": 1.4953,
      "step": 1684
    },
    {
      "epoch": 0.37032967032967035,
      "grad_norm": 0.2479780259362285,
      "learning_rate": 5.09802108115368e-07,
      "loss": 1.5105,
      "step": 1685
    },
    {
      "epoch": 0.37054945054945054,
      "grad_norm": 0.2379512224951037,
      "learning_rate": 5.096023821005793e-07,
      "loss": 1.5068,
      "step": 1686
    },
    {
      "epoch": 0.3707692307692308,
      "grad_norm": 0.27283424923387284,
      "learning_rate": 5.094025966834879e-07,
      "loss": 1.5117,
      "step": 1687
    },
    {
      "epoch": 0.370989010989011,
      "grad_norm": 0.2715125554339445,
      "learning_rate": 5.092027519593385e-07,
      "loss": 1.4377,
      "step": 1688
    },
    {
      "epoch": 0.3712087912087912,
      "grad_norm": 0.24841593700618475,
      "learning_rate": 5.090028480234041e-07,
      "loss": 1.4514,
      "step": 1689
    },
    {
      "epoch": 0.37142857142857144,
      "grad_norm": 0.24937969008749136,
      "learning_rate": 5.088028849709859e-07,
      "loss": 1.5119,
      "step": 1690
    },
    {
      "epoch": 0.37164835164835164,
      "grad_norm": 0.2524058970252941,
      "learning_rate": 5.086028628974133e-07,
      "loss": 1.4801,
      "step": 1691
    },
    {
      "epoch": 0.3718681318681319,
      "grad_norm": 0.2475098724614424,
      "learning_rate": 5.084027818980437e-07,
      "loss": 1.5341,
      "step": 1692
    },
    {
      "epoch": 0.3720879120879121,
      "grad_norm": 0.2627371302017303,
      "learning_rate": 5.082026420682629e-07,
      "loss": 1.484,
      "step": 1693
    },
    {
      "epoch": 0.3723076923076923,
      "grad_norm": 0.24337253730307487,
      "learning_rate": 5.080024435034843e-07,
      "loss": 1.4469,
      "step": 1694
    },
    {
      "epoch": 0.37252747252747254,
      "grad_norm": 0.27099817798239256,
      "learning_rate": 5.078021862991496e-07,
      "loss": 1.5334,
      "step": 1695
    },
    {
      "epoch": 0.37274725274725273,
      "grad_norm": 0.2352378169507532,
      "learning_rate": 5.076018705507285e-07,
      "loss": 1.3952,
      "step": 1696
    },
    {
      "epoch": 0.372967032967033,
      "grad_norm": 0.24398355494635385,
      "learning_rate": 5.074014963537183e-07,
      "loss": 1.4562,
      "step": 1697
    },
    {
      "epoch": 0.3731868131868132,
      "grad_norm": 0.24711736541024606,
      "learning_rate": 5.072010638036447e-07,
      "loss": 1.4462,
      "step": 1698
    },
    {
      "epoch": 0.3734065934065934,
      "grad_norm": 0.28791534650705003,
      "learning_rate": 5.070005729960605e-07,
      "loss": 1.5435,
      "step": 1699
    },
    {
      "epoch": 0.37362637362637363,
      "grad_norm": 0.23699801997694145,
      "learning_rate": 5.068000240265469e-07,
      "loss": 1.4298,
      "step": 1700
    },
    {
      "epoch": 0.37384615384615383,
      "grad_norm": 0.2677719563086052,
      "learning_rate": 5.065994169907127e-07,
      "loss": 1.4486,
      "step": 1701
    },
    {
      "epoch": 0.3740659340659341,
      "grad_norm": 0.24943279185346431,
      "learning_rate": 5.063987519841939e-07,
      "loss": 1.4759,
      "step": 1702
    },
    {
      "epoch": 0.3742857142857143,
      "grad_norm": 0.25568632667159863,
      "learning_rate": 5.061980291026549e-07,
      "loss": 1.4719,
      "step": 1703
    },
    {
      "epoch": 0.37450549450549453,
      "grad_norm": 0.24438577886394727,
      "learning_rate": 5.05997248441787e-07,
      "loss": 1.4523,
      "step": 1704
    },
    {
      "epoch": 0.3747252747252747,
      "grad_norm": 0.2586426162558187,
      "learning_rate": 5.057964100973096e-07,
      "loss": 1.5342,
      "step": 1705
    },
    {
      "epoch": 0.3749450549450549,
      "grad_norm": 0.24637124850913075,
      "learning_rate": 5.055955141649691e-07,
      "loss": 1.4716,
      "step": 1706
    },
    {
      "epoch": 0.3751648351648352,
      "grad_norm": 0.2626533839286568,
      "learning_rate": 5.053945607405396e-07,
      "loss": 1.5064,
      "step": 1707
    },
    {
      "epoch": 0.37538461538461537,
      "grad_norm": 0.2681991834944331,
      "learning_rate": 5.05193549919823e-07,
      "loss": 1.4551,
      "step": 1708
    },
    {
      "epoch": 0.3756043956043956,
      "grad_norm": 0.2489845477272852,
      "learning_rate": 5.049924817986477e-07,
      "loss": 1.4625,
      "step": 1709
    },
    {
      "epoch": 0.3758241758241758,
      "grad_norm": 0.24985291717622093,
      "learning_rate": 5.047913564728703e-07,
      "loss": 1.4062,
      "step": 1710
    },
    {
      "epoch": 0.376043956043956,
      "grad_norm": 0.24988919932688364,
      "learning_rate": 5.045901740383739e-07,
      "loss": 1.5092,
      "step": 1711
    },
    {
      "epoch": 0.37626373626373627,
      "grad_norm": 0.25271003886000754,
      "learning_rate": 5.043889345910693e-07,
      "loss": 1.4787,
      "step": 1712
    },
    {
      "epoch": 0.37648351648351647,
      "grad_norm": 0.2505745971018397,
      "learning_rate": 5.041876382268945e-07,
      "loss": 1.4293,
      "step": 1713
    },
    {
      "epoch": 0.3767032967032967,
      "grad_norm": 0.24482954309101237,
      "learning_rate": 5.039862850418143e-07,
      "loss": 1.4475,
      "step": 1714
    },
    {
      "epoch": 0.3769230769230769,
      "grad_norm": 0.24170670857954743,
      "learning_rate": 5.037848751318209e-07,
      "loss": 1.4701,
      "step": 1715
    },
    {
      "epoch": 0.37714285714285717,
      "grad_norm": 0.28057626991866186,
      "learning_rate": 5.035834085929334e-07,
      "loss": 1.478,
      "step": 1716
    },
    {
      "epoch": 0.37736263736263737,
      "grad_norm": 0.242264466938029,
      "learning_rate": 5.033818855211981e-07,
      "loss": 1.4919,
      "step": 1717
    },
    {
      "epoch": 0.37758241758241756,
      "grad_norm": 0.24614319025443693,
      "learning_rate": 5.031803060126877e-07,
      "loss": 1.4668,
      "step": 1718
    },
    {
      "epoch": 0.3778021978021978,
      "grad_norm": 0.24586533904885063,
      "learning_rate": 5.029786701635025e-07,
      "loss": 1.4821,
      "step": 1719
    },
    {
      "epoch": 0.378021978021978,
      "grad_norm": 0.25410869040339945,
      "learning_rate": 5.027769780697694e-07,
      "loss": 1.462,
      "step": 1720
    },
    {
      "epoch": 0.37824175824175826,
      "grad_norm": 0.30108580189647816,
      "learning_rate": 5.025752298276416e-07,
      "loss": 1.5574,
      "step": 1721
    },
    {
      "epoch": 0.37846153846153846,
      "grad_norm": 0.2604357948050287,
      "learning_rate": 5.023734255333e-07,
      "loss": 1.51,
      "step": 1722
    },
    {
      "epoch": 0.37868131868131866,
      "grad_norm": 0.2488240596006886,
      "learning_rate": 5.021715652829517e-07,
      "loss": 1.4783,
      "step": 1723
    },
    {
      "epoch": 0.3789010989010989,
      "grad_norm": 0.24486155407460908,
      "learning_rate": 5.019696491728302e-07,
      "loss": 1.4731,
      "step": 1724
    },
    {
      "epoch": 0.3791208791208791,
      "grad_norm": 0.24814047886420185,
      "learning_rate": 5.017676772991964e-07,
      "loss": 1.4932,
      "step": 1725
    },
    {
      "epoch": 0.37934065934065936,
      "grad_norm": 0.25141982006108415,
      "learning_rate": 5.015656497583367e-07,
      "loss": 1.4612,
      "step": 1726
    },
    {
      "epoch": 0.37956043956043956,
      "grad_norm": 0.24246762504972255,
      "learning_rate": 5.013635666465653e-07,
      "loss": 1.4322,
      "step": 1727
    },
    {
      "epoch": 0.3797802197802198,
      "grad_norm": 0.25178380724344507,
      "learning_rate": 5.011614280602222e-07,
      "loss": 1.4421,
      "step": 1728
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24372812872379204,
      "learning_rate": 5.009592340956736e-07,
      "loss": 1.4522,
      "step": 1729
    },
    {
      "epoch": 0.3802197802197802,
      "grad_norm": 0.25813976594211274,
      "learning_rate": 5.007569848493125e-07,
      "loss": 1.5024,
      "step": 1730
    },
    {
      "epoch": 0.38043956043956045,
      "grad_norm": 0.24717820104448904,
      "learning_rate": 5.005546804175584e-07,
      "loss": 1.5187,
      "step": 1731
    },
    {
      "epoch": 0.38065934065934065,
      "grad_norm": 0.24230905820266857,
      "learning_rate": 5.003523208968566e-07,
      "loss": 1.4505,
      "step": 1732
    },
    {
      "epoch": 0.3808791208791209,
      "grad_norm": 0.24157743771302173,
      "learning_rate": 5.001499063836791e-07,
      "loss": 1.445,
      "step": 1733
    },
    {
      "epoch": 0.3810989010989011,
      "grad_norm": 0.2463804707054971,
      "learning_rate": 4.99947436974524e-07,
      "loss": 1.4571,
      "step": 1734
    },
    {
      "epoch": 0.3813186813186813,
      "grad_norm": 0.251794984418654,
      "learning_rate": 4.997449127659155e-07,
      "loss": 1.4895,
      "step": 1735
    },
    {
      "epoch": 0.38153846153846155,
      "grad_norm": 0.2528992624714515,
      "learning_rate": 4.995423338544038e-07,
      "loss": 1.4259,
      "step": 1736
    },
    {
      "epoch": 0.38175824175824175,
      "grad_norm": 0.2435411471713826,
      "learning_rate": 4.993397003365654e-07,
      "loss": 1.5032,
      "step": 1737
    },
    {
      "epoch": 0.381978021978022,
      "grad_norm": 0.2517055998896912,
      "learning_rate": 4.99137012309003e-07,
      "loss": 1.5022,
      "step": 1738
    },
    {
      "epoch": 0.3821978021978022,
      "grad_norm": 0.26009088996567187,
      "learning_rate": 4.989342698683447e-07,
      "loss": 1.5164,
      "step": 1739
    },
    {
      "epoch": 0.3824175824175824,
      "grad_norm": 0.254362818031432,
      "learning_rate": 4.987314731112453e-07,
      "loss": 1.479,
      "step": 1740
    },
    {
      "epoch": 0.38263736263736264,
      "grad_norm": 0.2597714051834542,
      "learning_rate": 4.985286221343846e-07,
      "loss": 1.4416,
      "step": 1741
    },
    {
      "epoch": 0.38285714285714284,
      "grad_norm": 0.25528039678124415,
      "learning_rate": 4.983257170344692e-07,
      "loss": 1.5145,
      "step": 1742
    },
    {
      "epoch": 0.3830769230769231,
      "grad_norm": 0.2547350729810254,
      "learning_rate": 4.981227579082307e-07,
      "loss": 1.4664,
      "step": 1743
    },
    {
      "epoch": 0.3832967032967033,
      "grad_norm": 0.23974405079595368,
      "learning_rate": 4.97919744852427e-07,
      "loss": 1.3935,
      "step": 1744
    },
    {
      "epoch": 0.38351648351648354,
      "grad_norm": 0.2601908249110836,
      "learning_rate": 4.977166779638414e-07,
      "loss": 1.4609,
      "step": 1745
    },
    {
      "epoch": 0.38373626373626374,
      "grad_norm": 0.256488928208101,
      "learning_rate": 4.97513557339283e-07,
      "loss": 1.4967,
      "step": 1746
    },
    {
      "epoch": 0.38395604395604394,
      "grad_norm": 0.24641004140866646,
      "learning_rate": 4.973103830755865e-07,
      "loss": 1.4455,
      "step": 1747
    },
    {
      "epoch": 0.3841758241758242,
      "grad_norm": 0.5478959950888969,
      "learning_rate": 4.971071552696123e-07,
      "loss": 1.4091,
      "step": 1748
    },
    {
      "epoch": 0.3843956043956044,
      "grad_norm": 0.2472976803178337,
      "learning_rate": 4.969038740182459e-07,
      "loss": 1.4679,
      "step": 1749
    },
    {
      "epoch": 0.38461538461538464,
      "grad_norm": 0.24927157233560374,
      "learning_rate": 4.967005394183987e-07,
      "loss": 1.54,
      "step": 1750
    },
    {
      "epoch": 0.38483516483516483,
      "grad_norm": 0.3448194022458905,
      "learning_rate": 4.964971515670075e-07,
      "loss": 1.4847,
      "step": 1751
    },
    {
      "epoch": 0.38505494505494503,
      "grad_norm": 0.2542207946374973,
      "learning_rate": 4.962937105610342e-07,
      "loss": 1.4905,
      "step": 1752
    },
    {
      "epoch": 0.3852747252747253,
      "grad_norm": 0.2468556554844839,
      "learning_rate": 4.960902164974662e-07,
      "loss": 1.4631,
      "step": 1753
    },
    {
      "epoch": 0.3854945054945055,
      "grad_norm": 0.2539802160889825,
      "learning_rate": 4.958866694733165e-07,
      "loss": 1.4537,
      "step": 1754
    },
    {
      "epoch": 0.38571428571428573,
      "grad_norm": 0.24328159268134317,
      "learning_rate": 4.95683069585623e-07,
      "loss": 1.4392,
      "step": 1755
    },
    {
      "epoch": 0.38593406593406593,
      "grad_norm": 0.25444687986670267,
      "learning_rate": 4.954794169314485e-07,
      "loss": 1.4786,
      "step": 1756
    },
    {
      "epoch": 0.3861538461538462,
      "grad_norm": 0.2734835499805895,
      "learning_rate": 4.952757116078817e-07,
      "loss": 1.482,
      "step": 1757
    },
    {
      "epoch": 0.3863736263736264,
      "grad_norm": 0.24566099112744974,
      "learning_rate": 4.950719537120362e-07,
      "loss": 1.4364,
      "step": 1758
    },
    {
      "epoch": 0.3865934065934066,
      "grad_norm": 0.2596874713017255,
      "learning_rate": 4.948681433410498e-07,
      "loss": 1.5016,
      "step": 1759
    },
    {
      "epoch": 0.3868131868131868,
      "grad_norm": 0.246167195794601,
      "learning_rate": 4.946642805920866e-07,
      "loss": 1.4643,
      "step": 1760
    },
    {
      "epoch": 0.387032967032967,
      "grad_norm": 0.2574672715077331,
      "learning_rate": 4.944603655623351e-07,
      "loss": 1.48,
      "step": 1761
    },
    {
      "epoch": 0.3872527472527473,
      "grad_norm": 0.24588561989161187,
      "learning_rate": 4.942563983490084e-07,
      "loss": 1.4308,
      "step": 1762
    },
    {
      "epoch": 0.3874725274725275,
      "grad_norm": 0.2546874864594375,
      "learning_rate": 4.94052379049345e-07,
      "loss": 1.4329,
      "step": 1763
    },
    {
      "epoch": 0.38769230769230767,
      "grad_norm": 0.25113981074630753,
      "learning_rate": 4.938483077606077e-07,
      "loss": 1.5257,
      "step": 1764
    },
    {
      "epoch": 0.3879120879120879,
      "grad_norm": 0.24634928100053097,
      "learning_rate": 4.936441845800848e-07,
      "loss": 1.5301,
      "step": 1765
    },
    {
      "epoch": 0.3881318681318681,
      "grad_norm": 0.27798264903802494,
      "learning_rate": 4.934400096050888e-07,
      "loss": 1.4742,
      "step": 1766
    },
    {
      "epoch": 0.38835164835164837,
      "grad_norm": 0.2595476892119092,
      "learning_rate": 4.93235782932957e-07,
      "loss": 1.4621,
      "step": 1767
    },
    {
      "epoch": 0.38857142857142857,
      "grad_norm": 0.25005757368351345,
      "learning_rate": 4.930315046610513e-07,
      "loss": 1.5062,
      "step": 1768
    },
    {
      "epoch": 0.38879120879120876,
      "grad_norm": 0.2475965750531368,
      "learning_rate": 4.928271748867587e-07,
      "loss": 1.4672,
      "step": 1769
    },
    {
      "epoch": 0.389010989010989,
      "grad_norm": 0.25355442753356366,
      "learning_rate": 4.926227937074897e-07,
      "loss": 1.4644,
      "step": 1770
    },
    {
      "epoch": 0.3892307692307692,
      "grad_norm": 0.2448924646560554,
      "learning_rate": 4.924183612206803e-07,
      "loss": 1.443,
      "step": 1771
    },
    {
      "epoch": 0.38945054945054947,
      "grad_norm": 0.27020637123372326,
      "learning_rate": 4.922138775237905e-07,
      "loss": 1.5124,
      "step": 1772
    },
    {
      "epoch": 0.38967032967032966,
      "grad_norm": 0.2429161843850925,
      "learning_rate": 4.92009342714305e-07,
      "loss": 1.4743,
      "step": 1773
    },
    {
      "epoch": 0.3898901098901099,
      "grad_norm": 0.252594183781848,
      "learning_rate": 4.918047568897325e-07,
      "loss": 1.4622,
      "step": 1774
    },
    {
      "epoch": 0.3901098901098901,
      "grad_norm": 0.24048192930828624,
      "learning_rate": 4.916001201476062e-07,
      "loss": 1.4698,
      "step": 1775
    },
    {
      "epoch": 0.3903296703296703,
      "grad_norm": 0.2422641983900522,
      "learning_rate": 4.913954325854836e-07,
      "loss": 1.4791,
      "step": 1776
    },
    {
      "epoch": 0.39054945054945056,
      "grad_norm": 0.25461158110345045,
      "learning_rate": 4.911906943009464e-07,
      "loss": 1.4913,
      "step": 1777
    },
    {
      "epoch": 0.39076923076923076,
      "grad_norm": 0.29321949968392325,
      "learning_rate": 4.909859053916006e-07,
      "loss": 1.4653,
      "step": 1778
    },
    {
      "epoch": 0.390989010989011,
      "grad_norm": 0.25663682866390836,
      "learning_rate": 4.907810659550759e-07,
      "loss": 1.5317,
      "step": 1779
    },
    {
      "epoch": 0.3912087912087912,
      "grad_norm": 0.24724952499445593,
      "learning_rate": 4.905761760890269e-07,
      "loss": 1.4515,
      "step": 1780
    },
    {
      "epoch": 0.3914285714285714,
      "grad_norm": 0.26411372708333347,
      "learning_rate": 4.903712358911313e-07,
      "loss": 1.4773,
      "step": 1781
    },
    {
      "epoch": 0.39164835164835166,
      "grad_norm": 0.24846185378366778,
      "learning_rate": 4.901662454590915e-07,
      "loss": 1.4742,
      "step": 1782
    },
    {
      "epoch": 0.39186813186813185,
      "grad_norm": 0.2668387394027364,
      "learning_rate": 4.899612048906334e-07,
      "loss": 1.4363,
      "step": 1783
    },
    {
      "epoch": 0.3920879120879121,
      "grad_norm": 0.2611779062632974,
      "learning_rate": 4.89756114283507e-07,
      "loss": 1.4804,
      "step": 1784
    },
    {
      "epoch": 0.3923076923076923,
      "grad_norm": 0.2677177467642176,
      "learning_rate": 4.895509737354864e-07,
      "loss": 1.454,
      "step": 1785
    },
    {
      "epoch": 0.39252747252747255,
      "grad_norm": 0.30826115217999084,
      "learning_rate": 4.893457833443692e-07,
      "loss": 1.4808,
      "step": 1786
    },
    {
      "epoch": 0.39274725274725275,
      "grad_norm": 0.2793667782925087,
      "learning_rate": 4.891405432079765e-07,
      "loss": 1.4545,
      "step": 1787
    },
    {
      "epoch": 0.39296703296703295,
      "grad_norm": 0.24327235516279463,
      "learning_rate": 4.889352534241537e-07,
      "loss": 1.4653,
      "step": 1788
    },
    {
      "epoch": 0.3931868131868132,
      "grad_norm": 0.25491835794406864,
      "learning_rate": 4.887299140907695e-07,
      "loss": 1.4747,
      "step": 1789
    },
    {
      "epoch": 0.3934065934065934,
      "grad_norm": 0.2540225940638022,
      "learning_rate": 4.885245253057165e-07,
      "loss": 1.5261,
      "step": 1790
    },
    {
      "epoch": 0.39362637362637365,
      "grad_norm": 0.26796825571352123,
      "learning_rate": 4.883190871669105e-07,
      "loss": 1.474,
      "step": 1791
    },
    {
      "epoch": 0.39384615384615385,
      "grad_norm": 0.2531601757943148,
      "learning_rate": 4.881135997722911e-07,
      "loss": 1.4956,
      "step": 1792
    },
    {
      "epoch": 0.39406593406593404,
      "grad_norm": 0.28754737023725047,
      "learning_rate": 4.879080632198215e-07,
      "loss": 1.4302,
      "step": 1793
    },
    {
      "epoch": 0.3942857142857143,
      "grad_norm": 0.2676420335773287,
      "learning_rate": 4.877024776074877e-07,
      "loss": 1.4763,
      "step": 1794
    },
    {
      "epoch": 0.3945054945054945,
      "grad_norm": 0.24716390305532507,
      "learning_rate": 4.874968430332998e-07,
      "loss": 1.4721,
      "step": 1795
    },
    {
      "epoch": 0.39472527472527474,
      "grad_norm": 0.23344381632056682,
      "learning_rate": 4.87291159595291e-07,
      "loss": 1.4322,
      "step": 1796
    },
    {
      "epoch": 0.39494505494505494,
      "grad_norm": 0.24377427229410936,
      "learning_rate": 4.870854273915178e-07,
      "loss": 1.4341,
      "step": 1797
    },
    {
      "epoch": 0.39516483516483514,
      "grad_norm": 0.24609118241522102,
      "learning_rate": 4.868796465200599e-07,
      "loss": 1.463,
      "step": 1798
    },
    {
      "epoch": 0.3953846153846154,
      "grad_norm": 0.26229641967712986,
      "learning_rate": 4.866738170790202e-07,
      "loss": 1.4789,
      "step": 1799
    },
    {
      "epoch": 0.3956043956043956,
      "grad_norm": 0.25776871029947374,
      "learning_rate": 4.864679391665245e-07,
      "loss": 1.5444,
      "step": 1800
    },
    {
      "epoch": 0.39582417582417584,
      "grad_norm": 0.23892518567766813,
      "learning_rate": 4.862620128807225e-07,
      "loss": 1.4507,
      "step": 1801
    },
    {
      "epoch": 0.39604395604395604,
      "grad_norm": 0.24817053150800028,
      "learning_rate": 4.860560383197859e-07,
      "loss": 1.4869,
      "step": 1802
    },
    {
      "epoch": 0.3962637362637363,
      "grad_norm": 0.25572874720505023,
      "learning_rate": 4.858500155819105e-07,
      "loss": 1.4768,
      "step": 1803
    },
    {
      "epoch": 0.3964835164835165,
      "grad_norm": 0.3168810818857704,
      "learning_rate": 4.856439447653142e-07,
      "loss": 1.4569,
      "step": 1804
    },
    {
      "epoch": 0.3967032967032967,
      "grad_norm": 0.24009418394509754,
      "learning_rate": 4.854378259682381e-07,
      "loss": 1.4379,
      "step": 1805
    },
    {
      "epoch": 0.39692307692307693,
      "grad_norm": 0.25161774876511417,
      "learning_rate": 4.852316592889464e-07,
      "loss": 1.4404,
      "step": 1806
    },
    {
      "epoch": 0.39714285714285713,
      "grad_norm": 0.25938765062006036,
      "learning_rate": 4.85025444825726e-07,
      "loss": 1.4366,
      "step": 1807
    },
    {
      "epoch": 0.3973626373626374,
      "grad_norm": 0.24821673122609372,
      "learning_rate": 4.848191826768863e-07,
      "loss": 1.4356,
      "step": 1808
    },
    {
      "epoch": 0.3975824175824176,
      "grad_norm": 0.2482062446031562,
      "learning_rate": 4.846128729407597e-07,
      "loss": 1.4395,
      "step": 1809
    },
    {
      "epoch": 0.3978021978021978,
      "grad_norm": 0.2669109598969336,
      "learning_rate": 4.844065157157014e-07,
      "loss": 1.4541,
      "step": 1810
    },
    {
      "epoch": 0.39802197802197803,
      "grad_norm": 0.25771016681731657,
      "learning_rate": 4.842001111000891e-07,
      "loss": 1.4473,
      "step": 1811
    },
    {
      "epoch": 0.3982417582417582,
      "grad_norm": 0.3009492967522756,
      "learning_rate": 4.839936591923229e-07,
      "loss": 1.4814,
      "step": 1812
    },
    {
      "epoch": 0.3984615384615385,
      "grad_norm": 0.2464490607634403,
      "learning_rate": 4.837871600908257e-07,
      "loss": 1.5183,
      "step": 1813
    },
    {
      "epoch": 0.3986813186813187,
      "grad_norm": 0.259549663136423,
      "learning_rate": 4.835806138940428e-07,
      "loss": 1.4927,
      "step": 1814
    },
    {
      "epoch": 0.3989010989010989,
      "grad_norm": 0.2566912365495615,
      "learning_rate": 4.833740207004421e-07,
      "loss": 1.4423,
      "step": 1815
    },
    {
      "epoch": 0.3991208791208791,
      "grad_norm": 0.25162225407409183,
      "learning_rate": 4.831673806085138e-07,
      "loss": 1.4719,
      "step": 1816
    },
    {
      "epoch": 0.3993406593406593,
      "grad_norm": 0.2565978320181227,
      "learning_rate": 4.829606937167702e-07,
      "loss": 1.4203,
      "step": 1817
    },
    {
      "epoch": 0.3995604395604396,
      "grad_norm": 0.2447236818504802,
      "learning_rate": 4.827539601237463e-07,
      "loss": 1.4745,
      "step": 1818
    },
    {
      "epoch": 0.39978021978021977,
      "grad_norm": 0.25277170852857506,
      "learning_rate": 4.825471799279993e-07,
      "loss": 1.4642,
      "step": 1819
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.24220794243076613,
      "learning_rate": 4.823403532281084e-07,
      "loss": 1.5,
      "step": 1820
    },
    {
      "epoch": 0.4002197802197802,
      "grad_norm": 0.2427399800900977,
      "learning_rate": 4.821334801226752e-07,
      "loss": 1.4454,
      "step": 1821
    },
    {
      "epoch": 0.4004395604395604,
      "grad_norm": 0.24036850678664107,
      "learning_rate": 4.819265607103233e-07,
      "loss": 1.5058,
      "step": 1822
    },
    {
      "epoch": 0.40065934065934067,
      "grad_norm": 0.26847581950085464,
      "learning_rate": 4.817195950896983e-07,
      "loss": 1.393,
      "step": 1823
    },
    {
      "epoch": 0.40087912087912086,
      "grad_norm": 0.268721336380857,
      "learning_rate": 4.815125833594679e-07,
      "loss": 1.4737,
      "step": 1824
    },
    {
      "epoch": 0.4010989010989011,
      "grad_norm": 0.2910363742169384,
      "learning_rate": 4.813055256183222e-07,
      "loss": 1.5296,
      "step": 1825
    },
    {
      "epoch": 0.4013186813186813,
      "grad_norm": 0.24118564870426512,
      "learning_rate": 4.810984219649725e-07,
      "loss": 1.4804,
      "step": 1826
    },
    {
      "epoch": 0.4015384615384615,
      "grad_norm": 0.2566750028500243,
      "learning_rate": 4.808912724981524e-07,
      "loss": 1.4949,
      "step": 1827
    },
    {
      "epoch": 0.40175824175824176,
      "grad_norm": 0.23867865799133473,
      "learning_rate": 4.806840773166171e-07,
      "loss": 1.4066,
      "step": 1828
    },
    {
      "epoch": 0.40197802197802196,
      "grad_norm": 0.25533959891940117,
      "learning_rate": 4.804768365191441e-07,
      "loss": 1.5586,
      "step": 1829
    },
    {
      "epoch": 0.4021978021978022,
      "grad_norm": 0.2422944624106141,
      "learning_rate": 4.802695502045322e-07,
      "loss": 1.4104,
      "step": 1830
    },
    {
      "epoch": 0.4024175824175824,
      "grad_norm": 0.24466760588270278,
      "learning_rate": 4.80062218471602e-07,
      "loss": 1.4668,
      "step": 1831
    },
    {
      "epoch": 0.40263736263736266,
      "grad_norm": 0.2466937832986439,
      "learning_rate": 4.798548414191957e-07,
      "loss": 1.4212,
      "step": 1832
    },
    {
      "epoch": 0.40285714285714286,
      "grad_norm": 0.24768306549436672,
      "learning_rate": 4.796474191461773e-07,
      "loss": 1.4515,
      "step": 1833
    },
    {
      "epoch": 0.40307692307692305,
      "grad_norm": 0.2504352894051246,
      "learning_rate": 4.794399517514322e-07,
      "loss": 1.4858,
      "step": 1834
    },
    {
      "epoch": 0.4032967032967033,
      "grad_norm": 0.24699744571501392,
      "learning_rate": 4.792324393338674e-07,
      "loss": 1.5157,
      "step": 1835
    },
    {
      "epoch": 0.4035164835164835,
      "grad_norm": 0.24744382663721826,
      "learning_rate": 4.79024881992411e-07,
      "loss": 1.4565,
      "step": 1836
    },
    {
      "epoch": 0.40373626373626376,
      "grad_norm": 0.26316227935316405,
      "learning_rate": 4.788172798260132e-07,
      "loss": 1.4838,
      "step": 1837
    },
    {
      "epoch": 0.40395604395604395,
      "grad_norm": 0.25257085263895485,
      "learning_rate": 4.786096329336451e-07,
      "loss": 1.4919,
      "step": 1838
    },
    {
      "epoch": 0.40417582417582415,
      "grad_norm": 0.2625332035565822,
      "learning_rate": 4.784019414142992e-07,
      "loss": 1.4807,
      "step": 1839
    },
    {
      "epoch": 0.4043956043956044,
      "grad_norm": 0.24705421740064612,
      "learning_rate": 4.781942053669891e-07,
      "loss": 1.4699,
      "step": 1840
    },
    {
      "epoch": 0.4046153846153846,
      "grad_norm": 0.2429821361794914,
      "learning_rate": 4.7798642489075e-07,
      "loss": 1.5306,
      "step": 1841
    },
    {
      "epoch": 0.40483516483516485,
      "grad_norm": 0.26020097251838,
      "learning_rate": 4.777786000846383e-07,
      "loss": 1.4924,
      "step": 1842
    },
    {
      "epoch": 0.40505494505494505,
      "grad_norm": 0.24070971549065806,
      "learning_rate": 4.77570731047731e-07,
      "loss": 1.4587,
      "step": 1843
    },
    {
      "epoch": 0.4052747252747253,
      "grad_norm": 0.23694437718701358,
      "learning_rate": 4.773628178791265e-07,
      "loss": 1.4419,
      "step": 1844
    },
    {
      "epoch": 0.4054945054945055,
      "grad_norm": 0.24633286127674098,
      "learning_rate": 4.771548606779445e-07,
      "loss": 1.4264,
      "step": 1845
    },
    {
      "epoch": 0.4057142857142857,
      "grad_norm": 0.2585750366583406,
      "learning_rate": 4.769468595433256e-07,
      "loss": 1.533,
      "step": 1846
    },
    {
      "epoch": 0.40593406593406595,
      "grad_norm": 0.2622541535395647,
      "learning_rate": 4.767388145744307e-07,
      "loss": 1.4448,
      "step": 1847
    },
    {
      "epoch": 0.40615384615384614,
      "grad_norm": 0.2525586985577036,
      "learning_rate": 4.7653072587044237e-07,
      "loss": 1.4515,
      "step": 1848
    },
    {
      "epoch": 0.4063736263736264,
      "grad_norm": 0.24274686923220623,
      "learning_rate": 4.763225935305637e-07,
      "loss": 1.4441,
      "step": 1849
    },
    {
      "epoch": 0.4065934065934066,
      "grad_norm": 0.28403044996668436,
      "learning_rate": 4.7611441765401884e-07,
      "loss": 1.507,
      "step": 1850
    },
    {
      "epoch": 0.4068131868131868,
      "grad_norm": 0.24752559644983235,
      "learning_rate": 4.7590619834005223e-07,
      "loss": 1.4437,
      "step": 1851
    },
    {
      "epoch": 0.40703296703296704,
      "grad_norm": 0.2554203155855423,
      "learning_rate": 4.7569793568792933e-07,
      "loss": 1.4766,
      "step": 1852
    },
    {
      "epoch": 0.40725274725274724,
      "grad_norm": 0.25857678651541904,
      "learning_rate": 4.754896297969364e-07,
      "loss": 1.4844,
      "step": 1853
    },
    {
      "epoch": 0.4074725274725275,
      "grad_norm": 0.23871671831128147,
      "learning_rate": 4.7528128076637984e-07,
      "loss": 1.4997,
      "step": 1854
    },
    {
      "epoch": 0.4076923076923077,
      "grad_norm": 0.250982771013428,
      "learning_rate": 4.750728886955871e-07,
      "loss": 1.4352,
      "step": 1855
    },
    {
      "epoch": 0.40791208791208794,
      "grad_norm": 0.2492415405681126,
      "learning_rate": 4.748644536839058e-07,
      "loss": 1.4324,
      "step": 1856
    },
    {
      "epoch": 0.40813186813186814,
      "grad_norm": 0.31408680888199064,
      "learning_rate": 4.746559758307045e-07,
      "loss": 1.4883,
      "step": 1857
    },
    {
      "epoch": 0.40835164835164833,
      "grad_norm": 0.252977338230396,
      "learning_rate": 4.7444745523537166e-07,
      "loss": 1.4003,
      "step": 1858
    },
    {
      "epoch": 0.4085714285714286,
      "grad_norm": 0.27105566114926655,
      "learning_rate": 4.74238891997316e-07,
      "loss": 1.4994,
      "step": 1859
    },
    {
      "epoch": 0.4087912087912088,
      "grad_norm": 0.25952788816711503,
      "learning_rate": 4.7403028621596746e-07,
      "loss": 1.451,
      "step": 1860
    },
    {
      "epoch": 0.40901098901098903,
      "grad_norm": 0.2541118783054669,
      "learning_rate": 4.738216379907753e-07,
      "loss": 1.5014,
      "step": 1861
    },
    {
      "epoch": 0.40923076923076923,
      "grad_norm": 0.24630810400822947,
      "learning_rate": 4.7361294742120965e-07,
      "loss": 1.4538,
      "step": 1862
    },
    {
      "epoch": 0.40945054945054943,
      "grad_norm": 0.23836644075251986,
      "learning_rate": 4.7340421460676044e-07,
      "loss": 1.499,
      "step": 1863
    },
    {
      "epoch": 0.4096703296703297,
      "grad_norm": 0.2859251633068821,
      "learning_rate": 4.731954396469379e-07,
      "loss": 1.4851,
      "step": 1864
    },
    {
      "epoch": 0.4098901098901099,
      "grad_norm": 0.251065508090773,
      "learning_rate": 4.729866226412724e-07,
      "loss": 1.4631,
      "step": 1865
    },
    {
      "epoch": 0.41010989010989013,
      "grad_norm": 0.2577852475566407,
      "learning_rate": 4.727777636893142e-07,
      "loss": 1.4355,
      "step": 1866
    },
    {
      "epoch": 0.4103296703296703,
      "grad_norm": 0.2649747543641412,
      "learning_rate": 4.725688628906338e-07,
      "loss": 1.4417,
      "step": 1867
    },
    {
      "epoch": 0.4105494505494505,
      "grad_norm": 0.2972787725085871,
      "learning_rate": 4.723599203448213e-07,
      "loss": 1.5141,
      "step": 1868
    },
    {
      "epoch": 0.4107692307692308,
      "grad_norm": 0.2505888516758264,
      "learning_rate": 4.7215093615148703e-07,
      "loss": 1.4643,
      "step": 1869
    },
    {
      "epoch": 0.41098901098901097,
      "grad_norm": 0.2555008453693315,
      "learning_rate": 4.7194191041026096e-07,
      "loss": 1.4753,
      "step": 1870
    },
    {
      "epoch": 0.4112087912087912,
      "grad_norm": 0.24984656958376497,
      "learning_rate": 4.7173284322079305e-07,
      "loss": 1.4604,
      "step": 1871
    },
    {
      "epoch": 0.4114285714285714,
      "grad_norm": 0.24267064075432038,
      "learning_rate": 4.715237346827529e-07,
      "loss": 1.468,
      "step": 1872
    },
    {
      "epoch": 0.4116483516483517,
      "grad_norm": 0.2474257707883197,
      "learning_rate": 4.7131458489582984e-07,
      "loss": 1.4087,
      "step": 1873
    },
    {
      "epoch": 0.41186813186813187,
      "grad_norm": 0.2595786871517908,
      "learning_rate": 4.711053939597328e-07,
      "loss": 1.4757,
      "step": 1874
    },
    {
      "epoch": 0.41208791208791207,
      "grad_norm": 0.26515303845851107,
      "learning_rate": 4.7089616197419055e-07,
      "loss": 1.5202,
      "step": 1875
    },
    {
      "epoch": 0.4123076923076923,
      "grad_norm": 0.2580574704773794,
      "learning_rate": 4.706868890389511e-07,
      "loss": 1.4987,
      "step": 1876
    },
    {
      "epoch": 0.4125274725274725,
      "grad_norm": 0.24629621536998408,
      "learning_rate": 4.704775752537823e-07,
      "loss": 1.4456,
      "step": 1877
    },
    {
      "epoch": 0.41274725274725277,
      "grad_norm": 0.27158152619748616,
      "learning_rate": 4.702682207184712e-07,
      "loss": 1.4959,
      "step": 1878
    },
    {
      "epoch": 0.41296703296703297,
      "grad_norm": 0.242480743299886,
      "learning_rate": 4.7005882553282453e-07,
      "loss": 1.4061,
      "step": 1879
    },
    {
      "epoch": 0.41318681318681316,
      "grad_norm": 0.2561528857614439,
      "learning_rate": 4.6984938979666826e-07,
      "loss": 1.456,
      "step": 1880
    },
    {
      "epoch": 0.4134065934065934,
      "grad_norm": 0.26569408665037497,
      "learning_rate": 4.6963991360984765e-07,
      "loss": 1.5047,
      "step": 1881
    },
    {
      "epoch": 0.4136263736263736,
      "grad_norm": 0.25976395589489604,
      "learning_rate": 4.694303970722274e-07,
      "loss": 1.4173,
      "step": 1882
    },
    {
      "epoch": 0.41384615384615386,
      "grad_norm": 0.2509515485922121,
      "learning_rate": 4.692208402836912e-07,
      "loss": 1.4098,
      "step": 1883
    },
    {
      "epoch": 0.41406593406593406,
      "grad_norm": 0.25501761312969995,
      "learning_rate": 4.6901124334414223e-07,
      "loss": 1.5109,
      "step": 1884
    },
    {
      "epoch": 0.4142857142857143,
      "grad_norm": 0.29388399646471647,
      "learning_rate": 4.6880160635350265e-07,
      "loss": 1.4561,
      "step": 1885
    },
    {
      "epoch": 0.4145054945054945,
      "grad_norm": 0.25062995423747175,
      "learning_rate": 4.6859192941171365e-07,
      "loss": 1.4908,
      "step": 1886
    },
    {
      "epoch": 0.4147252747252747,
      "grad_norm": 0.24776975594772918,
      "learning_rate": 4.6838221261873564e-07,
      "loss": 1.4095,
      "step": 1887
    },
    {
      "epoch": 0.41494505494505496,
      "grad_norm": 0.24406414113091413,
      "learning_rate": 4.6817245607454785e-07,
      "loss": 1.4596,
      "step": 1888
    },
    {
      "epoch": 0.41516483516483516,
      "grad_norm": 0.24152548526912102,
      "learning_rate": 4.679626598791485e-07,
      "loss": 1.4641,
      "step": 1889
    },
    {
      "epoch": 0.4153846153846154,
      "grad_norm": 0.2591887155483339,
      "learning_rate": 4.6775282413255485e-07,
      "loss": 1.4536,
      "step": 1890
    },
    {
      "epoch": 0.4156043956043956,
      "grad_norm": 0.25372519647690656,
      "learning_rate": 4.67542948934803e-07,
      "loss": 1.4672,
      "step": 1891
    },
    {
      "epoch": 0.4158241758241758,
      "grad_norm": 0.24984707519866892,
      "learning_rate": 4.6733303438594764e-07,
      "loss": 1.4889,
      "step": 1892
    },
    {
      "epoch": 0.41604395604395605,
      "grad_norm": 0.2465455207121106,
      "learning_rate": 4.6712308058606217e-07,
      "loss": 1.4577,
      "step": 1893
    },
    {
      "epoch": 0.41626373626373625,
      "grad_norm": 0.2866705542456264,
      "learning_rate": 4.669130876352393e-07,
      "loss": 1.478,
      "step": 1894
    },
    {
      "epoch": 0.4164835164835165,
      "grad_norm": 0.2449021470786593,
      "learning_rate": 4.6670305563358984e-07,
      "loss": 1.4696,
      "step": 1895
    },
    {
      "epoch": 0.4167032967032967,
      "grad_norm": 0.24194367792760602,
      "learning_rate": 4.6649298468124336e-07,
      "loss": 1.4133,
      "step": 1896
    },
    {
      "epoch": 0.4169230769230769,
      "grad_norm": 0.23805554230224935,
      "learning_rate": 4.662828748783479e-07,
      "loss": 1.4735,
      "step": 1897
    },
    {
      "epoch": 0.41714285714285715,
      "grad_norm": 0.24487829405021447,
      "learning_rate": 4.6607272632507026e-07,
      "loss": 1.4194,
      "step": 1898
    },
    {
      "epoch": 0.41736263736263735,
      "grad_norm": 0.24576913269889125,
      "learning_rate": 4.658625391215956e-07,
      "loss": 1.4756,
      "step": 1899
    },
    {
      "epoch": 0.4175824175824176,
      "grad_norm": 0.24345437401255496,
      "learning_rate": 4.6565231336812736e-07,
      "loss": 1.489,
      "step": 1900
    },
    {
      "epoch": 0.4178021978021978,
      "grad_norm": 0.24951438935633438,
      "learning_rate": 4.654420491648876e-07,
      "loss": 1.4564,
      "step": 1901
    },
    {
      "epoch": 0.41802197802197805,
      "grad_norm": 0.25116043650662734,
      "learning_rate": 4.652317466121166e-07,
      "loss": 1.5066,
      "step": 1902
    },
    {
      "epoch": 0.41824175824175824,
      "grad_norm": 0.26962698285180703,
      "learning_rate": 4.6502140581007296e-07,
      "loss": 1.474,
      "step": 1903
    },
    {
      "epoch": 0.41846153846153844,
      "grad_norm": 0.24739657362303075,
      "learning_rate": 4.648110268590333e-07,
      "loss": 1.4814,
      "step": 1904
    },
    {
      "epoch": 0.4186813186813187,
      "grad_norm": 0.25940239165573875,
      "learning_rate": 4.646006098592928e-07,
      "loss": 1.5056,
      "step": 1905
    },
    {
      "epoch": 0.4189010989010989,
      "grad_norm": 0.24657560022021907,
      "learning_rate": 4.643901549111645e-07,
      "loss": 1.4625,
      "step": 1906
    },
    {
      "epoch": 0.41912087912087914,
      "grad_norm": 0.26164647648356376,
      "learning_rate": 4.641796621149796e-07,
      "loss": 1.4912,
      "step": 1907
    },
    {
      "epoch": 0.41934065934065934,
      "grad_norm": 0.2457366206140155,
      "learning_rate": 4.639691315710874e-07,
      "loss": 1.479,
      "step": 1908
    },
    {
      "epoch": 0.41956043956043954,
      "grad_norm": 0.24829951402302813,
      "learning_rate": 4.637585633798552e-07,
      "loss": 1.4507,
      "step": 1909
    },
    {
      "epoch": 0.4197802197802198,
      "grad_norm": 0.24137696225989796,
      "learning_rate": 4.6354795764166805e-07,
      "loss": 1.4788,
      "step": 1910
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.25059062683734157,
      "learning_rate": 4.633373144569293e-07,
      "loss": 1.4602,
      "step": 1911
    },
    {
      "epoch": 0.42021978021978024,
      "grad_norm": 0.2502811070900914,
      "learning_rate": 4.6312663392605954e-07,
      "loss": 1.4299,
      "step": 1912
    },
    {
      "epoch": 0.42043956043956043,
      "grad_norm": 0.24843538159409193,
      "learning_rate": 4.629159161494979e-07,
      "loss": 1.4624,
      "step": 1913
    },
    {
      "epoch": 0.4206593406593407,
      "grad_norm": 0.25679900310546494,
      "learning_rate": 4.6270516122770063e-07,
      "loss": 1.523,
      "step": 1914
    },
    {
      "epoch": 0.4208791208791209,
      "grad_norm": 0.2551073468372462,
      "learning_rate": 4.624943692611422e-07,
      "loss": 1.4654,
      "step": 1915
    },
    {
      "epoch": 0.4210989010989011,
      "grad_norm": 0.25520050194715943,
      "learning_rate": 4.622835403503143e-07,
      "loss": 1.4168,
      "step": 1916
    },
    {
      "epoch": 0.42131868131868133,
      "grad_norm": 0.2502794084468964,
      "learning_rate": 4.6207267459572644e-07,
      "loss": 1.434,
      "step": 1917
    },
    {
      "epoch": 0.42153846153846153,
      "grad_norm": 0.2711830636029782,
      "learning_rate": 4.6186177209790574e-07,
      "loss": 1.511,
      "step": 1918
    },
    {
      "epoch": 0.4217582417582418,
      "grad_norm": 0.2563225325946048,
      "learning_rate": 4.616508329573968e-07,
      "loss": 1.4495,
      "step": 1919
    },
    {
      "epoch": 0.421978021978022,
      "grad_norm": 0.3447193135994813,
      "learning_rate": 4.614398572747616e-07,
      "loss": 1.4517,
      "step": 1920
    },
    {
      "epoch": 0.4221978021978022,
      "grad_norm": 0.2679357815994255,
      "learning_rate": 4.612288451505797e-07,
      "loss": 1.4658,
      "step": 1921
    },
    {
      "epoch": 0.4224175824175824,
      "grad_norm": 0.2526983489954876,
      "learning_rate": 4.6101779668544784e-07,
      "loss": 1.438,
      "step": 1922
    },
    {
      "epoch": 0.4226373626373626,
      "grad_norm": 0.2425244700743695,
      "learning_rate": 4.608067119799802e-07,
      "loss": 1.4182,
      "step": 1923
    },
    {
      "epoch": 0.4228571428571429,
      "grad_norm": 0.2480925558781557,
      "learning_rate": 4.6059559113480823e-07,
      "loss": 1.4514,
      "step": 1924
    },
    {
      "epoch": 0.4230769230769231,
      "grad_norm": 0.24793658474452754,
      "learning_rate": 4.603844342505807e-07,
      "loss": 1.441,
      "step": 1925
    },
    {
      "epoch": 0.42329670329670327,
      "grad_norm": 0.24733060094121687,
      "learning_rate": 4.6017324142796337e-07,
      "loss": 1.4692,
      "step": 1926
    },
    {
      "epoch": 0.4235164835164835,
      "grad_norm": 0.2509952033182183,
      "learning_rate": 4.5996201276763916e-07,
      "loss": 1.434,
      "step": 1927
    },
    {
      "epoch": 0.4237362637362637,
      "grad_norm": 0.24111925275808668,
      "learning_rate": 4.5975074837030834e-07,
      "loss": 1.4882,
      "step": 1928
    },
    {
      "epoch": 0.42395604395604397,
      "grad_norm": 0.24589477494009745,
      "learning_rate": 4.595394483366878e-07,
      "loss": 1.4403,
      "step": 1929
    },
    {
      "epoch": 0.42417582417582417,
      "grad_norm": 0.23500968708420283,
      "learning_rate": 4.5932811276751186e-07,
      "loss": 1.5122,
      "step": 1930
    },
    {
      "epoch": 0.4243956043956044,
      "grad_norm": 0.2509348334848281,
      "learning_rate": 4.5911674176353126e-07,
      "loss": 1.4871,
      "step": 1931
    },
    {
      "epoch": 0.4246153846153846,
      "grad_norm": 0.2487398588487112,
      "learning_rate": 4.589053354255142e-07,
      "loss": 1.4703,
      "step": 1932
    },
    {
      "epoch": 0.4248351648351648,
      "grad_norm": 0.264975782675525,
      "learning_rate": 4.586938938542455e-07,
      "loss": 1.4318,
      "step": 1933
    },
    {
      "epoch": 0.42505494505494507,
      "grad_norm": 0.24653741023793363,
      "learning_rate": 4.584824171505265e-07,
      "loss": 1.4367,
      "step": 1934
    },
    {
      "epoch": 0.42527472527472526,
      "grad_norm": 0.244534014831125,
      "learning_rate": 4.582709054151755e-07,
      "loss": 1.51,
      "step": 1935
    },
    {
      "epoch": 0.4254945054945055,
      "grad_norm": 0.2434156563546049,
      "learning_rate": 4.5805935874902776e-07,
      "loss": 1.4115,
      "step": 1936
    },
    {
      "epoch": 0.4257142857142857,
      "grad_norm": 0.4409375634032495,
      "learning_rate": 4.578477772529349e-07,
      "loss": 1.449,
      "step": 1937
    },
    {
      "epoch": 0.4259340659340659,
      "grad_norm": 0.28008372588235186,
      "learning_rate": 4.5763616102776506e-07,
      "loss": 1.4542,
      "step": 1938
    },
    {
      "epoch": 0.42615384615384616,
      "grad_norm": 0.24733105124087043,
      "learning_rate": 4.5742451017440327e-07,
      "loss": 1.4666,
      "step": 1939
    },
    {
      "epoch": 0.42637362637362636,
      "grad_norm": 0.26214161289244853,
      "learning_rate": 4.5721282479375063e-07,
      "loss": 1.4603,
      "step": 1940
    },
    {
      "epoch": 0.4265934065934066,
      "grad_norm": 0.2600237795584968,
      "learning_rate": 4.570011049867252e-07,
      "loss": 1.4887,
      "step": 1941
    },
    {
      "epoch": 0.4268131868131868,
      "grad_norm": 0.2635085113676006,
      "learning_rate": 4.56789350854261e-07,
      "loss": 1.4473,
      "step": 1942
    },
    {
      "epoch": 0.42703296703296706,
      "grad_norm": 0.25549916126245564,
      "learning_rate": 4.565775624973087e-07,
      "loss": 1.5072,
      "step": 1943
    },
    {
      "epoch": 0.42725274725274726,
      "grad_norm": 0.2529790262311076,
      "learning_rate": 4.563657400168352e-07,
      "loss": 1.4892,
      "step": 1944
    },
    {
      "epoch": 0.42747252747252745,
      "grad_norm": 0.23877180896457764,
      "learning_rate": 4.561538835138238e-07,
      "loss": 1.4678,
      "step": 1945
    },
    {
      "epoch": 0.4276923076923077,
      "grad_norm": 0.252072813103344,
      "learning_rate": 4.559419930892736e-07,
      "loss": 1.4576,
      "step": 1946
    },
    {
      "epoch": 0.4279120879120879,
      "grad_norm": 0.267557026769763,
      "learning_rate": 4.557300688442003e-07,
      "loss": 1.5399,
      "step": 1947
    },
    {
      "epoch": 0.42813186813186815,
      "grad_norm": 0.2853942500311004,
      "learning_rate": 4.555181108796356e-07,
      "loss": 1.4752,
      "step": 1948
    },
    {
      "epoch": 0.42835164835164835,
      "grad_norm": 0.26206350182419685,
      "learning_rate": 4.553061192966273e-07,
      "loss": 1.5009,
      "step": 1949
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.24792865769176314,
      "learning_rate": 4.5509409419623896e-07,
      "loss": 1.4927,
      "step": 1950
    },
    {
      "epoch": 0.4287912087912088,
      "grad_norm": 0.3788666994107071,
      "learning_rate": 4.548820356795507e-07,
      "loss": 1.5395,
      "step": 1951
    },
    {
      "epoch": 0.429010989010989,
      "grad_norm": 0.250827362039741,
      "learning_rate": 4.5466994384765794e-07,
      "loss": 1.4867,
      "step": 1952
    },
    {
      "epoch": 0.42923076923076925,
      "grad_norm": 0.25333586193161073,
      "learning_rate": 4.5445781880167235e-07,
      "loss": 1.4877,
      "step": 1953
    },
    {
      "epoch": 0.42945054945054945,
      "grad_norm": 0.33638463275251274,
      "learning_rate": 4.5424566064272126e-07,
      "loss": 1.439,
      "step": 1954
    },
    {
      "epoch": 0.42967032967032964,
      "grad_norm": 0.24798145046955322,
      "learning_rate": 4.5403346947194805e-07,
      "loss": 1.4297,
      "step": 1955
    },
    {
      "epoch": 0.4298901098901099,
      "grad_norm": 0.266951146502588,
      "learning_rate": 4.5382124539051156e-07,
      "loss": 1.4988,
      "step": 1956
    },
    {
      "epoch": 0.4301098901098901,
      "grad_norm": 0.2487390391782417,
      "learning_rate": 4.5360898849958626e-07,
      "loss": 1.4884,
      "step": 1957
    },
    {
      "epoch": 0.43032967032967034,
      "grad_norm": 0.2622662025890682,
      "learning_rate": 4.533966989003626e-07,
      "loss": 1.4731,
      "step": 1958
    },
    {
      "epoch": 0.43054945054945054,
      "grad_norm": 0.24065722376539453,
      "learning_rate": 4.5318437669404633e-07,
      "loss": 1.4385,
      "step": 1959
    },
    {
      "epoch": 0.4307692307692308,
      "grad_norm": 0.2482545867959584,
      "learning_rate": 4.529720219818589e-07,
      "loss": 1.4443,
      "step": 1960
    },
    {
      "epoch": 0.430989010989011,
      "grad_norm": 0.25570330945984554,
      "learning_rate": 4.527596348650373e-07,
      "loss": 1.4119,
      "step": 1961
    },
    {
      "epoch": 0.4312087912087912,
      "grad_norm": 0.2410226952956021,
      "learning_rate": 4.525472154448337e-07,
      "loss": 1.4648,
      "step": 1962
    },
    {
      "epoch": 0.43142857142857144,
      "grad_norm": 0.25257701450368264,
      "learning_rate": 4.5233476382251605e-07,
      "loss": 1.5602,
      "step": 1963
    },
    {
      "epoch": 0.43164835164835164,
      "grad_norm": 0.2346379087711012,
      "learning_rate": 4.5212228009936735e-07,
      "loss": 1.4428,
      "step": 1964
    },
    {
      "epoch": 0.4318681318681319,
      "grad_norm": 0.3076212419062858,
      "learning_rate": 4.519097643766859e-07,
      "loss": 1.4448,
      "step": 1965
    },
    {
      "epoch": 0.4320879120879121,
      "grad_norm": 0.2558029366853796,
      "learning_rate": 4.5169721675578544e-07,
      "loss": 1.5036,
      "step": 1966
    },
    {
      "epoch": 0.4323076923076923,
      "grad_norm": 0.2523538767794331,
      "learning_rate": 4.5148463733799495e-07,
      "loss": 1.4498,
      "step": 1967
    },
    {
      "epoch": 0.43252747252747253,
      "grad_norm": 0.2730193900422513,
      "learning_rate": 4.5127202622465826e-07,
      "loss": 1.4951,
      "step": 1968
    },
    {
      "epoch": 0.43274725274725273,
      "grad_norm": 0.2706074518453508,
      "learning_rate": 4.5105938351713463e-07,
      "loss": 1.4692,
      "step": 1969
    },
    {
      "epoch": 0.432967032967033,
      "grad_norm": 0.2528399436655525,
      "learning_rate": 4.508467093167983e-07,
      "loss": 1.4844,
      "step": 1970
    },
    {
      "epoch": 0.4331868131868132,
      "grad_norm": 0.25125033559172794,
      "learning_rate": 4.5063400372503836e-07,
      "loss": 1.538,
      "step": 1971
    },
    {
      "epoch": 0.43340659340659343,
      "grad_norm": 0.24891370801079335,
      "learning_rate": 4.5042126684325906e-07,
      "loss": 1.3944,
      "step": 1972
    },
    {
      "epoch": 0.43362637362637363,
      "grad_norm": 0.24812753341113256,
      "learning_rate": 4.502084987728794e-07,
      "loss": 1.5183,
      "step": 1973
    },
    {
      "epoch": 0.4338461538461538,
      "grad_norm": 0.2701235151023572,
      "learning_rate": 4.4999569961533354e-07,
      "loss": 1.4777,
      "step": 1974
    },
    {
      "epoch": 0.4340659340659341,
      "grad_norm": 0.24538484487503745,
      "learning_rate": 4.4978286947206996e-07,
      "loss": 1.4778,
      "step": 1975
    },
    {
      "epoch": 0.4342857142857143,
      "grad_norm": 0.26435441514855984,
      "learning_rate": 4.495700084445526e-07,
      "loss": 1.4732,
      "step": 1976
    },
    {
      "epoch": 0.4345054945054945,
      "grad_norm": 0.2889115556355036,
      "learning_rate": 4.4935711663425935e-07,
      "loss": 1.4655,
      "step": 1977
    },
    {
      "epoch": 0.4347252747252747,
      "grad_norm": 0.24587544643910925,
      "learning_rate": 4.491441941426834e-07,
      "loss": 1.4345,
      "step": 1978
    },
    {
      "epoch": 0.4349450549450549,
      "grad_norm": 0.2634869755023885,
      "learning_rate": 4.489312410713323e-07,
      "loss": 1.4412,
      "step": 1979
    },
    {
      "epoch": 0.4351648351648352,
      "grad_norm": 0.2472304319512106,
      "learning_rate": 4.4871825752172813e-07,
      "loss": 1.4419,
      "step": 1980
    },
    {
      "epoch": 0.43538461538461537,
      "grad_norm": 0.2439296820321806,
      "learning_rate": 4.485052435954076e-07,
      "loss": 1.4787,
      "step": 1981
    },
    {
      "epoch": 0.4356043956043956,
      "grad_norm": 0.2420349645564503,
      "learning_rate": 4.4829219939392197e-07,
      "loss": 1.4406,
      "step": 1982
    },
    {
      "epoch": 0.4358241758241758,
      "grad_norm": 0.24806407871145997,
      "learning_rate": 4.4807912501883677e-07,
      "loss": 1.4712,
      "step": 1983
    },
    {
      "epoch": 0.43604395604395607,
      "grad_norm": 0.25960102110521244,
      "learning_rate": 4.47866020571732e-07,
      "loss": 1.4789,
      "step": 1984
    },
    {
      "epoch": 0.43626373626373627,
      "grad_norm": 0.31118938866936896,
      "learning_rate": 4.476528861542019e-07,
      "loss": 1.518,
      "step": 1985
    },
    {
      "epoch": 0.43648351648351646,
      "grad_norm": 0.24322874340836062,
      "learning_rate": 4.474397218678554e-07,
      "loss": 1.4709,
      "step": 1986
    },
    {
      "epoch": 0.4367032967032967,
      "grad_norm": 0.25468814898355213,
      "learning_rate": 4.4722652781431483e-07,
      "loss": 1.4672,
      "step": 1987
    },
    {
      "epoch": 0.4369230769230769,
      "grad_norm": 0.2557318512527907,
      "learning_rate": 4.4701330409521767e-07,
      "loss": 1.4958,
      "step": 1988
    },
    {
      "epoch": 0.43714285714285717,
      "grad_norm": 0.24240681824502205,
      "learning_rate": 4.4680005081221494e-07,
      "loss": 1.4409,
      "step": 1989
    },
    {
      "epoch": 0.43736263736263736,
      "grad_norm": 0.5309155135537983,
      "learning_rate": 4.46586768066972e-07,
      "loss": 1.5051,
      "step": 1990
    },
    {
      "epoch": 0.43758241758241756,
      "grad_norm": 0.30257411107460197,
      "learning_rate": 4.463734559611681e-07,
      "loss": 1.461,
      "step": 1991
    },
    {
      "epoch": 0.4378021978021978,
      "grad_norm": 0.24386625263667058,
      "learning_rate": 4.461601145964964e-07,
      "loss": 1.5288,
      "step": 1992
    },
    {
      "epoch": 0.438021978021978,
      "grad_norm": 0.29251326962414803,
      "learning_rate": 4.459467440746645e-07,
      "loss": 1.4851,
      "step": 1993
    },
    {
      "epoch": 0.43824175824175826,
      "grad_norm": 0.25691611671310566,
      "learning_rate": 4.457333444973934e-07,
      "loss": 1.4672,
      "step": 1994
    },
    {
      "epoch": 0.43846153846153846,
      "grad_norm": 0.24535319812993825,
      "learning_rate": 4.4551991596641816e-07,
      "loss": 1.4394,
      "step": 1995
    },
    {
      "epoch": 0.43868131868131865,
      "grad_norm": 0.2843181282489842,
      "learning_rate": 4.4530645858348756e-07,
      "loss": 1.4789,
      "step": 1996
    },
    {
      "epoch": 0.4389010989010989,
      "grad_norm": 0.2820459116215806,
      "learning_rate": 4.450929724503642e-07,
      "loss": 1.4839,
      "step": 1997
    },
    {
      "epoch": 0.4391208791208791,
      "grad_norm": 0.25298341155850945,
      "learning_rate": 4.448794576688244e-07,
      "loss": 1.4933,
      "step": 1998
    },
    {
      "epoch": 0.43934065934065936,
      "grad_norm": 0.288869264881916,
      "learning_rate": 4.446659143406581e-07,
      "loss": 1.4681,
      "step": 1999
    },
    {
      "epoch": 0.43956043956043955,
      "grad_norm": 0.2419493068887366,
      "learning_rate": 4.4445234256766873e-07,
      "loss": 1.4561,
      "step": 2000
    },
    {
      "epoch": 0.4397802197802198,
      "grad_norm": 0.24830821012270685,
      "learning_rate": 4.4423874245167355e-07,
      "loss": 1.4202,
      "step": 2001
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23866381009239548,
      "learning_rate": 4.440251140945033e-07,
      "loss": 1.4589,
      "step": 2002
    },
    {
      "epoch": 0.4402197802197802,
      "grad_norm": 0.2650481763620387,
      "learning_rate": 4.438114575980018e-07,
      "loss": 1.4912,
      "step": 2003
    },
    {
      "epoch": 0.44043956043956045,
      "grad_norm": 0.2672582520895399,
      "learning_rate": 4.4359777306402663e-07,
      "loss": 1.4875,
      "step": 2004
    },
    {
      "epoch": 0.44065934065934065,
      "grad_norm": 0.23896969422896724,
      "learning_rate": 4.433840605944488e-07,
      "loss": 1.5128,
      "step": 2005
    },
    {
      "epoch": 0.4408791208791209,
      "grad_norm": 0.2434239236442947,
      "learning_rate": 4.4317032029115234e-07,
      "loss": 1.4578,
      "step": 2006
    },
    {
      "epoch": 0.4410989010989011,
      "grad_norm": 0.24524504092689825,
      "learning_rate": 4.429565522560348e-07,
      "loss": 1.4882,
      "step": 2007
    },
    {
      "epoch": 0.4413186813186813,
      "grad_norm": 0.26319249674895484,
      "learning_rate": 4.427427565910069e-07,
      "loss": 1.5003,
      "step": 2008
    },
    {
      "epoch": 0.44153846153846155,
      "grad_norm": 0.24421254101152262,
      "learning_rate": 4.425289333979924e-07,
      "loss": 1.4428,
      "step": 2009
    },
    {
      "epoch": 0.44175824175824174,
      "grad_norm": 0.26257616504930686,
      "learning_rate": 4.4231508277892833e-07,
      "loss": 1.5429,
      "step": 2010
    },
    {
      "epoch": 0.441978021978022,
      "grad_norm": 0.2714002271767539,
      "learning_rate": 4.421012048357646e-07,
      "loss": 1.5147,
      "step": 2011
    },
    {
      "epoch": 0.4421978021978022,
      "grad_norm": 0.26141776740071254,
      "learning_rate": 4.418872996704645e-07,
      "loss": 1.4218,
      "step": 2012
    },
    {
      "epoch": 0.44241758241758244,
      "grad_norm": 0.2582404154466726,
      "learning_rate": 4.4167336738500395e-07,
      "loss": 1.517,
      "step": 2013
    },
    {
      "epoch": 0.44263736263736264,
      "grad_norm": 0.24120479739895342,
      "learning_rate": 4.414594080813721e-07,
      "loss": 1.4686,
      "step": 2014
    },
    {
      "epoch": 0.44285714285714284,
      "grad_norm": 0.2495583973180263,
      "learning_rate": 4.412454218615705e-07,
      "loss": 1.4777,
      "step": 2015
    },
    {
      "epoch": 0.4430769230769231,
      "grad_norm": 0.25250100386632984,
      "learning_rate": 4.4103140882761413e-07,
      "loss": 1.4964,
      "step": 2016
    },
    {
      "epoch": 0.4432967032967033,
      "grad_norm": 0.2602533544692144,
      "learning_rate": 4.4081736908153045e-07,
      "loss": 1.4982,
      "step": 2017
    },
    {
      "epoch": 0.44351648351648354,
      "grad_norm": 0.2448456408148941,
      "learning_rate": 4.406033027253595e-07,
      "loss": 1.4978,
      "step": 2018
    },
    {
      "epoch": 0.44373626373626374,
      "grad_norm": 0.26427459374956785,
      "learning_rate": 4.403892098611543e-07,
      "loss": 1.4515,
      "step": 2019
    },
    {
      "epoch": 0.44395604395604393,
      "grad_norm": 0.24692002348678865,
      "learning_rate": 4.4017509059098033e-07,
      "loss": 1.5037,
      "step": 2020
    },
    {
      "epoch": 0.4441758241758242,
      "grad_norm": 0.2429174226027057,
      "learning_rate": 4.399609450169158e-07,
      "loss": 1.4219,
      "step": 2021
    },
    {
      "epoch": 0.4443956043956044,
      "grad_norm": 0.2595607253689343,
      "learning_rate": 4.397467732410513e-07,
      "loss": 1.4642,
      "step": 2022
    },
    {
      "epoch": 0.44461538461538463,
      "grad_norm": 0.24262057566586,
      "learning_rate": 4.3953257536549e-07,
      "loss": 1.4686,
      "step": 2023
    },
    {
      "epoch": 0.44483516483516483,
      "grad_norm": 0.24935581520690986,
      "learning_rate": 4.393183514923476e-07,
      "loss": 1.4374,
      "step": 2024
    },
    {
      "epoch": 0.44505494505494503,
      "grad_norm": 0.25301345209781323,
      "learning_rate": 4.3910410172375197e-07,
      "loss": 1.4045,
      "step": 2025
    },
    {
      "epoch": 0.4452747252747253,
      "grad_norm": 0.3266191260053457,
      "learning_rate": 4.388898261618434e-07,
      "loss": 1.4664,
      "step": 2026
    },
    {
      "epoch": 0.4454945054945055,
      "grad_norm": 0.24422716200018552,
      "learning_rate": 4.3867552490877474e-07,
      "loss": 1.3922,
      "step": 2027
    },
    {
      "epoch": 0.44571428571428573,
      "grad_norm": 0.35083311768377134,
      "learning_rate": 4.3846119806671065e-07,
      "loss": 1.4702,
      "step": 2028
    },
    {
      "epoch": 0.4459340659340659,
      "grad_norm": 0.2561190306795904,
      "learning_rate": 4.382468457378284e-07,
      "loss": 1.4395,
      "step": 2029
    },
    {
      "epoch": 0.4461538461538462,
      "grad_norm": 0.30220352060657174,
      "learning_rate": 4.38032468024317e-07,
      "loss": 1.4538,
      "step": 2030
    },
    {
      "epoch": 0.4463736263736264,
      "grad_norm": 0.3318615787325118,
      "learning_rate": 4.378180650283781e-07,
      "loss": 1.4895,
      "step": 2031
    },
    {
      "epoch": 0.44659340659340657,
      "grad_norm": 0.23559489549850618,
      "learning_rate": 4.3760363685222495e-07,
      "loss": 1.4998,
      "step": 2032
    },
    {
      "epoch": 0.4468131868131868,
      "grad_norm": 0.2555988521994486,
      "learning_rate": 4.373891835980828e-07,
      "loss": 1.5352,
      "step": 2033
    },
    {
      "epoch": 0.447032967032967,
      "grad_norm": 0.24972905143409474,
      "learning_rate": 4.3717470536818903e-07,
      "loss": 1.5623,
      "step": 2034
    },
    {
      "epoch": 0.4472527472527473,
      "grad_norm": 0.2415536218839497,
      "learning_rate": 4.36960202264793e-07,
      "loss": 1.3939,
      "step": 2035
    },
    {
      "epoch": 0.44747252747252747,
      "grad_norm": 0.2620784864455322,
      "learning_rate": 4.367456743901559e-07,
      "loss": 1.4449,
      "step": 2036
    },
    {
      "epoch": 0.44769230769230767,
      "grad_norm": 0.2560134164059784,
      "learning_rate": 4.3653112184655035e-07,
      "loss": 1.3946,
      "step": 2037
    },
    {
      "epoch": 0.4479120879120879,
      "grad_norm": 0.2706183432557722,
      "learning_rate": 4.363165447362613e-07,
      "loss": 1.4553,
      "step": 2038
    },
    {
      "epoch": 0.4481318681318681,
      "grad_norm": 0.24148864334358544,
      "learning_rate": 4.3610194316158514e-07,
      "loss": 1.4521,
      "step": 2039
    },
    {
      "epoch": 0.44835164835164837,
      "grad_norm": 0.2477428186822883,
      "learning_rate": 4.358873172248298e-07,
      "loss": 1.4963,
      "step": 2040
    },
    {
      "epoch": 0.44857142857142857,
      "grad_norm": 0.2444094548120749,
      "learning_rate": 4.35672667028315e-07,
      "loss": 1.4787,
      "step": 2041
    },
    {
      "epoch": 0.4487912087912088,
      "grad_norm": 0.2819388152739819,
      "learning_rate": 4.3545799267437193e-07,
      "loss": 1.4535,
      "step": 2042
    },
    {
      "epoch": 0.449010989010989,
      "grad_norm": 0.25331600255846914,
      "learning_rate": 4.3524329426534356e-07,
      "loss": 1.4079,
      "step": 2043
    },
    {
      "epoch": 0.4492307692307692,
      "grad_norm": 0.2600742009308652,
      "learning_rate": 4.350285719035839e-07,
      "loss": 1.3983,
      "step": 2044
    },
    {
      "epoch": 0.44945054945054946,
      "grad_norm": 0.24806827114783264,
      "learning_rate": 4.3481382569145853e-07,
      "loss": 1.4367,
      "step": 2045
    },
    {
      "epoch": 0.44967032967032966,
      "grad_norm": 0.2527522514484014,
      "learning_rate": 4.3459905573134455e-07,
      "loss": 1.4561,
      "step": 2046
    },
    {
      "epoch": 0.4498901098901099,
      "grad_norm": 0.2416568168819981,
      "learning_rate": 4.343842621256304e-07,
      "loss": 1.4361,
      "step": 2047
    },
    {
      "epoch": 0.4501098901098901,
      "grad_norm": 0.25004779077616274,
      "learning_rate": 4.341694449767157e-07,
      "loss": 1.4807,
      "step": 2048
    },
    {
      "epoch": 0.4503296703296703,
      "grad_norm": 0.256423919404246,
      "learning_rate": 4.33954604387011e-07,
      "loss": 1.4729,
      "step": 2049
    },
    {
      "epoch": 0.45054945054945056,
      "grad_norm": 0.3029008617306128,
      "learning_rate": 4.337397404589385e-07,
      "loss": 1.478,
      "step": 2050
    },
    {
      "epoch": 0.45076923076923076,
      "grad_norm": 0.23925969015310744,
      "learning_rate": 4.335248532949315e-07,
      "loss": 1.428,
      "step": 2051
    },
    {
      "epoch": 0.450989010989011,
      "grad_norm": 0.27330281176762683,
      "learning_rate": 4.333099429974338e-07,
      "loss": 1.4692,
      "step": 2052
    },
    {
      "epoch": 0.4512087912087912,
      "grad_norm": 0.25284322084006117,
      "learning_rate": 4.3309500966890095e-07,
      "loss": 1.4964,
      "step": 2053
    },
    {
      "epoch": 0.4514285714285714,
      "grad_norm": 0.25212111305231066,
      "learning_rate": 4.3288005341179923e-07,
      "loss": 1.5439,
      "step": 2054
    },
    {
      "epoch": 0.45164835164835165,
      "grad_norm": 0.27897296674893207,
      "learning_rate": 4.3266507432860537e-07,
      "loss": 1.5189,
      "step": 2055
    },
    {
      "epoch": 0.45186813186813185,
      "grad_norm": 0.24574501087694758,
      "learning_rate": 4.3245007252180775e-07,
      "loss": 1.4427,
      "step": 2056
    },
    {
      "epoch": 0.4520879120879121,
      "grad_norm": 0.2567189836504206,
      "learning_rate": 4.322350480939052e-07,
      "loss": 1.4366,
      "step": 2057
    },
    {
      "epoch": 0.4523076923076923,
      "grad_norm": 0.2605807299139044,
      "learning_rate": 4.3202000114740727e-07,
      "loss": 1.4846,
      "step": 2058
    },
    {
      "epoch": 0.45252747252747255,
      "grad_norm": 0.2453753849791273,
      "learning_rate": 4.3180493178483435e-07,
      "loss": 1.4598,
      "step": 2059
    },
    {
      "epoch": 0.45274725274725275,
      "grad_norm": 0.24328461589670727,
      "learning_rate": 4.3158984010871765e-07,
      "loss": 1.4737,
      "step": 2060
    },
    {
      "epoch": 0.45296703296703295,
      "grad_norm": 0.25501626902401436,
      "learning_rate": 4.313747262215987e-07,
      "loss": 1.4738,
      "step": 2061
    },
    {
      "epoch": 0.4531868131868132,
      "grad_norm": 0.26784819737481885,
      "learning_rate": 4.3115959022602984e-07,
      "loss": 1.4225,
      "step": 2062
    },
    {
      "epoch": 0.4534065934065934,
      "grad_norm": 0.24525130988311333,
      "learning_rate": 4.30944432224574e-07,
      "loss": 1.5193,
      "step": 2063
    },
    {
      "epoch": 0.45362637362637365,
      "grad_norm": 0.3612247288721671,
      "learning_rate": 4.307292523198043e-07,
      "loss": 1.496,
      "step": 2064
    },
    {
      "epoch": 0.45384615384615384,
      "grad_norm": 0.23592061965966898,
      "learning_rate": 4.3051405061430465e-07,
      "loss": 1.4753,
      "step": 2065
    },
    {
      "epoch": 0.45406593406593404,
      "grad_norm": 0.25423611392034606,
      "learning_rate": 4.3029882721066934e-07,
      "loss": 1.5484,
      "step": 2066
    },
    {
      "epoch": 0.4542857142857143,
      "grad_norm": 0.2935670369186098,
      "learning_rate": 4.3008358221150267e-07,
      "loss": 1.4633,
      "step": 2067
    },
    {
      "epoch": 0.4545054945054945,
      "grad_norm": 0.24337580307309942,
      "learning_rate": 4.2986831571941944e-07,
      "loss": 1.4539,
      "step": 2068
    },
    {
      "epoch": 0.45472527472527474,
      "grad_norm": 0.25121797435260773,
      "learning_rate": 4.2965302783704474e-07,
      "loss": 1.4811,
      "step": 2069
    },
    {
      "epoch": 0.45494505494505494,
      "grad_norm": 0.24855948195337907,
      "learning_rate": 4.29437718667014e-07,
      "loss": 1.4963,
      "step": 2070
    },
    {
      "epoch": 0.4551648351648352,
      "grad_norm": 0.25810073656964916,
      "learning_rate": 4.2922238831197235e-07,
      "loss": 1.5138,
      "step": 2071
    },
    {
      "epoch": 0.4553846153846154,
      "grad_norm": 0.24979152300750987,
      "learning_rate": 4.290070368745754e-07,
      "loss": 1.5022,
      "step": 2072
    },
    {
      "epoch": 0.4556043956043956,
      "grad_norm": 0.28878714910143466,
      "learning_rate": 4.2879166445748865e-07,
      "loss": 1.4649,
      "step": 2073
    },
    {
      "epoch": 0.45582417582417584,
      "grad_norm": 0.24946059870744922,
      "learning_rate": 4.285762711633877e-07,
      "loss": 1.493,
      "step": 2074
    },
    {
      "epoch": 0.45604395604395603,
      "grad_norm": 0.2757308433732612,
      "learning_rate": 4.28360857094958e-07,
      "loss": 1.4789,
      "step": 2075
    },
    {
      "epoch": 0.4562637362637363,
      "grad_norm": 0.2596881745657374,
      "learning_rate": 4.2814542235489494e-07,
      "loss": 1.4998,
      "step": 2076
    },
    {
      "epoch": 0.4564835164835165,
      "grad_norm": 0.2603748068429344,
      "learning_rate": 4.2792996704590387e-07,
      "loss": 1.4733,
      "step": 2077
    },
    {
      "epoch": 0.4567032967032967,
      "grad_norm": 0.2443662480772662,
      "learning_rate": 4.277144912706998e-07,
      "loss": 1.4996,
      "step": 2078
    },
    {
      "epoch": 0.45692307692307693,
      "grad_norm": 0.2614697763154464,
      "learning_rate": 4.274989951320075e-07,
      "loss": 1.4367,
      "step": 2079
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 0.24731871090892066,
      "learning_rate": 4.272834787325615e-07,
      "loss": 1.4672,
      "step": 2080
    },
    {
      "epoch": 0.4573626373626374,
      "grad_norm": 0.2560989472472663,
      "learning_rate": 4.2706794217510603e-07,
      "loss": 1.4906,
      "step": 2081
    },
    {
      "epoch": 0.4575824175824176,
      "grad_norm": 0.24534994109634842,
      "learning_rate": 4.2685238556239495e-07,
      "loss": 1.4648,
      "step": 2082
    },
    {
      "epoch": 0.4578021978021978,
      "grad_norm": 0.28760353790968335,
      "learning_rate": 4.2663680899719143e-07,
      "loss": 1.4789,
      "step": 2083
    },
    {
      "epoch": 0.458021978021978,
      "grad_norm": 0.2396213350672438,
      "learning_rate": 4.264212125822686e-07,
      "loss": 1.4612,
      "step": 2084
    },
    {
      "epoch": 0.4582417582417582,
      "grad_norm": 0.26144684638887217,
      "learning_rate": 4.262055964204085e-07,
      "loss": 1.4013,
      "step": 2085
    },
    {
      "epoch": 0.4584615384615385,
      "grad_norm": 0.2607553202482003,
      "learning_rate": 4.259899606144031e-07,
      "loss": 1.5135,
      "step": 2086
    },
    {
      "epoch": 0.4586813186813187,
      "grad_norm": 0.2594931474385825,
      "learning_rate": 4.2577430526705343e-07,
      "loss": 1.416,
      "step": 2087
    },
    {
      "epoch": 0.4589010989010989,
      "grad_norm": 0.2611756491498896,
      "learning_rate": 4.255586304811699e-07,
      "loss": 1.5464,
      "step": 2088
    },
    {
      "epoch": 0.4591208791208791,
      "grad_norm": 0.25004996383787975,
      "learning_rate": 4.253429363595723e-07,
      "loss": 1.4762,
      "step": 2089
    },
    {
      "epoch": 0.4593406593406593,
      "grad_norm": 0.25427160923801684,
      "learning_rate": 4.2512722300508943e-07,
      "loss": 1.5437,
      "step": 2090
    },
    {
      "epoch": 0.45956043956043957,
      "grad_norm": 0.2610733982322852,
      "learning_rate": 4.2491149052055936e-07,
      "loss": 1.5014,
      "step": 2091
    },
    {
      "epoch": 0.45978021978021977,
      "grad_norm": 0.2561664692941615,
      "learning_rate": 4.2469573900882946e-07,
      "loss": 1.4396,
      "step": 2092
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26370019260727595,
      "learning_rate": 4.2447996857275583e-07,
      "loss": 1.5059,
      "step": 2093
    },
    {
      "epoch": 0.4602197802197802,
      "grad_norm": 0.3261297040343998,
      "learning_rate": 4.242641793152038e-07,
      "loss": 1.5002,
      "step": 2094
    },
    {
      "epoch": 0.4604395604395604,
      "grad_norm": 0.28908805699260925,
      "learning_rate": 4.240483713390477e-07,
      "loss": 1.5241,
      "step": 2095
    },
    {
      "epoch": 0.46065934065934067,
      "grad_norm": 0.26358486161361233,
      "learning_rate": 4.2383254474717064e-07,
      "loss": 1.5232,
      "step": 2096
    },
    {
      "epoch": 0.46087912087912086,
      "grad_norm": 0.24265924214246137,
      "learning_rate": 4.2361669964246477e-07,
      "loss": 1.4652,
      "step": 2097
    },
    {
      "epoch": 0.4610989010989011,
      "grad_norm": 0.23999489946480565,
      "learning_rate": 4.2340083612783094e-07,
      "loss": 1.4139,
      "step": 2098
    },
    {
      "epoch": 0.4613186813186813,
      "grad_norm": 0.2511447918001245,
      "learning_rate": 4.231849543061788e-07,
      "loss": 1.4413,
      "step": 2099
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 0.2434633609202519,
      "learning_rate": 4.229690542804267e-07,
      "loss": 1.448,
      "step": 2100
    },
    {
      "epoch": 0.46175824175824176,
      "grad_norm": 0.2680332101972384,
      "learning_rate": 4.2275313615350195e-07,
      "loss": 1.4894,
      "step": 2101
    },
    {
      "epoch": 0.46197802197802196,
      "grad_norm": 0.2553793825205718,
      "learning_rate": 4.225372000283399e-07,
      "loss": 1.4996,
      "step": 2102
    },
    {
      "epoch": 0.4621978021978022,
      "grad_norm": 0.25530929239675787,
      "learning_rate": 4.223212460078851e-07,
      "loss": 1.4353,
      "step": 2103
    },
    {
      "epoch": 0.4624175824175824,
      "grad_norm": 0.2467225152119547,
      "learning_rate": 4.2210527419509034e-07,
      "loss": 1.4512,
      "step": 2104
    },
    {
      "epoch": 0.46263736263736266,
      "grad_norm": 0.2542087333372453,
      "learning_rate": 4.218892846929167e-07,
      "loss": 1.4923,
      "step": 2105
    },
    {
      "epoch": 0.46285714285714286,
      "grad_norm": 0.24250588824628316,
      "learning_rate": 4.2167327760433415e-07,
      "loss": 1.4893,
      "step": 2106
    },
    {
      "epoch": 0.46307692307692305,
      "grad_norm": 0.248356494344754,
      "learning_rate": 4.214572530323209e-07,
      "loss": 1.4826,
      "step": 2107
    },
    {
      "epoch": 0.4632967032967033,
      "grad_norm": 0.25129696629053866,
      "learning_rate": 4.21241211079863e-07,
      "loss": 1.5164,
      "step": 2108
    },
    {
      "epoch": 0.4635164835164835,
      "grad_norm": 0.24500177287742092,
      "learning_rate": 4.210251518499555e-07,
      "loss": 1.412,
      "step": 2109
    },
    {
      "epoch": 0.46373626373626375,
      "grad_norm": 0.255636589279567,
      "learning_rate": 4.208090754456011e-07,
      "loss": 1.5499,
      "step": 2110
    },
    {
      "epoch": 0.46395604395604395,
      "grad_norm": 0.3803441252872352,
      "learning_rate": 4.205929819698114e-07,
      "loss": 1.4686,
      "step": 2111
    },
    {
      "epoch": 0.4641758241758242,
      "grad_norm": 0.24126324776431443,
      "learning_rate": 4.203768715256053e-07,
      "loss": 1.3918,
      "step": 2112
    },
    {
      "epoch": 0.4643956043956044,
      "grad_norm": 0.2496857026325197,
      "learning_rate": 4.2016074421601047e-07,
      "loss": 1.5016,
      "step": 2113
    },
    {
      "epoch": 0.4646153846153846,
      "grad_norm": 0.25998269132170193,
      "learning_rate": 4.199446001440621e-07,
      "loss": 1.4559,
      "step": 2114
    },
    {
      "epoch": 0.46483516483516485,
      "grad_norm": 0.25144116621936335,
      "learning_rate": 4.1972843941280375e-07,
      "loss": 1.4639,
      "step": 2115
    },
    {
      "epoch": 0.46505494505494505,
      "grad_norm": 0.24650473826043623,
      "learning_rate": 4.1951226212528674e-07,
      "loss": 1.5229,
      "step": 2116
    },
    {
      "epoch": 0.4652747252747253,
      "grad_norm": 0.2540484387629659,
      "learning_rate": 4.192960683845703e-07,
      "loss": 1.4339,
      "step": 2117
    },
    {
      "epoch": 0.4654945054945055,
      "grad_norm": 0.2752498752239846,
      "learning_rate": 4.1907985829372146e-07,
      "loss": 1.4687,
      "step": 2118
    },
    {
      "epoch": 0.4657142857142857,
      "grad_norm": 0.24540759562663214,
      "learning_rate": 4.188636319558154e-07,
      "loss": 1.4865,
      "step": 2119
    },
    {
      "epoch": 0.46593406593406594,
      "grad_norm": 0.2658690120881403,
      "learning_rate": 4.1864738947393443e-07,
      "loss": 1.4461,
      "step": 2120
    },
    {
      "epoch": 0.46615384615384614,
      "grad_norm": 0.38623091924961017,
      "learning_rate": 4.184311309511691e-07,
      "loss": 1.4963,
      "step": 2121
    },
    {
      "epoch": 0.4663736263736264,
      "grad_norm": 0.24199241715554962,
      "learning_rate": 4.182148564906173e-07,
      "loss": 1.4266,
      "step": 2122
    },
    {
      "epoch": 0.4665934065934066,
      "grad_norm": 0.24330896768729043,
      "learning_rate": 4.1799856619538456e-07,
      "loss": 1.3831,
      "step": 2123
    },
    {
      "epoch": 0.4668131868131868,
      "grad_norm": 0.24032074539738865,
      "learning_rate": 4.177822601685843e-07,
      "loss": 1.4185,
      "step": 2124
    },
    {
      "epoch": 0.46703296703296704,
      "grad_norm": 0.24960556255798652,
      "learning_rate": 4.1756593851333657e-07,
      "loss": 1.4808,
      "step": 2125
    },
    {
      "epoch": 0.46725274725274724,
      "grad_norm": 0.23711489610297176,
      "learning_rate": 4.1734960133277e-07,
      "loss": 1.4657,
      "step": 2126
    },
    {
      "epoch": 0.4674725274725275,
      "grad_norm": 0.32068962980375065,
      "learning_rate": 4.1713324873001985e-07,
      "loss": 1.4832,
      "step": 2127
    },
    {
      "epoch": 0.4676923076923077,
      "grad_norm": 0.2518391727634862,
      "learning_rate": 4.1691688080822906e-07,
      "loss": 1.4646,
      "step": 2128
    },
    {
      "epoch": 0.46791208791208794,
      "grad_norm": 0.24932634371386272,
      "learning_rate": 4.1670049767054757e-07,
      "loss": 1.4596,
      "step": 2129
    },
    {
      "epoch": 0.46813186813186813,
      "grad_norm": 0.251421232967717,
      "learning_rate": 4.16484099420133e-07,
      "loss": 1.509,
      "step": 2130
    },
    {
      "epoch": 0.46835164835164833,
      "grad_norm": 0.25696894067799475,
      "learning_rate": 4.162676861601498e-07,
      "loss": 1.4876,
      "step": 2131
    },
    {
      "epoch": 0.4685714285714286,
      "grad_norm": 0.23856366540520546,
      "learning_rate": 4.160512579937697e-07,
      "loss": 1.5087,
      "step": 2132
    },
    {
      "epoch": 0.4687912087912088,
      "grad_norm": 0.24999786642137214,
      "learning_rate": 4.1583481502417166e-07,
      "loss": 1.5268,
      "step": 2133
    },
    {
      "epoch": 0.46901098901098903,
      "grad_norm": 0.2463367846663434,
      "learning_rate": 4.156183573545415e-07,
      "loss": 1.4424,
      "step": 2134
    },
    {
      "epoch": 0.46923076923076923,
      "grad_norm": 0.28294377177431385,
      "learning_rate": 4.1540188508807225e-07,
      "loss": 1.4708,
      "step": 2135
    },
    {
      "epoch": 0.4694505494505494,
      "grad_norm": 0.25320109960323733,
      "learning_rate": 4.151853983279637e-07,
      "loss": 1.5059,
      "step": 2136
    },
    {
      "epoch": 0.4696703296703297,
      "grad_norm": 0.25529501987391623,
      "learning_rate": 4.1496889717742263e-07,
      "loss": 1.4886,
      "step": 2137
    },
    {
      "epoch": 0.4698901098901099,
      "grad_norm": 0.26722265943532014,
      "learning_rate": 4.147523817396629e-07,
      "loss": 1.5239,
      "step": 2138
    },
    {
      "epoch": 0.4701098901098901,
      "grad_norm": 0.2638999649207356,
      "learning_rate": 4.1453585211790465e-07,
      "loss": 1.5156,
      "step": 2139
    },
    {
      "epoch": 0.4703296703296703,
      "grad_norm": 0.25046035356861307,
      "learning_rate": 4.143193084153753e-07,
      "loss": 1.484,
      "step": 2140
    },
    {
      "epoch": 0.4705494505494506,
      "grad_norm": 0.2503101049980358,
      "learning_rate": 4.141027507353087e-07,
      "loss": 1.4616,
      "step": 2141
    },
    {
      "epoch": 0.4707692307692308,
      "grad_norm": 0.2605648383058907,
      "learning_rate": 4.1388617918094563e-07,
      "loss": 1.4535,
      "step": 2142
    },
    {
      "epoch": 0.47098901098901097,
      "grad_norm": 0.24344463446649003,
      "learning_rate": 4.1366959385553316e-07,
      "loss": 1.4759,
      "step": 2143
    },
    {
      "epoch": 0.4712087912087912,
      "grad_norm": 0.2721819098807181,
      "learning_rate": 4.134529948623249e-07,
      "loss": 1.5411,
      "step": 2144
    },
    {
      "epoch": 0.4714285714285714,
      "grad_norm": 0.2454384678281813,
      "learning_rate": 4.132363823045816e-07,
      "loss": 1.3788,
      "step": 2145
    },
    {
      "epoch": 0.47164835164835167,
      "grad_norm": 0.2521026799104069,
      "learning_rate": 4.130197562855696e-07,
      "loss": 1.4809,
      "step": 2146
    },
    {
      "epoch": 0.47186813186813187,
      "grad_norm": 0.24389100413775666,
      "learning_rate": 4.1280311690856234e-07,
      "loss": 1.4973,
      "step": 2147
    },
    {
      "epoch": 0.47208791208791206,
      "grad_norm": 0.24801335859937837,
      "learning_rate": 4.125864642768391e-07,
      "loss": 1.4622,
      "step": 2148
    },
    {
      "epoch": 0.4723076923076923,
      "grad_norm": 0.24582502995608502,
      "learning_rate": 4.12369798493686e-07,
      "loss": 1.441,
      "step": 2149
    },
    {
      "epoch": 0.4725274725274725,
      "grad_norm": 0.2556859453671942,
      "learning_rate": 4.121531196623951e-07,
      "loss": 1.4526,
      "step": 2150
    },
    {
      "epoch": 0.47274725274725277,
      "grad_norm": 0.6082569291110256,
      "learning_rate": 4.1193642788626474e-07,
      "loss": 1.4471,
      "step": 2151
    },
    {
      "epoch": 0.47296703296703296,
      "grad_norm": 0.24879058608612206,
      "learning_rate": 4.117197232685995e-07,
      "loss": 1.4823,
      "step": 2152
    },
    {
      "epoch": 0.47318681318681316,
      "grad_norm": 0.267231589560563,
      "learning_rate": 4.1150300591271e-07,
      "loss": 1.5142,
      "step": 2153
    },
    {
      "epoch": 0.4734065934065934,
      "grad_norm": 0.27099962653455645,
      "learning_rate": 4.1128627592191295e-07,
      "loss": 1.4974,
      "step": 2154
    },
    {
      "epoch": 0.4736263736263736,
      "grad_norm": 0.27173473889903393,
      "learning_rate": 4.1106953339953103e-07,
      "loss": 1.4799,
      "step": 2155
    },
    {
      "epoch": 0.47384615384615386,
      "grad_norm": 0.24540001875933956,
      "learning_rate": 4.10852778448893e-07,
      "loss": 1.453,
      "step": 2156
    },
    {
      "epoch": 0.47406593406593406,
      "grad_norm": 0.24341896446192007,
      "learning_rate": 4.106360111733337e-07,
      "loss": 1.422,
      "step": 2157
    },
    {
      "epoch": 0.4742857142857143,
      "grad_norm": 0.2605932533495345,
      "learning_rate": 4.1041923167619345e-07,
      "loss": 1.5059,
      "step": 2158
    },
    {
      "epoch": 0.4745054945054945,
      "grad_norm": 0.3739400105072832,
      "learning_rate": 4.102024400608186e-07,
      "loss": 1.434,
      "step": 2159
    },
    {
      "epoch": 0.4747252747252747,
      "grad_norm": 0.24029918941121878,
      "learning_rate": 4.099856364305614e-07,
      "loss": 1.4364,
      "step": 2160
    },
    {
      "epoch": 0.47494505494505496,
      "grad_norm": 0.25843652143322837,
      "learning_rate": 4.0976882088877955e-07,
      "loss": 1.4552,
      "step": 2161
    },
    {
      "epoch": 0.47516483516483515,
      "grad_norm": 0.36890007621991705,
      "learning_rate": 4.0955199353883666e-07,
      "loss": 1.4771,
      "step": 2162
    },
    {
      "epoch": 0.4753846153846154,
      "grad_norm": 0.2559317403094861,
      "learning_rate": 4.0933515448410197e-07,
      "loss": 1.471,
      "step": 2163
    },
    {
      "epoch": 0.4756043956043956,
      "grad_norm": 0.25008407923538156,
      "learning_rate": 4.0911830382795e-07,
      "loss": 1.5048,
      "step": 2164
    },
    {
      "epoch": 0.4758241758241758,
      "grad_norm": 0.2423485199780464,
      "learning_rate": 4.089014416737613e-07,
      "loss": 1.5425,
      "step": 2165
    },
    {
      "epoch": 0.47604395604395605,
      "grad_norm": 0.24572023579073374,
      "learning_rate": 4.086845681249215e-07,
      "loss": 1.4656,
      "step": 2166
    },
    {
      "epoch": 0.47626373626373625,
      "grad_norm": 0.27440521210145385,
      "learning_rate": 4.084676832848216e-07,
      "loss": 1.491,
      "step": 2167
    },
    {
      "epoch": 0.4764835164835165,
      "grad_norm": 0.3107747060908553,
      "learning_rate": 4.082507872568585e-07,
      "loss": 1.5429,
      "step": 2168
    },
    {
      "epoch": 0.4767032967032967,
      "grad_norm": 0.2537307127475986,
      "learning_rate": 4.08033880144434e-07,
      "loss": 1.4688,
      "step": 2169
    },
    {
      "epoch": 0.47692307692307695,
      "grad_norm": 0.2518383654657694,
      "learning_rate": 4.078169620509552e-07,
      "loss": 1.4755,
      "step": 2170
    },
    {
      "epoch": 0.47714285714285715,
      "grad_norm": 0.31769063812828074,
      "learning_rate": 4.076000330798345e-07,
      "loss": 1.4752,
      "step": 2171
    },
    {
      "epoch": 0.47736263736263734,
      "grad_norm": 0.26660323364676236,
      "learning_rate": 4.0738309333448964e-07,
      "loss": 1.4554,
      "step": 2172
    },
    {
      "epoch": 0.4775824175824176,
      "grad_norm": 0.25529946819906274,
      "learning_rate": 4.0716614291834327e-07,
      "loss": 1.5257,
      "step": 2173
    },
    {
      "epoch": 0.4778021978021978,
      "grad_norm": 0.25148741629104937,
      "learning_rate": 4.0694918193482317e-07,
      "loss": 1.4256,
      "step": 2174
    },
    {
      "epoch": 0.47802197802197804,
      "grad_norm": 0.23587639375669317,
      "learning_rate": 4.0673221048736234e-07,
      "loss": 1.4262,
      "step": 2175
    },
    {
      "epoch": 0.47824175824175824,
      "grad_norm": 0.25877167767567394,
      "learning_rate": 4.0651522867939863e-07,
      "loss": 1.4829,
      "step": 2176
    },
    {
      "epoch": 0.47846153846153844,
      "grad_norm": 0.24940377754896279,
      "learning_rate": 4.0629823661437485e-07,
      "loss": 1.4375,
      "step": 2177
    },
    {
      "epoch": 0.4786813186813187,
      "grad_norm": 0.25485570465304636,
      "learning_rate": 4.060812343957385e-07,
      "loss": 1.4751,
      "step": 2178
    },
    {
      "epoch": 0.4789010989010989,
      "grad_norm": 0.24195630123328987,
      "learning_rate": 4.0586422212694226e-07,
      "loss": 1.4817,
      "step": 2179
    },
    {
      "epoch": 0.47912087912087914,
      "grad_norm": 0.2646037576766741,
      "learning_rate": 4.0564719991144344e-07,
      "loss": 1.4383,
      "step": 2180
    },
    {
      "epoch": 0.47934065934065934,
      "grad_norm": 0.24472048765588794,
      "learning_rate": 4.0543016785270417e-07,
      "loss": 1.4679,
      "step": 2181
    },
    {
      "epoch": 0.47956043956043953,
      "grad_norm": 0.28942181978522935,
      "learning_rate": 4.052131260541911e-07,
      "loss": 1.4702,
      "step": 2182
    },
    {
      "epoch": 0.4797802197802198,
      "grad_norm": 0.24832404652400986,
      "learning_rate": 4.049960746193755e-07,
      "loss": 1.4315,
      "step": 2183
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4369544456809934,
      "learning_rate": 4.0477901365173375e-07,
      "loss": 1.4129,
      "step": 2184
    },
    {
      "epoch": 0.48021978021978023,
      "grad_norm": 0.2541583125518289,
      "learning_rate": 4.045619432547461e-07,
      "loss": 1.4702,
      "step": 2185
    },
    {
      "epoch": 0.48043956043956043,
      "grad_norm": 0.2523588685878539,
      "learning_rate": 4.043448635318975e-07,
      "loss": 1.4745,
      "step": 2186
    },
    {
      "epoch": 0.4806593406593407,
      "grad_norm": 0.2510926996602045,
      "learning_rate": 4.041277745866776e-07,
      "loss": 1.4902,
      "step": 2187
    },
    {
      "epoch": 0.4808791208791209,
      "grad_norm": 0.2592318111699245,
      "learning_rate": 4.039106765225803e-07,
      "loss": 1.49,
      "step": 2188
    },
    {
      "epoch": 0.4810989010989011,
      "grad_norm": 0.2511407064232407,
      "learning_rate": 4.036935694431038e-07,
      "loss": 1.4597,
      "step": 2189
    },
    {
      "epoch": 0.48131868131868133,
      "grad_norm": 0.24553195643338055,
      "learning_rate": 4.034764534517504e-07,
      "loss": 1.4164,
      "step": 2190
    },
    {
      "epoch": 0.4815384615384615,
      "grad_norm": 0.2544932429204096,
      "learning_rate": 4.032593286520272e-07,
      "loss": 1.4233,
      "step": 2191
    },
    {
      "epoch": 0.4817582417582418,
      "grad_norm": 0.246372819528969,
      "learning_rate": 4.030421951474449e-07,
      "loss": 1.4254,
      "step": 2192
    },
    {
      "epoch": 0.481978021978022,
      "grad_norm": 0.2444111471649207,
      "learning_rate": 4.028250530415188e-07,
      "loss": 1.502,
      "step": 2193
    },
    {
      "epoch": 0.48219780219780217,
      "grad_norm": 0.2484790732312108,
      "learning_rate": 4.0260790243776806e-07,
      "loss": 1.4628,
      "step": 2194
    },
    {
      "epoch": 0.4824175824175824,
      "grad_norm": 0.24749135295669722,
      "learning_rate": 4.0239074343971586e-07,
      "loss": 1.4266,
      "step": 2195
    },
    {
      "epoch": 0.4826373626373626,
      "grad_norm": 0.24861043417869974,
      "learning_rate": 4.021735761508896e-07,
      "loss": 1.4871,
      "step": 2196
    },
    {
      "epoch": 0.4828571428571429,
      "grad_norm": 0.2531956580470661,
      "learning_rate": 4.0195640067482035e-07,
      "loss": 1.4946,
      "step": 2197
    },
    {
      "epoch": 0.48307692307692307,
      "grad_norm": 0.25469456597604356,
      "learning_rate": 4.017392171150434e-07,
      "loss": 1.5239,
      "step": 2198
    },
    {
      "epoch": 0.4832967032967033,
      "grad_norm": 0.25846203750265634,
      "learning_rate": 4.015220255750977e-07,
      "loss": 1.4277,
      "step": 2199
    },
    {
      "epoch": 0.4835164835164835,
      "grad_norm": 0.24535480350801261,
      "learning_rate": 4.0130482615852604e-07,
      "loss": 1.4681,
      "step": 2200
    },
    {
      "epoch": 0.4837362637362637,
      "grad_norm": 0.2474661039500049,
      "learning_rate": 4.010876189688748e-07,
      "loss": 1.4978,
      "step": 2201
    },
    {
      "epoch": 0.48395604395604397,
      "grad_norm": 0.24811461610371785,
      "learning_rate": 4.008704041096943e-07,
      "loss": 1.4826,
      "step": 2202
    },
    {
      "epoch": 0.48417582417582417,
      "grad_norm": 0.23865921226506684,
      "learning_rate": 4.006531816845386e-07,
      "loss": 1.4128,
      "step": 2203
    },
    {
      "epoch": 0.4843956043956044,
      "grad_norm": 0.2536119805482763,
      "learning_rate": 4.004359517969651e-07,
      "loss": 1.5137,
      "step": 2204
    },
    {
      "epoch": 0.4846153846153846,
      "grad_norm": 0.2538459095239332,
      "learning_rate": 4.002187145505347e-07,
      "loss": 1.4692,
      "step": 2205
    },
    {
      "epoch": 0.4848351648351648,
      "grad_norm": 0.25037515727381765,
      "learning_rate": 4.000014700488123e-07,
      "loss": 1.486,
      "step": 2206
    },
    {
      "epoch": 0.48505494505494506,
      "grad_norm": 0.2534854405761311,
      "learning_rate": 3.9978421839536565e-07,
      "loss": 1.4716,
      "step": 2207
    },
    {
      "epoch": 0.48527472527472526,
      "grad_norm": 0.2516800776138757,
      "learning_rate": 3.995669596937664e-07,
      "loss": 1.4109,
      "step": 2208
    },
    {
      "epoch": 0.4854945054945055,
      "grad_norm": 0.26224105682457977,
      "learning_rate": 3.9934969404758916e-07,
      "loss": 1.502,
      "step": 2209
    },
    {
      "epoch": 0.4857142857142857,
      "grad_norm": 0.33275220074258516,
      "learning_rate": 3.9913242156041224e-07,
      "loss": 1.445,
      "step": 2210
    },
    {
      "epoch": 0.4859340659340659,
      "grad_norm": 0.2517760073190495,
      "learning_rate": 3.9891514233581694e-07,
      "loss": 1.5186,
      "step": 2211
    },
    {
      "epoch": 0.48615384615384616,
      "grad_norm": 0.24466627661225412,
      "learning_rate": 3.986978564773877e-07,
      "loss": 1.4935,
      "step": 2212
    },
    {
      "epoch": 0.48637362637362636,
      "grad_norm": 0.25131614810160363,
      "learning_rate": 3.984805640887124e-07,
      "loss": 1.4478,
      "step": 2213
    },
    {
      "epoch": 0.4865934065934066,
      "grad_norm": 0.2586676888317208,
      "learning_rate": 3.982632652733819e-07,
      "loss": 1.4601,
      "step": 2214
    },
    {
      "epoch": 0.4868131868131868,
      "grad_norm": 0.2484649524850849,
      "learning_rate": 3.980459601349902e-07,
      "loss": 1.4408,
      "step": 2215
    },
    {
      "epoch": 0.48703296703296706,
      "grad_norm": 0.25411892758832755,
      "learning_rate": 3.978286487771342e-07,
      "loss": 1.4455,
      "step": 2216
    },
    {
      "epoch": 0.48725274725274725,
      "grad_norm": 0.2527913746887048,
      "learning_rate": 3.9761133130341364e-07,
      "loss": 1.4259,
      "step": 2217
    },
    {
      "epoch": 0.48747252747252745,
      "grad_norm": 0.2524472976197228,
      "learning_rate": 3.9739400781743154e-07,
      "loss": 1.4366,
      "step": 2218
    },
    {
      "epoch": 0.4876923076923077,
      "grad_norm": 0.24749176289072802,
      "learning_rate": 3.9717667842279346e-07,
      "loss": 1.4817,
      "step": 2219
    },
    {
      "epoch": 0.4879120879120879,
      "grad_norm": 0.23411649445207888,
      "learning_rate": 3.96959343223108e-07,
      "loss": 1.4662,
      "step": 2220
    },
    {
      "epoch": 0.48813186813186815,
      "grad_norm": 0.259536911380817,
      "learning_rate": 3.967420023219864e-07,
      "loss": 1.48,
      "step": 2221
    },
    {
      "epoch": 0.48835164835164835,
      "grad_norm": 0.2492732422110916,
      "learning_rate": 3.965246558230426e-07,
      "loss": 1.4856,
      "step": 2222
    },
    {
      "epoch": 0.48857142857142855,
      "grad_norm": 0.2572209250914964,
      "learning_rate": 3.963073038298935e-07,
      "loss": 1.4741,
      "step": 2223
    },
    {
      "epoch": 0.4887912087912088,
      "grad_norm": 0.25895725275313164,
      "learning_rate": 3.9608994644615794e-07,
      "loss": 1.4347,
      "step": 2224
    },
    {
      "epoch": 0.489010989010989,
      "grad_norm": 0.245615821628319,
      "learning_rate": 3.9587258377545804e-07,
      "loss": 1.4458,
      "step": 2225
    },
    {
      "epoch": 0.48923076923076925,
      "grad_norm": 0.2563831201035074,
      "learning_rate": 3.9565521592141815e-07,
      "loss": 1.5211,
      "step": 2226
    },
    {
      "epoch": 0.48945054945054944,
      "grad_norm": 0.2541876466901656,
      "learning_rate": 3.9543784298766525e-07,
      "loss": 1.4274,
      "step": 2227
    },
    {
      "epoch": 0.4896703296703297,
      "grad_norm": 0.25398626933791496,
      "learning_rate": 3.952204650778282e-07,
      "loss": 1.4797,
      "step": 2228
    },
    {
      "epoch": 0.4898901098901099,
      "grad_norm": 0.25758327785635093,
      "learning_rate": 3.950030822955391e-07,
      "loss": 1.4612,
      "step": 2229
    },
    {
      "epoch": 0.4901098901098901,
      "grad_norm": 0.2541624023396581,
      "learning_rate": 3.9478569474443164e-07,
      "loss": 1.4715,
      "step": 2230
    },
    {
      "epoch": 0.49032967032967034,
      "grad_norm": 0.2442921249089427,
      "learning_rate": 3.9456830252814204e-07,
      "loss": 1.5103,
      "step": 2231
    },
    {
      "epoch": 0.49054945054945054,
      "grad_norm": 0.255858430011158,
      "learning_rate": 3.943509057503088e-07,
      "loss": 1.4637,
      "step": 2232
    },
    {
      "epoch": 0.4907692307692308,
      "grad_norm": 0.2507519521381344,
      "learning_rate": 3.9413350451457257e-07,
      "loss": 1.4694,
      "step": 2233
    },
    {
      "epoch": 0.490989010989011,
      "grad_norm": 0.2573155043460483,
      "learning_rate": 3.9391609892457617e-07,
      "loss": 1.4416,
      "step": 2234
    },
    {
      "epoch": 0.4912087912087912,
      "grad_norm": 0.28593458043217906,
      "learning_rate": 3.9369868908396425e-07,
      "loss": 1.4554,
      "step": 2235
    },
    {
      "epoch": 0.49142857142857144,
      "grad_norm": 0.24396185472777643,
      "learning_rate": 3.934812750963838e-07,
      "loss": 1.4465,
      "step": 2236
    },
    {
      "epoch": 0.49164835164835163,
      "grad_norm": 0.24446906800643653,
      "learning_rate": 3.932638570654836e-07,
      "loss": 1.4438,
      "step": 2237
    },
    {
      "epoch": 0.4918681318681319,
      "grad_norm": 0.23084238338734298,
      "learning_rate": 3.930464350949144e-07,
      "loss": 1.4375,
      "step": 2238
    },
    {
      "epoch": 0.4920879120879121,
      "grad_norm": 0.2528309575959943,
      "learning_rate": 3.9282900928832897e-07,
      "loss": 1.5037,
      "step": 2239
    },
    {
      "epoch": 0.49230769230769234,
      "grad_norm": 0.24839907705789768,
      "learning_rate": 3.9261157974938167e-07,
      "loss": 1.4749,
      "step": 2240
    },
    {
      "epoch": 0.49252747252747253,
      "grad_norm": 0.2492622242980725,
      "learning_rate": 3.9239414658172877e-07,
      "loss": 1.4654,
      "step": 2241
    },
    {
      "epoch": 0.49274725274725273,
      "grad_norm": 0.25282608627910996,
      "learning_rate": 3.921767098890282e-07,
      "loss": 1.4292,
      "step": 2242
    },
    {
      "epoch": 0.492967032967033,
      "grad_norm": 0.24870397375672293,
      "learning_rate": 3.919592697749395e-07,
      "loss": 1.4669,
      "step": 2243
    },
    {
      "epoch": 0.4931868131868132,
      "grad_norm": 0.2512220677317805,
      "learning_rate": 3.9174182634312436e-07,
      "loss": 1.4564,
      "step": 2244
    },
    {
      "epoch": 0.49340659340659343,
      "grad_norm": 0.24628718210590606,
      "learning_rate": 3.915243796972452e-07,
      "loss": 1.4781,
      "step": 2245
    },
    {
      "epoch": 0.4936263736263736,
      "grad_norm": 0.2611441766796232,
      "learning_rate": 3.913069299409667e-07,
      "loss": 1.5186,
      "step": 2246
    },
    {
      "epoch": 0.4938461538461538,
      "grad_norm": 0.2397190545356055,
      "learning_rate": 3.9108947717795463e-07,
      "loss": 1.4307,
      "step": 2247
    },
    {
      "epoch": 0.4940659340659341,
      "grad_norm": 0.23909845500613314,
      "learning_rate": 3.908720215118764e-07,
      "loss": 1.4208,
      "step": 2248
    },
    {
      "epoch": 0.4942857142857143,
      "grad_norm": 0.28326644231504894,
      "learning_rate": 3.906545630464007e-07,
      "loss": 1.4224,
      "step": 2249
    },
    {
      "epoch": 0.4945054945054945,
      "grad_norm": 0.24633437801473324,
      "learning_rate": 3.9043710188519735e-07,
      "loss": 1.4887,
      "step": 2250
    },
    {
      "epoch": 0.4947252747252747,
      "grad_norm": 0.23904916861557024,
      "learning_rate": 3.90219638131938e-07,
      "loss": 1.4118,
      "step": 2251
    },
    {
      "epoch": 0.4949450549450549,
      "grad_norm": 0.25677412191404325,
      "learning_rate": 3.90002171890295e-07,
      "loss": 1.5036,
      "step": 2252
    },
    {
      "epoch": 0.49516483516483517,
      "grad_norm": 0.25005759102015573,
      "learning_rate": 3.8978470326394214e-07,
      "loss": 1.4814,
      "step": 2253
    },
    {
      "epoch": 0.49538461538461537,
      "grad_norm": 0.2581993772151251,
      "learning_rate": 3.8956723235655426e-07,
      "loss": 1.4622,
      "step": 2254
    },
    {
      "epoch": 0.4956043956043956,
      "grad_norm": 0.2609294834516711,
      "learning_rate": 3.893497592718074e-07,
      "loss": 1.4481,
      "step": 2255
    },
    {
      "epoch": 0.4958241758241758,
      "grad_norm": 0.25723709715413734,
      "learning_rate": 3.891322841133786e-07,
      "loss": 1.4517,
      "step": 2256
    },
    {
      "epoch": 0.49604395604395607,
      "grad_norm": 0.25155775919517426,
      "learning_rate": 3.889148069849457e-07,
      "loss": 1.5171,
      "step": 2257
    },
    {
      "epoch": 0.49626373626373627,
      "grad_norm": 0.2620557653630709,
      "learning_rate": 3.8869732799018784e-07,
      "loss": 1.4489,
      "step": 2258
    },
    {
      "epoch": 0.49648351648351646,
      "grad_norm": 0.25002331863705707,
      "learning_rate": 3.884798472327848e-07,
      "loss": 1.5076,
      "step": 2259
    },
    {
      "epoch": 0.4967032967032967,
      "grad_norm": 0.2656186939595916,
      "learning_rate": 3.8826236481641695e-07,
      "loss": 1.5147,
      "step": 2260
    },
    {
      "epoch": 0.4969230769230769,
      "grad_norm": 0.26114526317827286,
      "learning_rate": 3.880448808447662e-07,
      "loss": 1.4475,
      "step": 2261
    },
    {
      "epoch": 0.49714285714285716,
      "grad_norm": 0.24379211496905664,
      "learning_rate": 3.878273954215143e-07,
      "loss": 1.4233,
      "step": 2262
    },
    {
      "epoch": 0.49736263736263736,
      "grad_norm": 0.242500199968574,
      "learning_rate": 3.876099086503445e-07,
      "loss": 1.4413,
      "step": 2263
    },
    {
      "epoch": 0.49758241758241756,
      "grad_norm": 0.2546738974714412,
      "learning_rate": 3.873924206349403e-07,
      "loss": 1.4875,
      "step": 2264
    },
    {
      "epoch": 0.4978021978021978,
      "grad_norm": 0.2514541612117798,
      "learning_rate": 3.871749314789855e-07,
      "loss": 1.4893,
      "step": 2265
    },
    {
      "epoch": 0.498021978021978,
      "grad_norm": 0.24914408880178057,
      "learning_rate": 3.8695744128616505e-07,
      "loss": 1.3969,
      "step": 2266
    },
    {
      "epoch": 0.49824175824175826,
      "grad_norm": 0.2474244656953845,
      "learning_rate": 3.867399501601641e-07,
      "loss": 1.486,
      "step": 2267
    },
    {
      "epoch": 0.49846153846153846,
      "grad_norm": 0.24438263305872812,
      "learning_rate": 3.865224582046682e-07,
      "loss": 1.419,
      "step": 2268
    },
    {
      "epoch": 0.4986813186813187,
      "grad_norm": 0.2629258652096395,
      "learning_rate": 3.8630496552336336e-07,
      "loss": 1.5157,
      "step": 2269
    },
    {
      "epoch": 0.4989010989010989,
      "grad_norm": 0.2409382566658195,
      "learning_rate": 3.8608747221993594e-07,
      "loss": 1.4329,
      "step": 2270
    },
    {
      "epoch": 0.4991208791208791,
      "grad_norm": 0.24623568377332633,
      "learning_rate": 3.858699783980726e-07,
      "loss": 1.4415,
      "step": 2271
    },
    {
      "epoch": 0.49934065934065935,
      "grad_norm": 0.2539052101987305,
      "learning_rate": 3.8565248416146007e-07,
      "loss": 1.4709,
      "step": 2272
    },
    {
      "epoch": 0.49956043956043955,
      "grad_norm": 0.24133238747541236,
      "learning_rate": 3.8543498961378567e-07,
      "loss": 1.4745,
      "step": 2273
    },
    {
      "epoch": 0.4997802197802198,
      "grad_norm": 0.2557776072848505,
      "learning_rate": 3.852174948587365e-07,
      "loss": 1.4667,
      "step": 2274
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5003240500311481,
      "learning_rate": 3.85e-07,
      "loss": 1.4184,
      "step": 2275
    },
    {
      "epoch": 0.5002197802197802,
      "grad_norm": 0.24841473830266936,
      "learning_rate": 3.8478250514126357e-07,
      "loss": 1.4785,
      "step": 2276
    },
    {
      "epoch": 0.5004395604395604,
      "grad_norm": 0.24033681185054634,
      "learning_rate": 3.8456501038621437e-07,
      "loss": 1.4664,
      "step": 2277
    },
    {
      "epoch": 0.5006593406593407,
      "grad_norm": 0.25073011844641474,
      "learning_rate": 3.8434751583853986e-07,
      "loss": 1.4608,
      "step": 2278
    },
    {
      "epoch": 0.5008791208791209,
      "grad_norm": 0.26313066571800764,
      "learning_rate": 3.8413002160192746e-07,
      "loss": 1.4739,
      "step": 2279
    },
    {
      "epoch": 0.5010989010989011,
      "grad_norm": 0.24628034670922833,
      "learning_rate": 3.8391252778006415e-07,
      "loss": 1.4514,
      "step": 2280
    },
    {
      "epoch": 0.5013186813186813,
      "grad_norm": 0.25819247247267896,
      "learning_rate": 3.836950344766366e-07,
      "loss": 1.4775,
      "step": 2281
    },
    {
      "epoch": 0.5015384615384615,
      "grad_norm": 0.32968273982251245,
      "learning_rate": 3.8347754179533187e-07,
      "loss": 1.4397,
      "step": 2282
    },
    {
      "epoch": 0.5017582417582418,
      "grad_norm": 0.25337506380009656,
      "learning_rate": 3.8326004983983584e-07,
      "loss": 1.4456,
      "step": 2283
    },
    {
      "epoch": 0.501978021978022,
      "grad_norm": 0.2509531393309275,
      "learning_rate": 3.83042558713835e-07,
      "loss": 1.4818,
      "step": 2284
    },
    {
      "epoch": 0.5021978021978022,
      "grad_norm": 0.2554138807544182,
      "learning_rate": 3.8282506852101446e-07,
      "loss": 1.4299,
      "step": 2285
    },
    {
      "epoch": 0.5024175824175824,
      "grad_norm": 0.24884345900933796,
      "learning_rate": 3.8260757936505983e-07,
      "loss": 1.4719,
      "step": 2286
    },
    {
      "epoch": 0.5026373626373626,
      "grad_norm": 0.2567318242500976,
      "learning_rate": 3.8239009134965556e-07,
      "loss": 1.5205,
      "step": 2287
    },
    {
      "epoch": 0.5028571428571429,
      "grad_norm": 0.28827141519980987,
      "learning_rate": 3.821726045784856e-07,
      "loss": 1.4994,
      "step": 2288
    },
    {
      "epoch": 0.5030769230769231,
      "grad_norm": 0.25402458703562336,
      "learning_rate": 3.8195511915523386e-07,
      "loss": 1.4988,
      "step": 2289
    },
    {
      "epoch": 0.5032967032967033,
      "grad_norm": 0.2568810361409908,
      "learning_rate": 3.8173763518358304e-07,
      "loss": 1.5159,
      "step": 2290
    },
    {
      "epoch": 0.5035164835164835,
      "grad_norm": 0.2509320041521234,
      "learning_rate": 3.8152015276721535e-07,
      "loss": 1.4442,
      "step": 2291
    },
    {
      "epoch": 0.5037362637362638,
      "grad_norm": 0.24210277332773947,
      "learning_rate": 3.813026720098121e-07,
      "loss": 1.473,
      "step": 2292
    },
    {
      "epoch": 0.503956043956044,
      "grad_norm": 0.23952405663008583,
      "learning_rate": 3.810851930150542e-07,
      "loss": 1.3874,
      "step": 2293
    },
    {
      "epoch": 0.5041758241758242,
      "grad_norm": 0.24393557987251238,
      "learning_rate": 3.8086771588662143e-07,
      "loss": 1.4558,
      "step": 2294
    },
    {
      "epoch": 0.5043956043956044,
      "grad_norm": 0.23736793767225717,
      "learning_rate": 3.806502407281926e-07,
      "loss": 1.4779,
      "step": 2295
    },
    {
      "epoch": 0.5046153846153846,
      "grad_norm": 0.2532361984171458,
      "learning_rate": 3.804327676434458e-07,
      "loss": 1.4287,
      "step": 2296
    },
    {
      "epoch": 0.5048351648351649,
      "grad_norm": 0.26654186227997895,
      "learning_rate": 3.8021529673605785e-07,
      "loss": 1.5512,
      "step": 2297
    },
    {
      "epoch": 0.5050549450549451,
      "grad_norm": 0.2427767085812242,
      "learning_rate": 3.799978281097051e-07,
      "loss": 1.4286,
      "step": 2298
    },
    {
      "epoch": 0.5052747252747253,
      "grad_norm": 0.26393950646934833,
      "learning_rate": 3.7978036186806204e-07,
      "loss": 1.5118,
      "step": 2299
    },
    {
      "epoch": 0.5054945054945055,
      "grad_norm": 0.27442419825958025,
      "learning_rate": 3.7956289811480263e-07,
      "loss": 1.5133,
      "step": 2300
    },
    {
      "epoch": 0.5057142857142857,
      "grad_norm": 0.25219356781392155,
      "learning_rate": 3.7934543695359947e-07,
      "loss": 1.4664,
      "step": 2301
    },
    {
      "epoch": 0.505934065934066,
      "grad_norm": 0.2552058890419149,
      "learning_rate": 3.7912797848812355e-07,
      "loss": 1.4977,
      "step": 2302
    },
    {
      "epoch": 0.5061538461538462,
      "grad_norm": 0.24694517782102704,
      "learning_rate": 3.789105228220453e-07,
      "loss": 1.4942,
      "step": 2303
    },
    {
      "epoch": 0.5063736263736264,
      "grad_norm": 0.25489159416406026,
      "learning_rate": 3.7869307005903325e-07,
      "loss": 1.4696,
      "step": 2304
    },
    {
      "epoch": 0.5065934065934066,
      "grad_norm": 0.25201538035174836,
      "learning_rate": 3.784756203027548e-07,
      "loss": 1.4254,
      "step": 2305
    },
    {
      "epoch": 0.5068131868131868,
      "grad_norm": 0.2738964359143157,
      "learning_rate": 3.7825817365687573e-07,
      "loss": 1.4718,
      "step": 2306
    },
    {
      "epoch": 0.5070329670329671,
      "grad_norm": 0.251695545088705,
      "learning_rate": 3.780407302250605e-07,
      "loss": 1.4369,
      "step": 2307
    },
    {
      "epoch": 0.5072527472527473,
      "grad_norm": 0.24599154493338135,
      "learning_rate": 3.7782329011097185e-07,
      "loss": 1.4624,
      "step": 2308
    },
    {
      "epoch": 0.5074725274725275,
      "grad_norm": 0.24667844501736866,
      "learning_rate": 3.776058534182713e-07,
      "loss": 1.4633,
      "step": 2309
    },
    {
      "epoch": 0.5076923076923077,
      "grad_norm": 0.29337737433785255,
      "learning_rate": 3.773884202506184e-07,
      "loss": 1.4483,
      "step": 2310
    },
    {
      "epoch": 0.5079120879120879,
      "grad_norm": 0.25017157663708117,
      "learning_rate": 3.7717099071167107e-07,
      "loss": 1.4527,
      "step": 2311
    },
    {
      "epoch": 0.5081318681318682,
      "grad_norm": 0.2493125023934487,
      "learning_rate": 3.7695356490508547e-07,
      "loss": 1.5364,
      "step": 2312
    },
    {
      "epoch": 0.5083516483516484,
      "grad_norm": 0.2552680950194569,
      "learning_rate": 3.767361429345164e-07,
      "loss": 1.3963,
      "step": 2313
    },
    {
      "epoch": 0.5085714285714286,
      "grad_norm": 0.2527747074759333,
      "learning_rate": 3.7651872490361625e-07,
      "loss": 1.4298,
      "step": 2314
    },
    {
      "epoch": 0.5087912087912088,
      "grad_norm": 0.270320247819433,
      "learning_rate": 3.763013109160358e-07,
      "loss": 1.4305,
      "step": 2315
    },
    {
      "epoch": 0.509010989010989,
      "grad_norm": 0.2501930406554101,
      "learning_rate": 3.760839010754239e-07,
      "loss": 1.4205,
      "step": 2316
    },
    {
      "epoch": 0.5092307692307693,
      "grad_norm": 0.24627255884646854,
      "learning_rate": 3.7586649548542736e-07,
      "loss": 1.4634,
      "step": 2317
    },
    {
      "epoch": 0.5094505494505495,
      "grad_norm": 0.2594413697324097,
      "learning_rate": 3.756490942496912e-07,
      "loss": 1.4296,
      "step": 2318
    },
    {
      "epoch": 0.5096703296703297,
      "grad_norm": 0.241443154074072,
      "learning_rate": 3.75431697471858e-07,
      "loss": 1.4214,
      "step": 2319
    },
    {
      "epoch": 0.5098901098901099,
      "grad_norm": 0.32001564128841054,
      "learning_rate": 3.7521430525556846e-07,
      "loss": 1.491,
      "step": 2320
    },
    {
      "epoch": 0.5101098901098902,
      "grad_norm": 0.24272898905556806,
      "learning_rate": 3.7499691770446083e-07,
      "loss": 1.432,
      "step": 2321
    },
    {
      "epoch": 0.5103296703296704,
      "grad_norm": 0.24399153979616894,
      "learning_rate": 3.7477953492217173e-07,
      "loss": 1.4484,
      "step": 2322
    },
    {
      "epoch": 0.5105494505494506,
      "grad_norm": 0.2424576699450368,
      "learning_rate": 3.7456215701233473e-07,
      "loss": 1.5475,
      "step": 2323
    },
    {
      "epoch": 0.5107692307692308,
      "grad_norm": 0.25864494452289266,
      "learning_rate": 3.743447840785818e-07,
      "loss": 1.4577,
      "step": 2324
    },
    {
      "epoch": 0.510989010989011,
      "grad_norm": 0.24964225806607046,
      "learning_rate": 3.7412741622454205e-07,
      "loss": 1.4903,
      "step": 2325
    },
    {
      "epoch": 0.5112087912087913,
      "grad_norm": 0.24474201190457495,
      "learning_rate": 3.7391005355384205e-07,
      "loss": 1.4525,
      "step": 2326
    },
    {
      "epoch": 0.5114285714285715,
      "grad_norm": 0.24910505365666538,
      "learning_rate": 3.7369269617010655e-07,
      "loss": 1.4498,
      "step": 2327
    },
    {
      "epoch": 0.5116483516483517,
      "grad_norm": 0.2393738418338155,
      "learning_rate": 3.7347534417695734e-07,
      "loss": 1.4433,
      "step": 2328
    },
    {
      "epoch": 0.5118681318681318,
      "grad_norm": 0.24786669020425164,
      "learning_rate": 3.7325799767801365e-07,
      "loss": 1.4739,
      "step": 2329
    },
    {
      "epoch": 0.512087912087912,
      "grad_norm": 0.2495661564832968,
      "learning_rate": 3.7304065677689206e-07,
      "loss": 1.4453,
      "step": 2330
    },
    {
      "epoch": 0.5123076923076924,
      "grad_norm": 0.25039328270074224,
      "learning_rate": 3.728233215772065e-07,
      "loss": 1.4564,
      "step": 2331
    },
    {
      "epoch": 0.5125274725274725,
      "grad_norm": 0.24682075686811833,
      "learning_rate": 3.726059921825685e-07,
      "loss": 1.4693,
      "step": 2332
    },
    {
      "epoch": 0.5127472527472527,
      "grad_norm": 0.2539640526472173,
      "learning_rate": 3.723886686965864e-07,
      "loss": 1.51,
      "step": 2333
    },
    {
      "epoch": 0.5129670329670329,
      "grad_norm": 0.238444959442042,
      "learning_rate": 3.7217135122286586e-07,
      "loss": 1.3958,
      "step": 2334
    },
    {
      "epoch": 0.5131868131868131,
      "grad_norm": 0.24104629604246275,
      "learning_rate": 3.719540398650099e-07,
      "loss": 1.4758,
      "step": 2335
    },
    {
      "epoch": 0.5134065934065934,
      "grad_norm": 0.3270237259906149,
      "learning_rate": 3.717367347266181e-07,
      "loss": 1.5671,
      "step": 2336
    },
    {
      "epoch": 0.5136263736263736,
      "grad_norm": 0.2456132693039993,
      "learning_rate": 3.7151943591128765e-07,
      "loss": 1.4872,
      "step": 2337
    },
    {
      "epoch": 0.5138461538461538,
      "grad_norm": 0.2448230376335242,
      "learning_rate": 3.7130214352261233e-07,
      "loss": 1.4548,
      "step": 2338
    },
    {
      "epoch": 0.514065934065934,
      "grad_norm": 0.24191589322092183,
      "learning_rate": 3.7108485766418315e-07,
      "loss": 1.4354,
      "step": 2339
    },
    {
      "epoch": 0.5142857142857142,
      "grad_norm": 0.26527616434014223,
      "learning_rate": 3.7086757843958796e-07,
      "loss": 1.4766,
      "step": 2340
    },
    {
      "epoch": 0.5145054945054945,
      "grad_norm": 0.2401466273453965,
      "learning_rate": 3.706503059524108e-07,
      "loss": 1.4633,
      "step": 2341
    },
    {
      "epoch": 0.5147252747252747,
      "grad_norm": 0.2476514638882906,
      "learning_rate": 3.704330403062336e-07,
      "loss": 1.491,
      "step": 2342
    },
    {
      "epoch": 0.5149450549450549,
      "grad_norm": 0.25424091238816915,
      "learning_rate": 3.702157816046344e-07,
      "loss": 1.4508,
      "step": 2343
    },
    {
      "epoch": 0.5151648351648351,
      "grad_norm": 0.2404497274348749,
      "learning_rate": 3.6999852995118784e-07,
      "loss": 1.4898,
      "step": 2344
    },
    {
      "epoch": 0.5153846153846153,
      "grad_norm": 0.2575112655941573,
      "learning_rate": 3.6978128544946524e-07,
      "loss": 1.4813,
      "step": 2345
    },
    {
      "epoch": 0.5156043956043956,
      "grad_norm": 0.35965713375661085,
      "learning_rate": 3.6956404820303496e-07,
      "loss": 1.4244,
      "step": 2346
    },
    {
      "epoch": 0.5158241758241758,
      "grad_norm": 0.25568518865603723,
      "learning_rate": 3.693468183154614e-07,
      "loss": 1.4547,
      "step": 2347
    },
    {
      "epoch": 0.516043956043956,
      "grad_norm": 0.254028026851914,
      "learning_rate": 3.6912959589030573e-07,
      "loss": 1.484,
      "step": 2348
    },
    {
      "epoch": 0.5162637362637362,
      "grad_norm": 0.26235289216398056,
      "learning_rate": 3.689123810311252e-07,
      "loss": 1.5028,
      "step": 2349
    },
    {
      "epoch": 0.5164835164835165,
      "grad_norm": 0.24235662101047564,
      "learning_rate": 3.68695173841474e-07,
      "loss": 1.4727,
      "step": 2350
    },
    {
      "epoch": 0.5167032967032967,
      "grad_norm": 0.24435031157677387,
      "learning_rate": 3.6847797442490226e-07,
      "loss": 1.4743,
      "step": 2351
    },
    {
      "epoch": 0.5169230769230769,
      "grad_norm": 0.2460452477834752,
      "learning_rate": 3.6826078288495654e-07,
      "loss": 1.4337,
      "step": 2352
    },
    {
      "epoch": 0.5171428571428571,
      "grad_norm": 0.23683375278787086,
      "learning_rate": 3.680435993251796e-07,
      "loss": 1.4824,
      "step": 2353
    },
    {
      "epoch": 0.5173626373626373,
      "grad_norm": 0.26491693796020277,
      "learning_rate": 3.678264238491105e-07,
      "loss": 1.48,
      "step": 2354
    },
    {
      "epoch": 0.5175824175824176,
      "grad_norm": 0.27656080165220287,
      "learning_rate": 3.6760925656028407e-07,
      "loss": 1.4828,
      "step": 2355
    },
    {
      "epoch": 0.5178021978021978,
      "grad_norm": 0.2538831484175234,
      "learning_rate": 3.67392097562232e-07,
      "loss": 1.4679,
      "step": 2356
    },
    {
      "epoch": 0.518021978021978,
      "grad_norm": 0.24301818447540044,
      "learning_rate": 3.671749469584811e-07,
      "loss": 1.5261,
      "step": 2357
    },
    {
      "epoch": 0.5182417582417582,
      "grad_norm": 0.6295033281908831,
      "learning_rate": 3.669578048525551e-07,
      "loss": 1.482,
      "step": 2358
    },
    {
      "epoch": 0.5184615384615384,
      "grad_norm": 0.24685397208179988,
      "learning_rate": 3.667406713479729e-07,
      "loss": 1.4516,
      "step": 2359
    },
    {
      "epoch": 0.5186813186813187,
      "grad_norm": 0.24234241454696095,
      "learning_rate": 3.6652354654824957e-07,
      "loss": 1.467,
      "step": 2360
    },
    {
      "epoch": 0.5189010989010989,
      "grad_norm": 0.2453436935073063,
      "learning_rate": 3.6630643055689623e-07,
      "loss": 1.477,
      "step": 2361
    },
    {
      "epoch": 0.5191208791208791,
      "grad_norm": 0.25091548919956697,
      "learning_rate": 3.6608932347741974e-07,
      "loss": 1.474,
      "step": 2362
    },
    {
      "epoch": 0.5193406593406593,
      "grad_norm": 0.26462417732227744,
      "learning_rate": 3.6587222541332244e-07,
      "loss": 1.4865,
      "step": 2363
    },
    {
      "epoch": 0.5195604395604395,
      "grad_norm": 0.23547926738031363,
      "learning_rate": 3.656551364681026e-07,
      "loss": 1.4548,
      "step": 2364
    },
    {
      "epoch": 0.5197802197802198,
      "grad_norm": 0.25248058408821283,
      "learning_rate": 3.65438056745254e-07,
      "loss": 1.4881,
      "step": 2365
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.24639762570360496,
      "learning_rate": 3.652209863482663e-07,
      "loss": 1.4333,
      "step": 2366
    },
    {
      "epoch": 0.5202197802197802,
      "grad_norm": 0.2384335820996811,
      "learning_rate": 3.6500392538062447e-07,
      "loss": 1.4922,
      "step": 2367
    },
    {
      "epoch": 0.5204395604395604,
      "grad_norm": 0.23938737681849095,
      "learning_rate": 3.6478687394580897e-07,
      "loss": 1.4098,
      "step": 2368
    },
    {
      "epoch": 0.5206593406593406,
      "grad_norm": 0.25775289904758764,
      "learning_rate": 3.645698321472959e-07,
      "loss": 1.5393,
      "step": 2369
    },
    {
      "epoch": 0.5208791208791209,
      "grad_norm": 0.2534509396457383,
      "learning_rate": 3.643528000885565e-07,
      "loss": 1.5131,
      "step": 2370
    },
    {
      "epoch": 0.5210989010989011,
      "grad_norm": 0.25161087694382234,
      "learning_rate": 3.641357778730578e-07,
      "loss": 1.5225,
      "step": 2371
    },
    {
      "epoch": 0.5213186813186813,
      "grad_norm": 0.2531132919378715,
      "learning_rate": 3.6391876560426156e-07,
      "loss": 1.3885,
      "step": 2372
    },
    {
      "epoch": 0.5215384615384615,
      "grad_norm": 0.27678118391000595,
      "learning_rate": 3.637017633856253e-07,
      "loss": 1.4796,
      "step": 2373
    },
    {
      "epoch": 0.5217582417582417,
      "grad_norm": 0.3372078287193749,
      "learning_rate": 3.6348477132060146e-07,
      "loss": 1.4785,
      "step": 2374
    },
    {
      "epoch": 0.521978021978022,
      "grad_norm": 0.3654889146461077,
      "learning_rate": 3.6326778951263765e-07,
      "loss": 1.4112,
      "step": 2375
    },
    {
      "epoch": 0.5221978021978022,
      "grad_norm": 0.2539579037334113,
      "learning_rate": 3.6305081806517676e-07,
      "loss": 1.5093,
      "step": 2376
    },
    {
      "epoch": 0.5224175824175824,
      "grad_norm": 0.24155137687989628,
      "learning_rate": 3.628338570816568e-07,
      "loss": 1.4028,
      "step": 2377
    },
    {
      "epoch": 0.5226373626373626,
      "grad_norm": 0.29159047625470025,
      "learning_rate": 3.626169066655105e-07,
      "loss": 1.5091,
      "step": 2378
    },
    {
      "epoch": 0.5228571428571429,
      "grad_norm": 0.25993978508890914,
      "learning_rate": 3.623999669201655e-07,
      "loss": 1.4699,
      "step": 2379
    },
    {
      "epoch": 0.5230769230769231,
      "grad_norm": 0.25977530015379996,
      "learning_rate": 3.621830379490449e-07,
      "loss": 1.4666,
      "step": 2380
    },
    {
      "epoch": 0.5232967032967033,
      "grad_norm": 0.2549597264906517,
      "learning_rate": 3.61966119855566e-07,
      "loss": 1.4371,
      "step": 2381
    },
    {
      "epoch": 0.5235164835164835,
      "grad_norm": 0.28112919059338826,
      "learning_rate": 3.6174921274314146e-07,
      "loss": 1.427,
      "step": 2382
    },
    {
      "epoch": 0.5237362637362637,
      "grad_norm": 0.25240146760536947,
      "learning_rate": 3.615323167151783e-07,
      "loss": 1.4486,
      "step": 2383
    },
    {
      "epoch": 0.523956043956044,
      "grad_norm": 0.2664580405551836,
      "learning_rate": 3.6131543187507853e-07,
      "loss": 1.4713,
      "step": 2384
    },
    {
      "epoch": 0.5241758241758242,
      "grad_norm": 0.2500998703829329,
      "learning_rate": 3.610985583262387e-07,
      "loss": 1.5272,
      "step": 2385
    },
    {
      "epoch": 0.5243956043956044,
      "grad_norm": 0.2525705996196096,
      "learning_rate": 3.6088169617204997e-07,
      "loss": 1.4412,
      "step": 2386
    },
    {
      "epoch": 0.5246153846153846,
      "grad_norm": 0.25000485621733065,
      "learning_rate": 3.6066484551589807e-07,
      "loss": 1.4597,
      "step": 2387
    },
    {
      "epoch": 0.5248351648351648,
      "grad_norm": 0.2410643399310685,
      "learning_rate": 3.604480064611634e-07,
      "loss": 1.4532,
      "step": 2388
    },
    {
      "epoch": 0.5250549450549451,
      "grad_norm": 0.24446312178585083,
      "learning_rate": 3.6023117911122043e-07,
      "loss": 1.5031,
      "step": 2389
    },
    {
      "epoch": 0.5252747252747253,
      "grad_norm": 0.3362095219381454,
      "learning_rate": 3.6001436356943865e-07,
      "loss": 1.5374,
      "step": 2390
    },
    {
      "epoch": 0.5254945054945055,
      "grad_norm": 0.30150323793043726,
      "learning_rate": 3.597975599391814e-07,
      "loss": 1.4848,
      "step": 2391
    },
    {
      "epoch": 0.5257142857142857,
      "grad_norm": 0.24788575059663162,
      "learning_rate": 3.595807683238066e-07,
      "loss": 1.4805,
      "step": 2392
    },
    {
      "epoch": 0.5259340659340659,
      "grad_norm": 0.2607277845856362,
      "learning_rate": 3.5936398882666633e-07,
      "loss": 1.4144,
      "step": 2393
    },
    {
      "epoch": 0.5261538461538462,
      "grad_norm": 0.24685913931142686,
      "learning_rate": 3.591472215511069e-07,
      "loss": 1.4582,
      "step": 2394
    },
    {
      "epoch": 0.5263736263736264,
      "grad_norm": 0.5962223978399409,
      "learning_rate": 3.58930466600469e-07,
      "loss": 1.4715,
      "step": 2395
    },
    {
      "epoch": 0.5265934065934066,
      "grad_norm": 0.26155916145065927,
      "learning_rate": 3.587137240780871e-07,
      "loss": 1.4348,
      "step": 2396
    },
    {
      "epoch": 0.5268131868131868,
      "grad_norm": 0.25033706471393447,
      "learning_rate": 3.5849699408729007e-07,
      "loss": 1.4629,
      "step": 2397
    },
    {
      "epoch": 0.527032967032967,
      "grad_norm": 0.24636054493260834,
      "learning_rate": 3.582802767314005e-07,
      "loss": 1.4718,
      "step": 2398
    },
    {
      "epoch": 0.5272527472527473,
      "grad_norm": 0.26061079364117784,
      "learning_rate": 3.580635721137353e-07,
      "loss": 1.425,
      "step": 2399
    },
    {
      "epoch": 0.5274725274725275,
      "grad_norm": 0.27835332660121304,
      "learning_rate": 3.5784688033760484e-07,
      "loss": 1.4561,
      "step": 2400
    },
    {
      "epoch": 0.5276923076923077,
      "grad_norm": 0.24729440919091983,
      "learning_rate": 3.5763020150631404e-07,
      "loss": 1.4787,
      "step": 2401
    },
    {
      "epoch": 0.5279120879120879,
      "grad_norm": 0.24420608479735176,
      "learning_rate": 3.574135357231609e-07,
      "loss": 1.4738,
      "step": 2402
    },
    {
      "epoch": 0.5281318681318682,
      "grad_norm": 0.2557867687587444,
      "learning_rate": 3.5719688309143775e-07,
      "loss": 1.509,
      "step": 2403
    },
    {
      "epoch": 0.5283516483516484,
      "grad_norm": 0.25074565043670655,
      "learning_rate": 3.569802437144304e-07,
      "loss": 1.524,
      "step": 2404
    },
    {
      "epoch": 0.5285714285714286,
      "grad_norm": 0.2461662436249225,
      "learning_rate": 3.5676361769541846e-07,
      "loss": 1.3901,
      "step": 2405
    },
    {
      "epoch": 0.5287912087912088,
      "grad_norm": 0.24870628819652119,
      "learning_rate": 3.565470051376751e-07,
      "loss": 1.4639,
      "step": 2406
    },
    {
      "epoch": 0.529010989010989,
      "grad_norm": 0.2578878580655407,
      "learning_rate": 3.5633040614446693e-07,
      "loss": 1.4337,
      "step": 2407
    },
    {
      "epoch": 0.5292307692307693,
      "grad_norm": 0.24416041120687818,
      "learning_rate": 3.561138208190543e-07,
      "loss": 1.4747,
      "step": 2408
    },
    {
      "epoch": 0.5294505494505495,
      "grad_norm": 0.25531378396760357,
      "learning_rate": 3.558972492646912e-07,
      "loss": 1.4904,
      "step": 2409
    },
    {
      "epoch": 0.5296703296703297,
      "grad_norm": 0.2363396760204816,
      "learning_rate": 3.556806915846247e-07,
      "loss": 1.4463,
      "step": 2410
    },
    {
      "epoch": 0.5298901098901099,
      "grad_norm": 0.24943120497323146,
      "learning_rate": 3.554641478820954e-07,
      "loss": 1.4236,
      "step": 2411
    },
    {
      "epoch": 0.5301098901098901,
      "grad_norm": 0.24732171440482204,
      "learning_rate": 3.552476182603372e-07,
      "loss": 1.4489,
      "step": 2412
    },
    {
      "epoch": 0.5303296703296704,
      "grad_norm": 0.24306256922666883,
      "learning_rate": 3.5503110282257725e-07,
      "loss": 1.4377,
      "step": 2413
    },
    {
      "epoch": 0.5305494505494506,
      "grad_norm": 0.24425677091309583,
      "learning_rate": 3.5481460167203626e-07,
      "loss": 1.4757,
      "step": 2414
    },
    {
      "epoch": 0.5307692307692308,
      "grad_norm": 0.2529037931280104,
      "learning_rate": 3.545981149119277e-07,
      "loss": 1.4868,
      "step": 2415
    },
    {
      "epoch": 0.530989010989011,
      "grad_norm": 0.24632188014095946,
      "learning_rate": 3.5438164264545854e-07,
      "loss": 1.4856,
      "step": 2416
    },
    {
      "epoch": 0.5312087912087912,
      "grad_norm": 0.2513676700213414,
      "learning_rate": 3.5416518497582844e-07,
      "loss": 1.4474,
      "step": 2417
    },
    {
      "epoch": 0.5314285714285715,
      "grad_norm": 0.24316858435167316,
      "learning_rate": 3.5394874200623025e-07,
      "loss": 1.4611,
      "step": 2418
    },
    {
      "epoch": 0.5316483516483517,
      "grad_norm": 0.30859414643629013,
      "learning_rate": 3.537323138398502e-07,
      "loss": 1.4665,
      "step": 2419
    },
    {
      "epoch": 0.5318681318681319,
      "grad_norm": 0.25132634887271454,
      "learning_rate": 3.5351590057986706e-07,
      "loss": 1.4642,
      "step": 2420
    },
    {
      "epoch": 0.5320879120879121,
      "grad_norm": 0.24990117582506366,
      "learning_rate": 3.5329950232945247e-07,
      "loss": 1.4652,
      "step": 2421
    },
    {
      "epoch": 0.5323076923076923,
      "grad_norm": 0.25409742429800175,
      "learning_rate": 3.5308311919177103e-07,
      "loss": 1.471,
      "step": 2422
    },
    {
      "epoch": 0.5325274725274726,
      "grad_norm": 0.32455414234173374,
      "learning_rate": 3.5286675126998003e-07,
      "loss": 1.4346,
      "step": 2423
    },
    {
      "epoch": 0.5327472527472528,
      "grad_norm": 0.3291216222833815,
      "learning_rate": 3.5265039866722997e-07,
      "loss": 1.4416,
      "step": 2424
    },
    {
      "epoch": 0.532967032967033,
      "grad_norm": 0.39783825599437156,
      "learning_rate": 3.5243406148666347e-07,
      "loss": 1.475,
      "step": 2425
    },
    {
      "epoch": 0.5331868131868132,
      "grad_norm": 0.24359859701096273,
      "learning_rate": 3.522177398314159e-07,
      "loss": 1.5323,
      "step": 2426
    },
    {
      "epoch": 0.5334065934065934,
      "grad_norm": 0.2505957093291331,
      "learning_rate": 3.5200143380461553e-07,
      "loss": 1.5131,
      "step": 2427
    },
    {
      "epoch": 0.5336263736263737,
      "grad_norm": 0.2476887249243456,
      "learning_rate": 3.517851435093827e-07,
      "loss": 1.4365,
      "step": 2428
    },
    {
      "epoch": 0.5338461538461539,
      "grad_norm": 0.251634764989001,
      "learning_rate": 3.5156886904883096e-07,
      "loss": 1.5045,
      "step": 2429
    },
    {
      "epoch": 0.5340659340659341,
      "grad_norm": 0.24693035470009495,
      "learning_rate": 3.5135261052606556e-07,
      "loss": 1.4867,
      "step": 2430
    },
    {
      "epoch": 0.5342857142857143,
      "grad_norm": 0.2510114323493616,
      "learning_rate": 3.511363680441847e-07,
      "loss": 1.4704,
      "step": 2431
    },
    {
      "epoch": 0.5345054945054946,
      "grad_norm": 0.24413830266218428,
      "learning_rate": 3.5092014170627847e-07,
      "loss": 1.4791,
      "step": 2432
    },
    {
      "epoch": 0.5347252747252748,
      "grad_norm": 0.2549529871977361,
      "learning_rate": 3.5070393161542974e-07,
      "loss": 1.4105,
      "step": 2433
    },
    {
      "epoch": 0.534945054945055,
      "grad_norm": 0.24085825020125307,
      "learning_rate": 3.504877378747133e-07,
      "loss": 1.4169,
      "step": 2434
    },
    {
      "epoch": 0.5351648351648352,
      "grad_norm": 0.2460619199237114,
      "learning_rate": 3.5027156058719635e-07,
      "loss": 1.4424,
      "step": 2435
    },
    {
      "epoch": 0.5353846153846153,
      "grad_norm": 0.26898264466723765,
      "learning_rate": 3.50055399855938e-07,
      "loss": 1.4813,
      "step": 2436
    },
    {
      "epoch": 0.5356043956043957,
      "grad_norm": 0.23253205429484922,
      "learning_rate": 3.4983925578398957e-07,
      "loss": 1.416,
      "step": 2437
    },
    {
      "epoch": 0.5358241758241759,
      "grad_norm": 0.31045840835273486,
      "learning_rate": 3.496231284743946e-07,
      "loss": 1.4473,
      "step": 2438
    },
    {
      "epoch": 0.536043956043956,
      "grad_norm": 0.24932598271297657,
      "learning_rate": 3.494070180301886e-07,
      "loss": 1.4535,
      "step": 2439
    },
    {
      "epoch": 0.5362637362637362,
      "grad_norm": 0.25065753370385874,
      "learning_rate": 3.491909245543989e-07,
      "loss": 1.4426,
      "step": 2440
    },
    {
      "epoch": 0.5364835164835164,
      "grad_norm": 0.26493054254675447,
      "learning_rate": 3.4897484815004457e-07,
      "loss": 1.5287,
      "step": 2441
    },
    {
      "epoch": 0.5367032967032968,
      "grad_norm": 0.2609973851037939,
      "learning_rate": 3.4875878892013703e-07,
      "loss": 1.4704,
      "step": 2442
    },
    {
      "epoch": 0.536923076923077,
      "grad_norm": 0.25556647371724434,
      "learning_rate": 3.4854274696767927e-07,
      "loss": 1.4117,
      "step": 2443
    },
    {
      "epoch": 0.5371428571428571,
      "grad_norm": 0.26521611235385095,
      "learning_rate": 3.4832672239566584e-07,
      "loss": 1.4343,
      "step": 2444
    },
    {
      "epoch": 0.5373626373626373,
      "grad_norm": 0.24173566804949828,
      "learning_rate": 3.481107153070833e-07,
      "loss": 1.435,
      "step": 2445
    },
    {
      "epoch": 0.5375824175824175,
      "grad_norm": 0.24424325127841187,
      "learning_rate": 3.4789472580490975e-07,
      "loss": 1.45,
      "step": 2446
    },
    {
      "epoch": 0.5378021978021978,
      "grad_norm": 0.244678264719358,
      "learning_rate": 3.476787539921149e-07,
      "loss": 1.4613,
      "step": 2447
    },
    {
      "epoch": 0.538021978021978,
      "grad_norm": 0.25787588823262375,
      "learning_rate": 3.4746279997166003e-07,
      "loss": 1.5005,
      "step": 2448
    },
    {
      "epoch": 0.5382417582417582,
      "grad_norm": 0.23894036858444787,
      "learning_rate": 3.472468638464981e-07,
      "loss": 1.4903,
      "step": 2449
    },
    {
      "epoch": 0.5384615384615384,
      "grad_norm": 0.2637014488813191,
      "learning_rate": 3.470309457195733e-07,
      "loss": 1.4404,
      "step": 2450
    },
    {
      "epoch": 0.5386813186813186,
      "grad_norm": 0.24646079245948196,
      "learning_rate": 3.468150456938213e-07,
      "loss": 1.5028,
      "step": 2451
    },
    {
      "epoch": 0.5389010989010989,
      "grad_norm": 0.2573309172419129,
      "learning_rate": 3.46599163872169e-07,
      "loss": 1.4448,
      "step": 2452
    },
    {
      "epoch": 0.5391208791208791,
      "grad_norm": 0.2589449655414008,
      "learning_rate": 3.463833003575352e-07,
      "loss": 1.445,
      "step": 2453
    },
    {
      "epoch": 0.5393406593406593,
      "grad_norm": 0.5852414658697316,
      "learning_rate": 3.4616745525282934e-07,
      "loss": 1.4688,
      "step": 2454
    },
    {
      "epoch": 0.5395604395604395,
      "grad_norm": 0.2519595272985753,
      "learning_rate": 3.459516286609524e-07,
      "loss": 1.4515,
      "step": 2455
    },
    {
      "epoch": 0.5397802197802197,
      "grad_norm": 0.2900347008805642,
      "learning_rate": 3.457358206847962e-07,
      "loss": 1.4487,
      "step": 2456
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.25661106951172385,
      "learning_rate": 3.455200314272441e-07,
      "loss": 1.4514,
      "step": 2457
    },
    {
      "epoch": 0.5402197802197802,
      "grad_norm": 0.25732154087119125,
      "learning_rate": 3.453042609911706e-07,
      "loss": 1.4282,
      "step": 2458
    },
    {
      "epoch": 0.5404395604395604,
      "grad_norm": 0.24889408132909518,
      "learning_rate": 3.4508850947944057e-07,
      "loss": 1.4445,
      "step": 2459
    },
    {
      "epoch": 0.5406593406593406,
      "grad_norm": 0.2701462310034605,
      "learning_rate": 3.4487277699491056e-07,
      "loss": 1.4432,
      "step": 2460
    },
    {
      "epoch": 0.5408791208791209,
      "grad_norm": 0.2553952628077091,
      "learning_rate": 3.4465706364042764e-07,
      "loss": 1.5246,
      "step": 2461
    },
    {
      "epoch": 0.5410989010989011,
      "grad_norm": 0.27429146561848944,
      "learning_rate": 3.4444136951883004e-07,
      "loss": 1.4802,
      "step": 2462
    },
    {
      "epoch": 0.5413186813186813,
      "grad_norm": 0.3847981198035457,
      "learning_rate": 3.442256947329466e-07,
      "loss": 1.426,
      "step": 2463
    },
    {
      "epoch": 0.5415384615384615,
      "grad_norm": 0.2505304096279959,
      "learning_rate": 3.4401003938559693e-07,
      "loss": 1.4251,
      "step": 2464
    },
    {
      "epoch": 0.5417582417582417,
      "grad_norm": 0.2811376482113325,
      "learning_rate": 3.437944035795915e-07,
      "loss": 1.4811,
      "step": 2465
    },
    {
      "epoch": 0.541978021978022,
      "grad_norm": 0.25685745384779063,
      "learning_rate": 3.4357878741773145e-07,
      "loss": 1.5058,
      "step": 2466
    },
    {
      "epoch": 0.5421978021978022,
      "grad_norm": 0.2589697373282326,
      "learning_rate": 3.433631910028086e-07,
      "loss": 1.4322,
      "step": 2467
    },
    {
      "epoch": 0.5424175824175824,
      "grad_norm": 0.2653156802744679,
      "learning_rate": 3.431476144376051e-07,
      "loss": 1.4406,
      "step": 2468
    },
    {
      "epoch": 0.5426373626373626,
      "grad_norm": 0.2425175948787124,
      "learning_rate": 3.4293205782489406e-07,
      "loss": 1.4273,
      "step": 2469
    },
    {
      "epoch": 0.5428571428571428,
      "grad_norm": 0.28131479717165775,
      "learning_rate": 3.4271652126743866e-07,
      "loss": 1.4875,
      "step": 2470
    },
    {
      "epoch": 0.5430769230769231,
      "grad_norm": 0.2377766036111552,
      "learning_rate": 3.4250100486799256e-07,
      "loss": 1.4524,
      "step": 2471
    },
    {
      "epoch": 0.5432967032967033,
      "grad_norm": 0.2811295000176585,
      "learning_rate": 3.422855087293002e-07,
      "loss": 1.4719,
      "step": 2472
    },
    {
      "epoch": 0.5435164835164835,
      "grad_norm": 0.2533281053392722,
      "learning_rate": 3.4207003295409617e-07,
      "loss": 1.4821,
      "step": 2473
    },
    {
      "epoch": 0.5437362637362637,
      "grad_norm": 0.2523850823471961,
      "learning_rate": 3.418545776451051e-07,
      "loss": 1.4996,
      "step": 2474
    },
    {
      "epoch": 0.5439560439560439,
      "grad_norm": 0.26054012455088077,
      "learning_rate": 3.416391429050421e-07,
      "loss": 1.4399,
      "step": 2475
    },
    {
      "epoch": 0.5441758241758242,
      "grad_norm": 0.2623042118045894,
      "learning_rate": 3.4142372883661224e-07,
      "loss": 1.4756,
      "step": 2476
    },
    {
      "epoch": 0.5443956043956044,
      "grad_norm": 0.2523932975509699,
      "learning_rate": 3.412083355425113e-07,
      "loss": 1.5235,
      "step": 2477
    },
    {
      "epoch": 0.5446153846153846,
      "grad_norm": 0.24750151171386292,
      "learning_rate": 3.4099296312542465e-07,
      "loss": 1.4234,
      "step": 2478
    },
    {
      "epoch": 0.5448351648351648,
      "grad_norm": 0.24840472821833404,
      "learning_rate": 3.407776116880277e-07,
      "loss": 1.4051,
      "step": 2479
    },
    {
      "epoch": 0.545054945054945,
      "grad_norm": 0.2432141910966809,
      "learning_rate": 3.40562281332986e-07,
      "loss": 1.4631,
      "step": 2480
    },
    {
      "epoch": 0.5452747252747253,
      "grad_norm": 0.24425875081241077,
      "learning_rate": 3.4034697216295514e-07,
      "loss": 1.4807,
      "step": 2481
    },
    {
      "epoch": 0.5454945054945055,
      "grad_norm": 0.24913540952082738,
      "learning_rate": 3.401316842805806e-07,
      "loss": 1.5285,
      "step": 2482
    },
    {
      "epoch": 0.5457142857142857,
      "grad_norm": 0.2561140740036112,
      "learning_rate": 3.399164177884974e-07,
      "loss": 1.4678,
      "step": 2483
    },
    {
      "epoch": 0.5459340659340659,
      "grad_norm": 0.24527619813444076,
      "learning_rate": 3.3970117278933075e-07,
      "loss": 1.4615,
      "step": 2484
    },
    {
      "epoch": 0.5461538461538461,
      "grad_norm": 0.24093477258527357,
      "learning_rate": 3.394859493856954e-07,
      "loss": 1.4754,
      "step": 2485
    },
    {
      "epoch": 0.5463736263736264,
      "grad_norm": 0.2483508001877219,
      "learning_rate": 3.392707476801957e-07,
      "loss": 1.4846,
      "step": 2486
    },
    {
      "epoch": 0.5465934065934066,
      "grad_norm": 0.251498999901592,
      "learning_rate": 3.3905556777542603e-07,
      "loss": 1.5126,
      "step": 2487
    },
    {
      "epoch": 0.5468131868131868,
      "grad_norm": 0.292984358149494,
      "learning_rate": 3.388404097739702e-07,
      "loss": 1.4384,
      "step": 2488
    },
    {
      "epoch": 0.547032967032967,
      "grad_norm": 0.2471393152596394,
      "learning_rate": 3.386252737784014e-07,
      "loss": 1.4826,
      "step": 2489
    },
    {
      "epoch": 0.5472527472527473,
      "grad_norm": 0.26047142753984537,
      "learning_rate": 3.384101598912823e-07,
      "loss": 1.4981,
      "step": 2490
    },
    {
      "epoch": 0.5474725274725275,
      "grad_norm": 0.2461878978044061,
      "learning_rate": 3.3819506821516553e-07,
      "loss": 1.47,
      "step": 2491
    },
    {
      "epoch": 0.5476923076923077,
      "grad_norm": 0.23805161395936614,
      "learning_rate": 3.3797999885259277e-07,
      "loss": 1.4635,
      "step": 2492
    },
    {
      "epoch": 0.5479120879120879,
      "grad_norm": 0.35282272725126546,
      "learning_rate": 3.377649519060948e-07,
      "loss": 1.5261,
      "step": 2493
    },
    {
      "epoch": 0.5481318681318681,
      "grad_norm": 0.2466539109687882,
      "learning_rate": 3.3754992747819224e-07,
      "loss": 1.45,
      "step": 2494
    },
    {
      "epoch": 0.5483516483516484,
      "grad_norm": 0.2603889600133571,
      "learning_rate": 3.3733492567139457e-07,
      "loss": 1.4689,
      "step": 2495
    },
    {
      "epoch": 0.5485714285714286,
      "grad_norm": 0.25709555281243185,
      "learning_rate": 3.371199465882008e-07,
      "loss": 1.4679,
      "step": 2496
    },
    {
      "epoch": 0.5487912087912088,
      "grad_norm": 0.2568533673561895,
      "learning_rate": 3.36904990331099e-07,
      "loss": 1.4868,
      "step": 2497
    },
    {
      "epoch": 0.549010989010989,
      "grad_norm": 0.24963793309196405,
      "learning_rate": 3.366900570025661e-07,
      "loss": 1.4055,
      "step": 2498
    },
    {
      "epoch": 0.5492307692307692,
      "grad_norm": 0.2475589390440301,
      "learning_rate": 3.364751467050686e-07,
      "loss": 1.4755,
      "step": 2499
    },
    {
      "epoch": 0.5494505494505495,
      "grad_norm": 0.2747454061930324,
      "learning_rate": 3.362602595410613e-07,
      "loss": 1.4102,
      "step": 2500
    },
    {
      "epoch": 0.5496703296703297,
      "grad_norm": 0.2627576856680716,
      "learning_rate": 3.3604539561298903e-07,
      "loss": 1.47,
      "step": 2501
    },
    {
      "epoch": 0.5498901098901099,
      "grad_norm": 0.26130260962110075,
      "learning_rate": 3.3583055502328437e-07,
      "loss": 1.4188,
      "step": 2502
    },
    {
      "epoch": 0.5501098901098901,
      "grad_norm": 0.2533795146156557,
      "learning_rate": 3.3561573787436954e-07,
      "loss": 1.4421,
      "step": 2503
    },
    {
      "epoch": 0.5503296703296703,
      "grad_norm": 0.2549711093400903,
      "learning_rate": 3.354009442686555e-07,
      "loss": 1.4564,
      "step": 2504
    },
    {
      "epoch": 0.5505494505494506,
      "grad_norm": 0.2596310640387029,
      "learning_rate": 3.351861743085415e-07,
      "loss": 1.4048,
      "step": 2505
    },
    {
      "epoch": 0.5507692307692308,
      "grad_norm": 0.2585193367500666,
      "learning_rate": 3.3497142809641615e-07,
      "loss": 1.4362,
      "step": 2506
    },
    {
      "epoch": 0.550989010989011,
      "grad_norm": 0.25759289383156936,
      "learning_rate": 3.347567057346565e-07,
      "loss": 1.44,
      "step": 2507
    },
    {
      "epoch": 0.5512087912087912,
      "grad_norm": 0.25242840136946804,
      "learning_rate": 3.345420073256281e-07,
      "loss": 1.4454,
      "step": 2508
    },
    {
      "epoch": 0.5514285714285714,
      "grad_norm": 0.2774655042786759,
      "learning_rate": 3.343273329716851e-07,
      "loss": 1.4512,
      "step": 2509
    },
    {
      "epoch": 0.5516483516483517,
      "grad_norm": 0.24697466261756673,
      "learning_rate": 3.341126827751701e-07,
      "loss": 1.5146,
      "step": 2510
    },
    {
      "epoch": 0.5518681318681319,
      "grad_norm": 0.24608028817615893,
      "learning_rate": 3.338980568384149e-07,
      "loss": 1.4418,
      "step": 2511
    },
    {
      "epoch": 0.5520879120879121,
      "grad_norm": 0.24884221696868708,
      "learning_rate": 3.3368345526373874e-07,
      "loss": 1.4368,
      "step": 2512
    },
    {
      "epoch": 0.5523076923076923,
      "grad_norm": 0.2744378881294568,
      "learning_rate": 3.334688781534497e-07,
      "loss": 1.4112,
      "step": 2513
    },
    {
      "epoch": 0.5525274725274725,
      "grad_norm": 0.24812712111009483,
      "learning_rate": 3.332543256098443e-07,
      "loss": 1.4973,
      "step": 2514
    },
    {
      "epoch": 0.5527472527472528,
      "grad_norm": 0.24336979659611635,
      "learning_rate": 3.3303979773520694e-07,
      "loss": 1.403,
      "step": 2515
    },
    {
      "epoch": 0.552967032967033,
      "grad_norm": 0.2492977419816484,
      "learning_rate": 3.32825294631811e-07,
      "loss": 1.5108,
      "step": 2516
    },
    {
      "epoch": 0.5531868131868132,
      "grad_norm": 0.26839021597873497,
      "learning_rate": 3.3261081640191725e-07,
      "loss": 1.5203,
      "step": 2517
    },
    {
      "epoch": 0.5534065934065934,
      "grad_norm": 0.24322649847388625,
      "learning_rate": 3.3239636314777514e-07,
      "loss": 1.4795,
      "step": 2518
    },
    {
      "epoch": 0.5536263736263737,
      "grad_norm": 0.24895096900469302,
      "learning_rate": 3.321819349716218e-07,
      "loss": 1.4335,
      "step": 2519
    },
    {
      "epoch": 0.5538461538461539,
      "grad_norm": 0.2459517510639334,
      "learning_rate": 3.3196753197568287e-07,
      "loss": 1.526,
      "step": 2520
    },
    {
      "epoch": 0.5540659340659341,
      "grad_norm": 0.26015872791263117,
      "learning_rate": 3.317531542621715e-07,
      "loss": 1.4665,
      "step": 2521
    },
    {
      "epoch": 0.5542857142857143,
      "grad_norm": 0.25941471154155876,
      "learning_rate": 3.315388019332893e-07,
      "loss": 1.4784,
      "step": 2522
    },
    {
      "epoch": 0.5545054945054945,
      "grad_norm": 0.24569955453892808,
      "learning_rate": 3.313244750912253e-07,
      "loss": 1.4537,
      "step": 2523
    },
    {
      "epoch": 0.5547252747252748,
      "grad_norm": 0.24186931977538126,
      "learning_rate": 3.311101738381565e-07,
      "loss": 1.5036,
      "step": 2524
    },
    {
      "epoch": 0.554945054945055,
      "grad_norm": 0.2704259669906531,
      "learning_rate": 3.3089589827624796e-07,
      "loss": 1.4056,
      "step": 2525
    },
    {
      "epoch": 0.5551648351648352,
      "grad_norm": 0.2607421854076853,
      "learning_rate": 3.3068164850765246e-07,
      "loss": 1.4222,
      "step": 2526
    },
    {
      "epoch": 0.5553846153846154,
      "grad_norm": 0.2434896276585296,
      "learning_rate": 3.3046742463451004e-07,
      "loss": 1.4041,
      "step": 2527
    },
    {
      "epoch": 0.5556043956043956,
      "grad_norm": 0.26658832252410564,
      "learning_rate": 3.302532267589487e-07,
      "loss": 1.4563,
      "step": 2528
    },
    {
      "epoch": 0.5558241758241759,
      "grad_norm": 0.24760698091118363,
      "learning_rate": 3.3003905498308414e-07,
      "loss": 1.4811,
      "step": 2529
    },
    {
      "epoch": 0.5560439560439561,
      "grad_norm": 0.2448022617942516,
      "learning_rate": 3.298249094090196e-07,
      "loss": 1.4732,
      "step": 2530
    },
    {
      "epoch": 0.5562637362637363,
      "grad_norm": 0.2676645906836815,
      "learning_rate": 3.2961079013884575e-07,
      "loss": 1.5212,
      "step": 2531
    },
    {
      "epoch": 0.5564835164835165,
      "grad_norm": 0.2592275178329446,
      "learning_rate": 3.2939669727464054e-07,
      "loss": 1.4354,
      "step": 2532
    },
    {
      "epoch": 0.5567032967032967,
      "grad_norm": 0.24202438550836206,
      "learning_rate": 3.2918263091846964e-07,
      "loss": 1.4716,
      "step": 2533
    },
    {
      "epoch": 0.556923076923077,
      "grad_norm": 0.25619769090494693,
      "learning_rate": 3.289685911723858e-07,
      "loss": 1.3794,
      "step": 2534
    },
    {
      "epoch": 0.5571428571428572,
      "grad_norm": 0.25163360088271464,
      "learning_rate": 3.2875457813842944e-07,
      "loss": 1.5385,
      "step": 2535
    },
    {
      "epoch": 0.5573626373626374,
      "grad_norm": 0.2474039155565428,
      "learning_rate": 3.2854059191862796e-07,
      "loss": 1.4646,
      "step": 2536
    },
    {
      "epoch": 0.5575824175824176,
      "grad_norm": 0.24790481799111025,
      "learning_rate": 3.2832663261499604e-07,
      "loss": 1.4936,
      "step": 2537
    },
    {
      "epoch": 0.5578021978021978,
      "grad_norm": 0.24333738553491652,
      "learning_rate": 3.2811270032953557e-07,
      "loss": 1.4335,
      "step": 2538
    },
    {
      "epoch": 0.5580219780219781,
      "grad_norm": 0.2605555266761169,
      "learning_rate": 3.2789879516423534e-07,
      "loss": 1.4951,
      "step": 2539
    },
    {
      "epoch": 0.5582417582417583,
      "grad_norm": 0.23841468143353503,
      "learning_rate": 3.276849172210717e-07,
      "loss": 1.4632,
      "step": 2540
    },
    {
      "epoch": 0.5584615384615385,
      "grad_norm": 0.2415403860543191,
      "learning_rate": 3.2747106660200766e-07,
      "loss": 1.4545,
      "step": 2541
    },
    {
      "epoch": 0.5586813186813187,
      "grad_norm": 0.26275919709985424,
      "learning_rate": 3.272572434089931e-07,
      "loss": 1.5114,
      "step": 2542
    },
    {
      "epoch": 0.5589010989010988,
      "grad_norm": 0.23853140946010676,
      "learning_rate": 3.2704344774396524e-07,
      "loss": 1.437,
      "step": 2543
    },
    {
      "epoch": 0.5591208791208792,
      "grad_norm": 0.2441808103072464,
      "learning_rate": 3.2682967970884754e-07,
      "loss": 1.4607,
      "step": 2544
    },
    {
      "epoch": 0.5593406593406594,
      "grad_norm": 0.2523839705339858,
      "learning_rate": 3.266159394055512e-07,
      "loss": 1.5381,
      "step": 2545
    },
    {
      "epoch": 0.5595604395604395,
      "grad_norm": 0.2730961053168548,
      "learning_rate": 3.264022269359734e-07,
      "loss": 1.4453,
      "step": 2546
    },
    {
      "epoch": 0.5597802197802197,
      "grad_norm": 0.24842846462662363,
      "learning_rate": 3.2618854240199823e-07,
      "loss": 1.4806,
      "step": 2547
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2449526251039824,
      "learning_rate": 3.259748859054967e-07,
      "loss": 1.5009,
      "step": 2548
    },
    {
      "epoch": 0.5602197802197803,
      "grad_norm": 0.25274673887224086,
      "learning_rate": 3.2576125754832634e-07,
      "loss": 1.4966,
      "step": 2549
    },
    {
      "epoch": 0.5604395604395604,
      "grad_norm": 0.23865412819854098,
      "learning_rate": 3.2554765743233125e-07,
      "loss": 1.465,
      "step": 2550
    },
    {
      "epoch": 0.5606593406593406,
      "grad_norm": 0.2541274318400704,
      "learning_rate": 3.2533408565934194e-07,
      "loss": 1.5231,
      "step": 2551
    },
    {
      "epoch": 0.5608791208791208,
      "grad_norm": 0.381478326789079,
      "learning_rate": 3.2512054233117564e-07,
      "loss": 1.542,
      "step": 2552
    },
    {
      "epoch": 0.5610989010989011,
      "grad_norm": 0.2503718315460643,
      "learning_rate": 3.2490702754963574e-07,
      "loss": 1.488,
      "step": 2553
    },
    {
      "epoch": 0.5613186813186813,
      "grad_norm": 0.3229378528011373,
      "learning_rate": 3.246935414165125e-07,
      "loss": 1.4384,
      "step": 2554
    },
    {
      "epoch": 0.5615384615384615,
      "grad_norm": 0.2474829544995195,
      "learning_rate": 3.244800840335819e-07,
      "loss": 1.4971,
      "step": 2555
    },
    {
      "epoch": 0.5617582417582417,
      "grad_norm": 2.3272715985664276,
      "learning_rate": 3.242666555026066e-07,
      "loss": 1.4205,
      "step": 2556
    },
    {
      "epoch": 0.5619780219780219,
      "grad_norm": 0.2532428428215496,
      "learning_rate": 3.2405325592533553e-07,
      "loss": 1.4453,
      "step": 2557
    },
    {
      "epoch": 0.5621978021978022,
      "grad_norm": 0.25886253529821535,
      "learning_rate": 3.2383988540350355e-07,
      "loss": 1.5173,
      "step": 2558
    },
    {
      "epoch": 0.5624175824175824,
      "grad_norm": 0.2518593405781085,
      "learning_rate": 3.2362654403883196e-07,
      "loss": 1.4558,
      "step": 2559
    },
    {
      "epoch": 0.5626373626373626,
      "grad_norm": 0.24827338476745162,
      "learning_rate": 3.23413231933028e-07,
      "loss": 1.5319,
      "step": 2560
    },
    {
      "epoch": 0.5628571428571428,
      "grad_norm": 0.3072532487745244,
      "learning_rate": 3.231999491877851e-07,
      "loss": 1.433,
      "step": 2561
    },
    {
      "epoch": 0.563076923076923,
      "grad_norm": 0.2585982268843787,
      "learning_rate": 3.2298669590478237e-07,
      "loss": 1.4268,
      "step": 2562
    },
    {
      "epoch": 0.5632967032967033,
      "grad_norm": 0.2629033786882321,
      "learning_rate": 3.2277347218568505e-07,
      "loss": 1.5091,
      "step": 2563
    },
    {
      "epoch": 0.5635164835164835,
      "grad_norm": 0.24837050939337624,
      "learning_rate": 3.2256027813214466e-07,
      "loss": 1.4253,
      "step": 2564
    },
    {
      "epoch": 0.5637362637362637,
      "grad_norm": 0.2566859245028406,
      "learning_rate": 3.2234711384579806e-07,
      "loss": 1.4301,
      "step": 2565
    },
    {
      "epoch": 0.5639560439560439,
      "grad_norm": 0.24792427739814885,
      "learning_rate": 3.2213397942826803e-07,
      "loss": 1.4601,
      "step": 2566
    },
    {
      "epoch": 0.5641758241758241,
      "grad_norm": 0.2443964933568627,
      "learning_rate": 3.2192087498116327e-07,
      "loss": 1.4502,
      "step": 2567
    },
    {
      "epoch": 0.5643956043956044,
      "grad_norm": 0.24082959777065138,
      "learning_rate": 3.2170780060607796e-07,
      "loss": 1.4533,
      "step": 2568
    },
    {
      "epoch": 0.5646153846153846,
      "grad_norm": 0.24382212241251466,
      "learning_rate": 3.2149475640459243e-07,
      "loss": 1.4929,
      "step": 2569
    },
    {
      "epoch": 0.5648351648351648,
      "grad_norm": 0.25018130599851895,
      "learning_rate": 3.212817424782719e-07,
      "loss": 1.4488,
      "step": 2570
    },
    {
      "epoch": 0.565054945054945,
      "grad_norm": 0.25574900908295944,
      "learning_rate": 3.210687589286677e-07,
      "loss": 1.4997,
      "step": 2571
    },
    {
      "epoch": 0.5652747252747252,
      "grad_norm": 0.27781853993394434,
      "learning_rate": 3.208558058573167e-07,
      "loss": 1.5537,
      "step": 2572
    },
    {
      "epoch": 0.5654945054945055,
      "grad_norm": 0.2601298765458066,
      "learning_rate": 3.206428833657407e-07,
      "loss": 1.4737,
      "step": 2573
    },
    {
      "epoch": 0.5657142857142857,
      "grad_norm": 0.25377617743154335,
      "learning_rate": 3.204299915554474e-07,
      "loss": 1.4547,
      "step": 2574
    },
    {
      "epoch": 0.5659340659340659,
      "grad_norm": 0.25072831180917604,
      "learning_rate": 3.2021713052792997e-07,
      "loss": 1.4487,
      "step": 2575
    },
    {
      "epoch": 0.5661538461538461,
      "grad_norm": 0.2433127043909169,
      "learning_rate": 3.2000430038466655e-07,
      "loss": 1.4171,
      "step": 2576
    },
    {
      "epoch": 0.5663736263736264,
      "grad_norm": 0.25400901437083023,
      "learning_rate": 3.1979150122712053e-07,
      "loss": 1.4724,
      "step": 2577
    },
    {
      "epoch": 0.5665934065934066,
      "grad_norm": 0.2486947090902992,
      "learning_rate": 3.1957873315674087e-07,
      "loss": 1.4275,
      "step": 2578
    },
    {
      "epoch": 0.5668131868131868,
      "grad_norm": 0.24727789962424468,
      "learning_rate": 3.1936599627496157e-07,
      "loss": 1.4607,
      "step": 2579
    },
    {
      "epoch": 0.567032967032967,
      "grad_norm": 0.24360407672577455,
      "learning_rate": 3.1915329068320176e-07,
      "loss": 1.4809,
      "step": 2580
    },
    {
      "epoch": 0.5672527472527472,
      "grad_norm": 0.26404224889084515,
      "learning_rate": 3.189406164828653e-07,
      "loss": 1.4369,
      "step": 2581
    },
    {
      "epoch": 0.5674725274725275,
      "grad_norm": 0.30193832823623906,
      "learning_rate": 3.187279737753416e-07,
      "loss": 1.4478,
      "step": 2582
    },
    {
      "epoch": 0.5676923076923077,
      "grad_norm": 0.254953736010794,
      "learning_rate": 3.18515362662005e-07,
      "loss": 1.4535,
      "step": 2583
    },
    {
      "epoch": 0.5679120879120879,
      "grad_norm": 0.25779652046949264,
      "learning_rate": 3.1830278324421455e-07,
      "loss": 1.446,
      "step": 2584
    },
    {
      "epoch": 0.5681318681318681,
      "grad_norm": 0.2791095963130846,
      "learning_rate": 3.180902356233141e-07,
      "loss": 1.476,
      "step": 2585
    },
    {
      "epoch": 0.5683516483516483,
      "grad_norm": 0.25209541243231776,
      "learning_rate": 3.1787771990063274e-07,
      "loss": 1.4546,
      "step": 2586
    },
    {
      "epoch": 0.5685714285714286,
      "grad_norm": 0.2463060068160242,
      "learning_rate": 3.176652361774839e-07,
      "loss": 1.4493,
      "step": 2587
    },
    {
      "epoch": 0.5687912087912088,
      "grad_norm": 0.2640472167615685,
      "learning_rate": 3.174527845551662e-07,
      "loss": 1.5246,
      "step": 2588
    },
    {
      "epoch": 0.569010989010989,
      "grad_norm": 0.25538103608343604,
      "learning_rate": 3.172403651349627e-07,
      "loss": 1.4742,
      "step": 2589
    },
    {
      "epoch": 0.5692307692307692,
      "grad_norm": 0.246145099871271,
      "learning_rate": 3.170279780181411e-07,
      "loss": 1.4781,
      "step": 2590
    },
    {
      "epoch": 0.5694505494505494,
      "grad_norm": 0.25269496869080404,
      "learning_rate": 3.1681562330595376e-07,
      "loss": 1.4693,
      "step": 2591
    },
    {
      "epoch": 0.5696703296703297,
      "grad_norm": 0.25688121599076813,
      "learning_rate": 3.1660330109963746e-07,
      "loss": 1.4672,
      "step": 2592
    },
    {
      "epoch": 0.5698901098901099,
      "grad_norm": 0.2407271097018519,
      "learning_rate": 3.163910115004138e-07,
      "loss": 1.4054,
      "step": 2593
    },
    {
      "epoch": 0.5701098901098901,
      "grad_norm": 0.2597704242979883,
      "learning_rate": 3.161787546094886e-07,
      "loss": 1.4555,
      "step": 2594
    },
    {
      "epoch": 0.5703296703296703,
      "grad_norm": 0.23975496843985386,
      "learning_rate": 3.1596653052805204e-07,
      "loss": 1.4669,
      "step": 2595
    },
    {
      "epoch": 0.5705494505494505,
      "grad_norm": 0.24621303519748397,
      "learning_rate": 3.1575433935727873e-07,
      "loss": 1.4984,
      "step": 2596
    },
    {
      "epoch": 0.5707692307692308,
      "grad_norm": 0.24101580541246978,
      "learning_rate": 3.155421811983277e-07,
      "loss": 1.4704,
      "step": 2597
    },
    {
      "epoch": 0.570989010989011,
      "grad_norm": 0.25761613526285804,
      "learning_rate": 3.1533005615234205e-07,
      "loss": 1.5003,
      "step": 2598
    },
    {
      "epoch": 0.5712087912087912,
      "grad_norm": 0.23982293041807357,
      "learning_rate": 3.1511796432044935e-07,
      "loss": 1.3696,
      "step": 2599
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.2551912185766934,
      "learning_rate": 3.1490590580376097e-07,
      "loss": 1.5132,
      "step": 2600
    },
    {
      "epoch": 0.5716483516483516,
      "grad_norm": 0.35324427853948154,
      "learning_rate": 3.1469388070337284e-07,
      "loss": 1.4856,
      "step": 2601
    },
    {
      "epoch": 0.5718681318681319,
      "grad_norm": 0.2488082673436781,
      "learning_rate": 3.144818891203644e-07,
      "loss": 1.488,
      "step": 2602
    },
    {
      "epoch": 0.5720879120879121,
      "grad_norm": 0.25294317307726777,
      "learning_rate": 3.1426993115579973e-07,
      "loss": 1.4542,
      "step": 2603
    },
    {
      "epoch": 0.5723076923076923,
      "grad_norm": 0.2572557661095975,
      "learning_rate": 3.1405800691072645e-07,
      "loss": 1.4924,
      "step": 2604
    },
    {
      "epoch": 0.5725274725274725,
      "grad_norm": 0.24894955663998664,
      "learning_rate": 3.1384611648617633e-07,
      "loss": 1.5289,
      "step": 2605
    },
    {
      "epoch": 0.5727472527472528,
      "grad_norm": 0.2478132076775968,
      "learning_rate": 3.1363425998316467e-07,
      "loss": 1.483,
      "step": 2606
    },
    {
      "epoch": 0.572967032967033,
      "grad_norm": 0.24078033897355672,
      "learning_rate": 3.1342243750269125e-07,
      "loss": 1.4472,
      "step": 2607
    },
    {
      "epoch": 0.5731868131868132,
      "grad_norm": 0.2475037589947621,
      "learning_rate": 3.13210649145739e-07,
      "loss": 1.4196,
      "step": 2608
    },
    {
      "epoch": 0.5734065934065934,
      "grad_norm": 0.23372037850745242,
      "learning_rate": 3.1299889501327483e-07,
      "loss": 1.4447,
      "step": 2609
    },
    {
      "epoch": 0.5736263736263736,
      "grad_norm": 0.2585502894539701,
      "learning_rate": 3.127871752062494e-07,
      "loss": 1.4731,
      "step": 2610
    },
    {
      "epoch": 0.5738461538461539,
      "grad_norm": 0.25331621953124167,
      "learning_rate": 3.125754898255967e-07,
      "loss": 1.5192,
      "step": 2611
    },
    {
      "epoch": 0.5740659340659341,
      "grad_norm": 0.2832784069839386,
      "learning_rate": 3.1236383897223487e-07,
      "loss": 1.4283,
      "step": 2612
    },
    {
      "epoch": 0.5742857142857143,
      "grad_norm": 0.2562379215472176,
      "learning_rate": 3.1215222274706507e-07,
      "loss": 1.5193,
      "step": 2613
    },
    {
      "epoch": 0.5745054945054945,
      "grad_norm": 0.2565464755119887,
      "learning_rate": 3.119406412509722e-07,
      "loss": 1.4648,
      "step": 2614
    },
    {
      "epoch": 0.5747252747252747,
      "grad_norm": 0.24239285072109812,
      "learning_rate": 3.1172909458482447e-07,
      "loss": 1.3815,
      "step": 2615
    },
    {
      "epoch": 0.574945054945055,
      "grad_norm": 0.24237000060226999,
      "learning_rate": 3.1151758284947356e-07,
      "loss": 1.4599,
      "step": 2616
    },
    {
      "epoch": 0.5751648351648352,
      "grad_norm": 0.24201060162258123,
      "learning_rate": 3.1130610614575453e-07,
      "loss": 1.4926,
      "step": 2617
    },
    {
      "epoch": 0.5753846153846154,
      "grad_norm": 0.24110541902147878,
      "learning_rate": 3.1109466457448574e-07,
      "loss": 1.4269,
      "step": 2618
    },
    {
      "epoch": 0.5756043956043956,
      "grad_norm": 0.5151905357191532,
      "learning_rate": 3.108832582364687e-07,
      "loss": 1.4174,
      "step": 2619
    },
    {
      "epoch": 0.5758241758241758,
      "grad_norm": 0.264400423492667,
      "learning_rate": 3.1067188723248823e-07,
      "loss": 1.5061,
      "step": 2620
    },
    {
      "epoch": 0.5760439560439561,
      "grad_norm": 0.2533826514388564,
      "learning_rate": 3.1046055166331217e-07,
      "loss": 1.4184,
      "step": 2621
    },
    {
      "epoch": 0.5762637362637363,
      "grad_norm": 0.2762190273399301,
      "learning_rate": 3.102492516296917e-07,
      "loss": 1.4942,
      "step": 2622
    },
    {
      "epoch": 0.5764835164835165,
      "grad_norm": 0.24797308297241358,
      "learning_rate": 3.100379872323609e-07,
      "loss": 1.4833,
      "step": 2623
    },
    {
      "epoch": 0.5767032967032967,
      "grad_norm": 0.2447992141664247,
      "learning_rate": 3.098267585720367e-07,
      "loss": 1.4154,
      "step": 2624
    },
    {
      "epoch": 0.5769230769230769,
      "grad_norm": 0.25456065851251086,
      "learning_rate": 3.096155657494194e-07,
      "loss": 1.5042,
      "step": 2625
    },
    {
      "epoch": 0.5771428571428572,
      "grad_norm": 0.24192886052994556,
      "learning_rate": 3.094044088651917e-07,
      "loss": 1.4607,
      "step": 2626
    },
    {
      "epoch": 0.5773626373626374,
      "grad_norm": 0.2680401406754433,
      "learning_rate": 3.091932880200199e-07,
      "loss": 1.4202,
      "step": 2627
    },
    {
      "epoch": 0.5775824175824176,
      "grad_norm": 0.3005987432819412,
      "learning_rate": 3.089822033145522e-07,
      "loss": 1.428,
      "step": 2628
    },
    {
      "epoch": 0.5778021978021978,
      "grad_norm": 0.26120398677275525,
      "learning_rate": 3.087711548494204e-07,
      "loss": 1.4763,
      "step": 2629
    },
    {
      "epoch": 0.578021978021978,
      "grad_norm": 0.32458714972291736,
      "learning_rate": 3.085601427252384e-07,
      "loss": 1.4499,
      "step": 2630
    },
    {
      "epoch": 0.5782417582417583,
      "grad_norm": 0.27004227251398977,
      "learning_rate": 3.0834916704260314e-07,
      "loss": 1.4768,
      "step": 2631
    },
    {
      "epoch": 0.5784615384615385,
      "grad_norm": 0.243766314754373,
      "learning_rate": 3.081382279020942e-07,
      "loss": 1.4529,
      "step": 2632
    },
    {
      "epoch": 0.5786813186813187,
      "grad_norm": 0.24606129828590848,
      "learning_rate": 3.079273254042736e-07,
      "loss": 1.4939,
      "step": 2633
    },
    {
      "epoch": 0.5789010989010989,
      "grad_norm": 0.2500961939135463,
      "learning_rate": 3.077164596496857e-07,
      "loss": 1.3928,
      "step": 2634
    },
    {
      "epoch": 0.5791208791208792,
      "grad_norm": 0.24275683762503814,
      "learning_rate": 3.0750563073885774e-07,
      "loss": 1.4908,
      "step": 2635
    },
    {
      "epoch": 0.5793406593406594,
      "grad_norm": 0.255648446292445,
      "learning_rate": 3.0729483877229925e-07,
      "loss": 1.4695,
      "step": 2636
    },
    {
      "epoch": 0.5795604395604396,
      "grad_norm": 0.2505917464893049,
      "learning_rate": 3.070840838505021e-07,
      "loss": 1.4361,
      "step": 2637
    },
    {
      "epoch": 0.5797802197802198,
      "grad_norm": 0.4164525347416848,
      "learning_rate": 3.068733660739404e-07,
      "loss": 1.4378,
      "step": 2638
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.25534232957880826,
      "learning_rate": 3.066626855430708e-07,
      "loss": 1.4871,
      "step": 2639
    },
    {
      "epoch": 0.5802197802197803,
      "grad_norm": 0.2625756004252582,
      "learning_rate": 3.064520423583319e-07,
      "loss": 1.4771,
      "step": 2640
    },
    {
      "epoch": 0.5804395604395605,
      "grad_norm": 0.24767947681665137,
      "learning_rate": 3.062414366201448e-07,
      "loss": 1.435,
      "step": 2641
    },
    {
      "epoch": 0.5806593406593407,
      "grad_norm": 0.24937095071964677,
      "learning_rate": 3.060308684289126e-07,
      "loss": 1.4655,
      "step": 2642
    },
    {
      "epoch": 0.5808791208791209,
      "grad_norm": 0.2349785411716748,
      "learning_rate": 3.0582033788502044e-07,
      "loss": 1.4373,
      "step": 2643
    },
    {
      "epoch": 0.581098901098901,
      "grad_norm": 0.24143193424826362,
      "learning_rate": 3.0560984508883556e-07,
      "loss": 1.4567,
      "step": 2644
    },
    {
      "epoch": 0.5813186813186814,
      "grad_norm": 0.2403897129083357,
      "learning_rate": 3.0539939014070724e-07,
      "loss": 1.4898,
      "step": 2645
    },
    {
      "epoch": 0.5815384615384616,
      "grad_norm": 0.2506923828615349,
      "learning_rate": 3.051889731409667e-07,
      "loss": 1.438,
      "step": 2646
    },
    {
      "epoch": 0.5817582417582418,
      "grad_norm": 0.25845481960208516,
      "learning_rate": 3.049785941899271e-07,
      "loss": 1.4776,
      "step": 2647
    },
    {
      "epoch": 0.581978021978022,
      "grad_norm": 0.29332390853994833,
      "learning_rate": 3.0476825338788345e-07,
      "loss": 1.4469,
      "step": 2648
    },
    {
      "epoch": 0.5821978021978022,
      "grad_norm": 0.3250187148849675,
      "learning_rate": 3.0455795083511254e-07,
      "loss": 1.4956,
      "step": 2649
    },
    {
      "epoch": 0.5824175824175825,
      "grad_norm": 0.24149785917052716,
      "learning_rate": 3.043476866318727e-07,
      "loss": 1.502,
      "step": 2650
    },
    {
      "epoch": 0.5826373626373627,
      "grad_norm": 0.26100227997137226,
      "learning_rate": 3.0413746087840447e-07,
      "loss": 1.5276,
      "step": 2651
    },
    {
      "epoch": 0.5828571428571429,
      "grad_norm": 0.24741217343462235,
      "learning_rate": 3.0392727367492983e-07,
      "loss": 1.4491,
      "step": 2652
    },
    {
      "epoch": 0.583076923076923,
      "grad_norm": 0.2365057831730838,
      "learning_rate": 3.037171251216522e-07,
      "loss": 1.4954,
      "step": 2653
    },
    {
      "epoch": 0.5832967032967032,
      "grad_norm": 0.27826573782475317,
      "learning_rate": 3.0350701531875673e-07,
      "loss": 1.4737,
      "step": 2654
    },
    {
      "epoch": 0.5835164835164836,
      "grad_norm": 0.24485011334493212,
      "learning_rate": 3.032969443664101e-07,
      "loss": 1.5342,
      "step": 2655
    },
    {
      "epoch": 0.5837362637362637,
      "grad_norm": 0.2520291467226276,
      "learning_rate": 3.030869123647606e-07,
      "loss": 1.5136,
      "step": 2656
    },
    {
      "epoch": 0.583956043956044,
      "grad_norm": 0.24737609327471743,
      "learning_rate": 3.028769194139378e-07,
      "loss": 1.5458,
      "step": 2657
    },
    {
      "epoch": 0.5841758241758241,
      "grad_norm": 0.27859430322534867,
      "learning_rate": 3.0266696561405246e-07,
      "loss": 1.4492,
      "step": 2658
    },
    {
      "epoch": 0.5843956043956045,
      "grad_norm": 0.24858071059972395,
      "learning_rate": 3.0245705106519705e-07,
      "loss": 1.4589,
      "step": 2659
    },
    {
      "epoch": 0.5846153846153846,
      "grad_norm": 0.251028121597539,
      "learning_rate": 3.022471758674451e-07,
      "loss": 1.4367,
      "step": 2660
    },
    {
      "epoch": 0.5848351648351648,
      "grad_norm": 0.23276113173526022,
      "learning_rate": 3.0203734012085155e-07,
      "loss": 1.4627,
      "step": 2661
    },
    {
      "epoch": 0.585054945054945,
      "grad_norm": 0.2978633736080548,
      "learning_rate": 3.018275439254522e-07,
      "loss": 1.4127,
      "step": 2662
    },
    {
      "epoch": 0.5852747252747252,
      "grad_norm": 0.2410866841392644,
      "learning_rate": 3.0161778738126445e-07,
      "loss": 1.3678,
      "step": 2663
    },
    {
      "epoch": 0.5854945054945055,
      "grad_norm": 0.24165254504816364,
      "learning_rate": 3.0140807058828634e-07,
      "loss": 1.476,
      "step": 2664
    },
    {
      "epoch": 0.5857142857142857,
      "grad_norm": 0.2624777663014652,
      "learning_rate": 3.011983936464974e-07,
      "loss": 1.4766,
      "step": 2665
    },
    {
      "epoch": 0.5859340659340659,
      "grad_norm": 0.2634247230469347,
      "learning_rate": 3.009887566558577e-07,
      "loss": 1.4598,
      "step": 2666
    },
    {
      "epoch": 0.5861538461538461,
      "grad_norm": 0.25875444607367315,
      "learning_rate": 3.007791597163088e-07,
      "loss": 1.509,
      "step": 2667
    },
    {
      "epoch": 0.5863736263736263,
      "grad_norm": 0.26630387524065624,
      "learning_rate": 3.0056960292777275e-07,
      "loss": 1.461,
      "step": 2668
    },
    {
      "epoch": 0.5865934065934066,
      "grad_norm": 0.24078612966483726,
      "learning_rate": 3.003600863901524e-07,
      "loss": 1.4968,
      "step": 2669
    },
    {
      "epoch": 0.5868131868131868,
      "grad_norm": 0.24566965714964653,
      "learning_rate": 3.0015061020333173e-07,
      "loss": 1.4463,
      "step": 2670
    },
    {
      "epoch": 0.587032967032967,
      "grad_norm": 0.26238688810875044,
      "learning_rate": 2.9994117446717546e-07,
      "loss": 1.4323,
      "step": 2671
    },
    {
      "epoch": 0.5872527472527472,
      "grad_norm": 0.26934926974738016,
      "learning_rate": 2.9973177928152886e-07,
      "loss": 1.4642,
      "step": 2672
    },
    {
      "epoch": 0.5874725274725274,
      "grad_norm": 0.252353065643034,
      "learning_rate": 2.9952242474621774e-07,
      "loss": 1.433,
      "step": 2673
    },
    {
      "epoch": 0.5876923076923077,
      "grad_norm": 0.2526591795699064,
      "learning_rate": 2.993131109610488e-07,
      "loss": 1.4873,
      "step": 2674
    },
    {
      "epoch": 0.5879120879120879,
      "grad_norm": 0.24745614937116045,
      "learning_rate": 2.991038380258095e-07,
      "loss": 1.4721,
      "step": 2675
    },
    {
      "epoch": 0.5881318681318681,
      "grad_norm": 0.25320330697301335,
      "learning_rate": 2.9889460604026724e-07,
      "loss": 1.503,
      "step": 2676
    },
    {
      "epoch": 0.5883516483516483,
      "grad_norm": 0.25404437739800473,
      "learning_rate": 2.9868541510417015e-07,
      "loss": 1.5045,
      "step": 2677
    },
    {
      "epoch": 0.5885714285714285,
      "grad_norm": 0.24604682796478008,
      "learning_rate": 2.9847626531724715e-07,
      "loss": 1.478,
      "step": 2678
    },
    {
      "epoch": 0.5887912087912088,
      "grad_norm": 0.24484594462571194,
      "learning_rate": 2.9826715677920693e-07,
      "loss": 1.433,
      "step": 2679
    },
    {
      "epoch": 0.589010989010989,
      "grad_norm": 0.24938898195160425,
      "learning_rate": 2.980580895897391e-07,
      "loss": 1.4724,
      "step": 2680
    },
    {
      "epoch": 0.5892307692307692,
      "grad_norm": 0.24404664218132727,
      "learning_rate": 2.97849063848513e-07,
      "loss": 1.4203,
      "step": 2681
    },
    {
      "epoch": 0.5894505494505494,
      "grad_norm": 0.25727167626009134,
      "learning_rate": 2.9764007965517875e-07,
      "loss": 1.5806,
      "step": 2682
    },
    {
      "epoch": 0.5896703296703296,
      "grad_norm": 0.25741622544742127,
      "learning_rate": 2.974311371093664e-07,
      "loss": 1.4805,
      "step": 2683
    },
    {
      "epoch": 0.5898901098901099,
      "grad_norm": 0.25109130456679774,
      "learning_rate": 2.972222363106858e-07,
      "loss": 1.4668,
      "step": 2684
    },
    {
      "epoch": 0.5901098901098901,
      "grad_norm": 0.25413762077941837,
      "learning_rate": 2.970133773587276e-07,
      "loss": 1.3821,
      "step": 2685
    },
    {
      "epoch": 0.5903296703296703,
      "grad_norm": 0.2481136357766227,
      "learning_rate": 2.968045603530621e-07,
      "loss": 1.4982,
      "step": 2686
    },
    {
      "epoch": 0.5905494505494505,
      "grad_norm": 0.26060840911471206,
      "learning_rate": 2.9659578539323966e-07,
      "loss": 1.536,
      "step": 2687
    },
    {
      "epoch": 0.5907692307692308,
      "grad_norm": 0.24894375397809232,
      "learning_rate": 2.963870525787903e-07,
      "loss": 1.4488,
      "step": 2688
    },
    {
      "epoch": 0.590989010989011,
      "grad_norm": 0.24549929315416463,
      "learning_rate": 2.961783620092246e-07,
      "loss": 1.4934,
      "step": 2689
    },
    {
      "epoch": 0.5912087912087912,
      "grad_norm": 0.24841004174419143,
      "learning_rate": 2.959697137840326e-07,
      "loss": 1.4358,
      "step": 2690
    },
    {
      "epoch": 0.5914285714285714,
      "grad_norm": 0.25265455873382414,
      "learning_rate": 2.9576110800268403e-07,
      "loss": 1.4453,
      "step": 2691
    },
    {
      "epoch": 0.5916483516483516,
      "grad_norm": 0.25304837017926046,
      "learning_rate": 2.9555254476462854e-07,
      "loss": 1.4942,
      "step": 2692
    },
    {
      "epoch": 0.5918681318681319,
      "grad_norm": 0.2564751955741434,
      "learning_rate": 2.953440241692954e-07,
      "loss": 1.5028,
      "step": 2693
    },
    {
      "epoch": 0.5920879120879121,
      "grad_norm": 0.4637593906261774,
      "learning_rate": 2.9513554631609403e-07,
      "loss": 1.449,
      "step": 2694
    },
    {
      "epoch": 0.5923076923076923,
      "grad_norm": 0.26801487565820803,
      "learning_rate": 2.9492711130441296e-07,
      "loss": 1.5407,
      "step": 2695
    },
    {
      "epoch": 0.5925274725274725,
      "grad_norm": 0.25455198449595984,
      "learning_rate": 2.947187192336202e-07,
      "loss": 1.4823,
      "step": 2696
    },
    {
      "epoch": 0.5927472527472527,
      "grad_norm": 0.3107751988612622,
      "learning_rate": 2.945103702030637e-07,
      "loss": 1.4964,
      "step": 2697
    },
    {
      "epoch": 0.592967032967033,
      "grad_norm": 0.2472063244291013,
      "learning_rate": 2.9430206431207066e-07,
      "loss": 1.5277,
      "step": 2698
    },
    {
      "epoch": 0.5931868131868132,
      "grad_norm": 0.24236934943271227,
      "learning_rate": 2.9409380165994786e-07,
      "loss": 1.4392,
      "step": 2699
    },
    {
      "epoch": 0.5934065934065934,
      "grad_norm": 0.25532961404856547,
      "learning_rate": 2.938855823459812e-07,
      "loss": 1.4332,
      "step": 2700
    },
    {
      "epoch": 0.5936263736263736,
      "grad_norm": 0.24241481637153153,
      "learning_rate": 2.936774064694363e-07,
      "loss": 1.4496,
      "step": 2701
    },
    {
      "epoch": 0.5938461538461538,
      "grad_norm": 0.24352597552449362,
      "learning_rate": 2.934692741295577e-07,
      "loss": 1.5056,
      "step": 2702
    },
    {
      "epoch": 0.5940659340659341,
      "grad_norm": 0.3906887755320128,
      "learning_rate": 2.9326118542556934e-07,
      "loss": 1.4237,
      "step": 2703
    },
    {
      "epoch": 0.5942857142857143,
      "grad_norm": 0.2394689170130002,
      "learning_rate": 2.9305314045667453e-07,
      "loss": 1.4957,
      "step": 2704
    },
    {
      "epoch": 0.5945054945054945,
      "grad_norm": 0.2568718639849738,
      "learning_rate": 2.9284513932205545e-07,
      "loss": 1.4349,
      "step": 2705
    },
    {
      "epoch": 0.5947252747252747,
      "grad_norm": 0.27977339720146194,
      "learning_rate": 2.926371821208735e-07,
      "loss": 1.449,
      "step": 2706
    },
    {
      "epoch": 0.5949450549450549,
      "grad_norm": 0.2544818020840947,
      "learning_rate": 2.924292689522691e-07,
      "loss": 1.4792,
      "step": 2707
    },
    {
      "epoch": 0.5951648351648352,
      "grad_norm": 0.24755744130994373,
      "learning_rate": 2.9222139991536177e-07,
      "loss": 1.5612,
      "step": 2708
    },
    {
      "epoch": 0.5953846153846154,
      "grad_norm": 0.2578695246691789,
      "learning_rate": 2.920135751092499e-07,
      "loss": 1.5198,
      "step": 2709
    },
    {
      "epoch": 0.5956043956043956,
      "grad_norm": 0.2872310702737208,
      "learning_rate": 2.918057946330109e-07,
      "loss": 1.4424,
      "step": 2710
    },
    {
      "epoch": 0.5958241758241758,
      "grad_norm": 0.2566019432078165,
      "learning_rate": 2.9159805858570093e-07,
      "loss": 1.4996,
      "step": 2711
    },
    {
      "epoch": 0.596043956043956,
      "grad_norm": 0.3174090061953062,
      "learning_rate": 2.9139036706635494e-07,
      "loss": 1.4493,
      "step": 2712
    },
    {
      "epoch": 0.5962637362637363,
      "grad_norm": 0.4448257107050925,
      "learning_rate": 2.911827201739867e-07,
      "loss": 1.4859,
      "step": 2713
    },
    {
      "epoch": 0.5964835164835165,
      "grad_norm": 0.25015361228476474,
      "learning_rate": 2.9097511800758894e-07,
      "loss": 1.4481,
      "step": 2714
    },
    {
      "epoch": 0.5967032967032967,
      "grad_norm": 0.3752665868215014,
      "learning_rate": 2.907675606661327e-07,
      "loss": 1.4822,
      "step": 2715
    },
    {
      "epoch": 0.5969230769230769,
      "grad_norm": 0.24421542260831594,
      "learning_rate": 2.9056004824856784e-07,
      "loss": 1.4789,
      "step": 2716
    },
    {
      "epoch": 0.5971428571428572,
      "grad_norm": 0.23402351613022046,
      "learning_rate": 2.903525808538226e-07,
      "loss": 1.4615,
      "step": 2717
    },
    {
      "epoch": 0.5973626373626374,
      "grad_norm": 0.2551073910167814,
      "learning_rate": 2.9014515858080424e-07,
      "loss": 1.4488,
      "step": 2718
    },
    {
      "epoch": 0.5975824175824176,
      "grad_norm": 0.2924320056245985,
      "learning_rate": 2.8993778152839806e-07,
      "loss": 1.425,
      "step": 2719
    },
    {
      "epoch": 0.5978021978021978,
      "grad_norm": 0.25662166205007986,
      "learning_rate": 2.8973044979546783e-07,
      "loss": 1.501,
      "step": 2720
    },
    {
      "epoch": 0.598021978021978,
      "grad_norm": 0.2562770092861731,
      "learning_rate": 2.89523163480856e-07,
      "loss": 1.4995,
      "step": 2721
    },
    {
      "epoch": 0.5982417582417583,
      "grad_norm": 0.2426367136724267,
      "learning_rate": 2.893159226833829e-07,
      "loss": 1.4782,
      "step": 2722
    },
    {
      "epoch": 0.5984615384615385,
      "grad_norm": 0.24777685332988814,
      "learning_rate": 2.8910872750184777e-07,
      "loss": 1.4702,
      "step": 2723
    },
    {
      "epoch": 0.5986813186813187,
      "grad_norm": 0.240135100898524,
      "learning_rate": 2.8890157803502757e-07,
      "loss": 1.4864,
      "step": 2724
    },
    {
      "epoch": 0.5989010989010989,
      "grad_norm": 0.29277346818395716,
      "learning_rate": 2.886944743816779e-07,
      "loss": 1.4546,
      "step": 2725
    },
    {
      "epoch": 0.5991208791208791,
      "grad_norm": 0.257649706301125,
      "learning_rate": 2.88487416640532e-07,
      "loss": 1.539,
      "step": 2726
    },
    {
      "epoch": 0.5993406593406594,
      "grad_norm": 0.2619500410076014,
      "learning_rate": 2.882804049103017e-07,
      "loss": 1.5068,
      "step": 2727
    },
    {
      "epoch": 0.5995604395604396,
      "grad_norm": 0.2528374253627584,
      "learning_rate": 2.880734392896768e-07,
      "loss": 1.4843,
      "step": 2728
    },
    {
      "epoch": 0.5997802197802198,
      "grad_norm": 0.5273316811041006,
      "learning_rate": 2.878665198773248e-07,
      "loss": 1.4479,
      "step": 2729
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2627518876587247,
      "learning_rate": 2.876596467718916e-07,
      "loss": 1.5063,
      "step": 2730
    },
    {
      "epoch": 0.6002197802197802,
      "grad_norm": 0.25536190043986196,
      "learning_rate": 2.874528200720007e-07,
      "loss": 1.5356,
      "step": 2731
    },
    {
      "epoch": 0.6004395604395605,
      "grad_norm": 0.23583426044581032,
      "learning_rate": 2.872460398762536e-07,
      "loss": 1.4199,
      "step": 2732
    },
    {
      "epoch": 0.6006593406593407,
      "grad_norm": 0.25871441334810275,
      "learning_rate": 2.870393062832298e-07,
      "loss": 1.5,
      "step": 2733
    },
    {
      "epoch": 0.6008791208791209,
      "grad_norm": 0.2536571432183577,
      "learning_rate": 2.868326193914862e-07,
      "loss": 1.4921,
      "step": 2734
    },
    {
      "epoch": 0.6010989010989011,
      "grad_norm": 0.25024205420416806,
      "learning_rate": 2.8662597929955785e-07,
      "loss": 1.3915,
      "step": 2735
    },
    {
      "epoch": 0.6013186813186813,
      "grad_norm": 0.24610958354668805,
      "learning_rate": 2.864193861059572e-07,
      "loss": 1.4565,
      "step": 2736
    },
    {
      "epoch": 0.6015384615384616,
      "grad_norm": 0.2503297851221268,
      "learning_rate": 2.8621283990917424e-07,
      "loss": 1.4075,
      "step": 2737
    },
    {
      "epoch": 0.6017582417582418,
      "grad_norm": 0.2538166839525025,
      "learning_rate": 2.8600634080767706e-07,
      "loss": 1.4641,
      "step": 2738
    },
    {
      "epoch": 0.601978021978022,
      "grad_norm": 0.25979470001679406,
      "learning_rate": 2.85799888899911e-07,
      "loss": 1.535,
      "step": 2739
    },
    {
      "epoch": 0.6021978021978022,
      "grad_norm": 0.24576730845308514,
      "learning_rate": 2.8559348428429866e-07,
      "loss": 1.4609,
      "step": 2740
    },
    {
      "epoch": 0.6024175824175824,
      "grad_norm": 0.2534655548238086,
      "learning_rate": 2.853871270592403e-07,
      "loss": 1.4879,
      "step": 2741
    },
    {
      "epoch": 0.6026373626373627,
      "grad_norm": 0.2789829200576066,
      "learning_rate": 2.8518081732311365e-07,
      "loss": 1.4383,
      "step": 2742
    },
    {
      "epoch": 0.6028571428571429,
      "grad_norm": 0.24500911334681083,
      "learning_rate": 2.8497455517427406e-07,
      "loss": 1.4767,
      "step": 2743
    },
    {
      "epoch": 0.6030769230769231,
      "grad_norm": 0.24063586795282615,
      "learning_rate": 2.847683407110536e-07,
      "loss": 1.409,
      "step": 2744
    },
    {
      "epoch": 0.6032967032967033,
      "grad_norm": 0.253914208061666,
      "learning_rate": 2.8456217403176183e-07,
      "loss": 1.5053,
      "step": 2745
    },
    {
      "epoch": 0.6035164835164836,
      "grad_norm": 0.24074409688538426,
      "learning_rate": 2.8435605523468575e-07,
      "loss": 1.4828,
      "step": 2746
    },
    {
      "epoch": 0.6037362637362638,
      "grad_norm": 0.2553825030734294,
      "learning_rate": 2.8414998441808944e-07,
      "loss": 1.5263,
      "step": 2747
    },
    {
      "epoch": 0.603956043956044,
      "grad_norm": 0.2382423781962556,
      "learning_rate": 2.8394396168021405e-07,
      "loss": 1.4564,
      "step": 2748
    },
    {
      "epoch": 0.6041758241758242,
      "grad_norm": 0.24876751520215948,
      "learning_rate": 2.8373798711927757e-07,
      "loss": 1.4817,
      "step": 2749
    },
    {
      "epoch": 0.6043956043956044,
      "grad_norm": 0.2516003827718687,
      "learning_rate": 2.835320608334755e-07,
      "loss": 1.4137,
      "step": 2750
    },
    {
      "epoch": 0.6046153846153847,
      "grad_norm": 0.2749483301934398,
      "learning_rate": 2.8332618292097993e-07,
      "loss": 1.4397,
      "step": 2751
    },
    {
      "epoch": 0.6048351648351649,
      "grad_norm": 0.286111946427972,
      "learning_rate": 2.831203534799401e-07,
      "loss": 1.4399,
      "step": 2752
    },
    {
      "epoch": 0.6050549450549451,
      "grad_norm": 0.29218144959981274,
      "learning_rate": 2.829145726084821e-07,
      "loss": 1.4436,
      "step": 2753
    },
    {
      "epoch": 0.6052747252747253,
      "grad_norm": 0.2525068779790069,
      "learning_rate": 2.827088404047089e-07,
      "loss": 1.4637,
      "step": 2754
    },
    {
      "epoch": 0.6054945054945055,
      "grad_norm": 0.2543220711372504,
      "learning_rate": 2.825031569667003e-07,
      "loss": 1.4778,
      "step": 2755
    },
    {
      "epoch": 0.6057142857142858,
      "grad_norm": 0.24789127883030665,
      "learning_rate": 2.8229752239251236e-07,
      "loss": 1.462,
      "step": 2756
    },
    {
      "epoch": 0.605934065934066,
      "grad_norm": 0.25967686683388536,
      "learning_rate": 2.820919367801786e-07,
      "loss": 1.5454,
      "step": 2757
    },
    {
      "epoch": 0.6061538461538462,
      "grad_norm": 0.2604169620992817,
      "learning_rate": 2.8188640022770883e-07,
      "loss": 1.5082,
      "step": 2758
    },
    {
      "epoch": 0.6063736263736264,
      "grad_norm": 0.26860630268928987,
      "learning_rate": 2.8168091283308945e-07,
      "loss": 1.5118,
      "step": 2759
    },
    {
      "epoch": 0.6065934065934065,
      "grad_norm": 0.39784011679850734,
      "learning_rate": 2.814754746942835e-07,
      "loss": 1.4461,
      "step": 2760
    },
    {
      "epoch": 0.6068131868131869,
      "grad_norm": 0.24390577596970886,
      "learning_rate": 2.8127008590923033e-07,
      "loss": 1.4586,
      "step": 2761
    },
    {
      "epoch": 0.607032967032967,
      "grad_norm": 0.23898812807040618,
      "learning_rate": 2.810647465758463e-07,
      "loss": 1.4653,
      "step": 2762
    },
    {
      "epoch": 0.6072527472527472,
      "grad_norm": 0.2630961492357384,
      "learning_rate": 2.808594567920236e-07,
      "loss": 1.483,
      "step": 2763
    },
    {
      "epoch": 0.6074725274725274,
      "grad_norm": 0.2549766101813333,
      "learning_rate": 2.806542166556309e-07,
      "loss": 1.4526,
      "step": 2764
    },
    {
      "epoch": 0.6076923076923076,
      "grad_norm": 0.2492678697325956,
      "learning_rate": 2.8044902626451363e-07,
      "loss": 1.5081,
      "step": 2765
    },
    {
      "epoch": 0.607912087912088,
      "grad_norm": 0.5191154837257347,
      "learning_rate": 2.802438857164928e-07,
      "loss": 1.4791,
      "step": 2766
    },
    {
      "epoch": 0.6081318681318681,
      "grad_norm": 0.24638747553873072,
      "learning_rate": 2.8003879510936665e-07,
      "loss": 1.5079,
      "step": 2767
    },
    {
      "epoch": 0.6083516483516483,
      "grad_norm": 0.25502158688573806,
      "learning_rate": 2.7983375454090857e-07,
      "loss": 1.4368,
      "step": 2768
    },
    {
      "epoch": 0.6085714285714285,
      "grad_norm": 0.24980133419717593,
      "learning_rate": 2.796287641088687e-07,
      "loss": 1.5135,
      "step": 2769
    },
    {
      "epoch": 0.6087912087912087,
      "grad_norm": 0.242376182426195,
      "learning_rate": 2.7942382391097324e-07,
      "loss": 1.4315,
      "step": 2770
    },
    {
      "epoch": 0.609010989010989,
      "grad_norm": 0.2580739428886733,
      "learning_rate": 2.7921893404492396e-07,
      "loss": 1.4545,
      "step": 2771
    },
    {
      "epoch": 0.6092307692307692,
      "grad_norm": 0.26204011355952916,
      "learning_rate": 2.790140946083994e-07,
      "loss": 1.4674,
      "step": 2772
    },
    {
      "epoch": 0.6094505494505494,
      "grad_norm": 0.25198584266570817,
      "learning_rate": 2.7880930569905366e-07,
      "loss": 1.4641,
      "step": 2773
    },
    {
      "epoch": 0.6096703296703296,
      "grad_norm": 0.28798028115276636,
      "learning_rate": 2.7860456741451645e-07,
      "loss": 1.5141,
      "step": 2774
    },
    {
      "epoch": 0.6098901098901099,
      "grad_norm": 0.24340120570712515,
      "learning_rate": 2.7839987985239386e-07,
      "loss": 1.4365,
      "step": 2775
    },
    {
      "epoch": 0.6101098901098901,
      "grad_norm": 0.2570269674608408,
      "learning_rate": 2.781952431102675e-07,
      "loss": 1.4716,
      "step": 2776
    },
    {
      "epoch": 0.6103296703296703,
      "grad_norm": 0.2387418227608774,
      "learning_rate": 2.7799065728569497e-07,
      "loss": 1.4219,
      "step": 2777
    },
    {
      "epoch": 0.6105494505494505,
      "grad_norm": 0.26350495969877236,
      "learning_rate": 2.777861224762095e-07,
      "loss": 1.4539,
      "step": 2778
    },
    {
      "epoch": 0.6107692307692307,
      "grad_norm": 0.24599371937262848,
      "learning_rate": 2.7758163877931974e-07,
      "loss": 1.4554,
      "step": 2779
    },
    {
      "epoch": 0.610989010989011,
      "grad_norm": 0.2504081196618615,
      "learning_rate": 2.7737720629251027e-07,
      "loss": 1.4543,
      "step": 2780
    },
    {
      "epoch": 0.6112087912087912,
      "grad_norm": 0.2543328826681762,
      "learning_rate": 2.771728251132414e-07,
      "loss": 1.4969,
      "step": 2781
    },
    {
      "epoch": 0.6114285714285714,
      "grad_norm": 0.2466344005097448,
      "learning_rate": 2.769684953389486e-07,
      "loss": 1.4792,
      "step": 2782
    },
    {
      "epoch": 0.6116483516483516,
      "grad_norm": 0.2441171202539346,
      "learning_rate": 2.76764217067043e-07,
      "loss": 1.4599,
      "step": 2783
    },
    {
      "epoch": 0.6118681318681318,
      "grad_norm": 0.25282448539558716,
      "learning_rate": 2.765599903949112e-07,
      "loss": 1.4729,
      "step": 2784
    },
    {
      "epoch": 0.6120879120879121,
      "grad_norm": 0.23823618339636068,
      "learning_rate": 2.763558154199151e-07,
      "loss": 1.4864,
      "step": 2785
    },
    {
      "epoch": 0.6123076923076923,
      "grad_norm": 0.2392802927769334,
      "learning_rate": 2.761516922393922e-07,
      "loss": 1.432,
      "step": 2786
    },
    {
      "epoch": 0.6125274725274725,
      "grad_norm": 0.24138215208784955,
      "learning_rate": 2.7594762095065507e-07,
      "loss": 1.4992,
      "step": 2787
    },
    {
      "epoch": 0.6127472527472527,
      "grad_norm": 0.24723985590687675,
      "learning_rate": 2.757436016509917e-07,
      "loss": 1.3818,
      "step": 2788
    },
    {
      "epoch": 0.6129670329670329,
      "grad_norm": 0.24340468458681574,
      "learning_rate": 2.75539634437665e-07,
      "loss": 1.4829,
      "step": 2789
    },
    {
      "epoch": 0.6131868131868132,
      "grad_norm": 0.3274566947817604,
      "learning_rate": 2.753357194079133e-07,
      "loss": 1.4621,
      "step": 2790
    },
    {
      "epoch": 0.6134065934065934,
      "grad_norm": 0.24919490504124517,
      "learning_rate": 2.751318566589501e-07,
      "loss": 1.4903,
      "step": 2791
    },
    {
      "epoch": 0.6136263736263736,
      "grad_norm": 0.25916066602605464,
      "learning_rate": 2.74928046287964e-07,
      "loss": 1.4575,
      "step": 2792
    },
    {
      "epoch": 0.6138461538461538,
      "grad_norm": 0.24608490593825452,
      "learning_rate": 2.747242883921182e-07,
      "loss": 1.5024,
      "step": 2793
    },
    {
      "epoch": 0.614065934065934,
      "grad_norm": 0.2441503092999619,
      "learning_rate": 2.7452058306855143e-07,
      "loss": 1.4518,
      "step": 2794
    },
    {
      "epoch": 0.6142857142857143,
      "grad_norm": 0.2592931844175535,
      "learning_rate": 2.7431693041437706e-07,
      "loss": 1.3783,
      "step": 2795
    },
    {
      "epoch": 0.6145054945054945,
      "grad_norm": 0.2992583932379197,
      "learning_rate": 2.741133305266834e-07,
      "loss": 1.4372,
      "step": 2796
    },
    {
      "epoch": 0.6147252747252747,
      "grad_norm": 0.24221330064764257,
      "learning_rate": 2.7390978350253374e-07,
      "loss": 1.4672,
      "step": 2797
    },
    {
      "epoch": 0.6149450549450549,
      "grad_norm": 0.28105113913083424,
      "learning_rate": 2.737062894389659e-07,
      "loss": 1.465,
      "step": 2798
    },
    {
      "epoch": 0.6151648351648351,
      "grad_norm": 0.2665490478140696,
      "learning_rate": 2.735028484329926e-07,
      "loss": 1.4044,
      "step": 2799
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 0.24423573775881333,
      "learning_rate": 2.7329946058160124e-07,
      "loss": 1.4441,
      "step": 2800
    },
    {
      "epoch": 0.6156043956043956,
      "grad_norm": 0.24854542416073258,
      "learning_rate": 2.730961259817542e-07,
      "loss": 1.5131,
      "step": 2801
    },
    {
      "epoch": 0.6158241758241758,
      "grad_norm": 0.2568503196130161,
      "learning_rate": 2.728928447303877e-07,
      "loss": 1.435,
      "step": 2802
    },
    {
      "epoch": 0.616043956043956,
      "grad_norm": 0.25751972570130277,
      "learning_rate": 2.7268961692441346e-07,
      "loss": 1.4825,
      "step": 2803
    },
    {
      "epoch": 0.6162637362637363,
      "grad_norm": 0.26758436474037745,
      "learning_rate": 2.724864426607169e-07,
      "loss": 1.488,
      "step": 2804
    },
    {
      "epoch": 0.6164835164835165,
      "grad_norm": 0.25463513002359184,
      "learning_rate": 2.722833220361586e-07,
      "loss": 1.4665,
      "step": 2805
    },
    {
      "epoch": 0.6167032967032967,
      "grad_norm": 0.24033732821235326,
      "learning_rate": 2.72080255147573e-07,
      "loss": 1.4403,
      "step": 2806
    },
    {
      "epoch": 0.6169230769230769,
      "grad_norm": 0.2537429863828503,
      "learning_rate": 2.718772420917693e-07,
      "loss": 1.4958,
      "step": 2807
    },
    {
      "epoch": 0.6171428571428571,
      "grad_norm": 0.3070515269029011,
      "learning_rate": 2.716742829655309e-07,
      "loss": 1.533,
      "step": 2808
    },
    {
      "epoch": 0.6173626373626374,
      "grad_norm": 0.26070131619033404,
      "learning_rate": 2.714713778656153e-07,
      "loss": 1.4411,
      "step": 2809
    },
    {
      "epoch": 0.6175824175824176,
      "grad_norm": 0.2448312377581891,
      "learning_rate": 2.7126852688875474e-07,
      "loss": 1.4124,
      "step": 2810
    },
    {
      "epoch": 0.6178021978021978,
      "grad_norm": 0.24533074910693486,
      "learning_rate": 2.710657301316552e-07,
      "loss": 1.4092,
      "step": 2811
    },
    {
      "epoch": 0.618021978021978,
      "grad_norm": 0.2622020381668056,
      "learning_rate": 2.70862987690997e-07,
      "loss": 1.4131,
      "step": 2812
    },
    {
      "epoch": 0.6182417582417582,
      "grad_norm": 0.24164843532840727,
      "learning_rate": 2.706602996634345e-07,
      "loss": 1.4463,
      "step": 2813
    },
    {
      "epoch": 0.6184615384615385,
      "grad_norm": 0.26821984758842227,
      "learning_rate": 2.7045766614559613e-07,
      "loss": 1.4513,
      "step": 2814
    },
    {
      "epoch": 0.6186813186813187,
      "grad_norm": 0.2527769764049168,
      "learning_rate": 2.702550872340846e-07,
      "loss": 1.4803,
      "step": 2815
    },
    {
      "epoch": 0.6189010989010989,
      "grad_norm": 0.24481534745646374,
      "learning_rate": 2.7005256302547604e-07,
      "loss": 1.4737,
      "step": 2816
    },
    {
      "epoch": 0.6191208791208791,
      "grad_norm": 0.24772386003665042,
      "learning_rate": 2.698500936163209e-07,
      "loss": 1.4522,
      "step": 2817
    },
    {
      "epoch": 0.6193406593406593,
      "grad_norm": 0.2505975677436217,
      "learning_rate": 2.696476791031434e-07,
      "loss": 1.5338,
      "step": 2818
    },
    {
      "epoch": 0.6195604395604396,
      "grad_norm": 0.25392803101198486,
      "learning_rate": 2.6944531958244165e-07,
      "loss": 1.5052,
      "step": 2819
    },
    {
      "epoch": 0.6197802197802198,
      "grad_norm": 0.23309877944596225,
      "learning_rate": 2.6924301515068745e-07,
      "loss": 1.4745,
      "step": 2820
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.26377695945308144,
      "learning_rate": 2.690407659043264e-07,
      "loss": 1.5323,
      "step": 2821
    },
    {
      "epoch": 0.6202197802197802,
      "grad_norm": 0.23633592621511978,
      "learning_rate": 2.688385719397779e-07,
      "loss": 1.3892,
      "step": 2822
    },
    {
      "epoch": 0.6204395604395604,
      "grad_norm": 0.2463089099013308,
      "learning_rate": 2.686364333534347e-07,
      "loss": 1.453,
      "step": 2823
    },
    {
      "epoch": 0.6206593406593407,
      "grad_norm": 0.2534261083630188,
      "learning_rate": 2.684343502416632e-07,
      "loss": 1.5113,
      "step": 2824
    },
    {
      "epoch": 0.6208791208791209,
      "grad_norm": 0.24994646596974002,
      "learning_rate": 2.682323227008038e-07,
      "loss": 1.451,
      "step": 2825
    },
    {
      "epoch": 0.6210989010989011,
      "grad_norm": 0.30067136753213547,
      "learning_rate": 2.680303508271698e-07,
      "loss": 1.4584,
      "step": 2826
    },
    {
      "epoch": 0.6213186813186813,
      "grad_norm": 0.25560280194093127,
      "learning_rate": 2.678284347170484e-07,
      "loss": 1.4125,
      "step": 2827
    },
    {
      "epoch": 0.6215384615384615,
      "grad_norm": 0.2941711091966941,
      "learning_rate": 2.676265744667e-07,
      "loss": 1.5021,
      "step": 2828
    },
    {
      "epoch": 0.6217582417582418,
      "grad_norm": 0.24605546671769116,
      "learning_rate": 2.6742477017235833e-07,
      "loss": 1.4621,
      "step": 2829
    },
    {
      "epoch": 0.621978021978022,
      "grad_norm": 0.2563288605501652,
      "learning_rate": 2.672230219302307e-07,
      "loss": 1.4661,
      "step": 2830
    },
    {
      "epoch": 0.6221978021978022,
      "grad_norm": 0.2483960316808464,
      "learning_rate": 2.670213298364975e-07,
      "loss": 1.48,
      "step": 2831
    },
    {
      "epoch": 0.6224175824175824,
      "grad_norm": 0.24389055588115652,
      "learning_rate": 2.6681969398731225e-07,
      "loss": 1.4399,
      "step": 2832
    },
    {
      "epoch": 0.6226373626373627,
      "grad_norm": 0.2461762059616389,
      "learning_rate": 2.6661811447880197e-07,
      "loss": 1.4945,
      "step": 2833
    },
    {
      "epoch": 0.6228571428571429,
      "grad_norm": 0.2471990459257803,
      "learning_rate": 2.664165914070665e-07,
      "loss": 1.522,
      "step": 2834
    },
    {
      "epoch": 0.6230769230769231,
      "grad_norm": 0.258707408088636,
      "learning_rate": 2.66215124868179e-07,
      "loss": 1.5095,
      "step": 2835
    },
    {
      "epoch": 0.6232967032967033,
      "grad_norm": 0.2507656223835275,
      "learning_rate": 2.660137149581857e-07,
      "loss": 1.4768,
      "step": 2836
    },
    {
      "epoch": 0.6235164835164835,
      "grad_norm": 0.26040518587086814,
      "learning_rate": 2.658123617731056e-07,
      "loss": 1.4672,
      "step": 2837
    },
    {
      "epoch": 0.6237362637362638,
      "grad_norm": 0.36898059415392886,
      "learning_rate": 2.6561106540893057e-07,
      "loss": 1.5357,
      "step": 2838
    },
    {
      "epoch": 0.623956043956044,
      "grad_norm": 0.2652248897625437,
      "learning_rate": 2.654098259616261e-07,
      "loss": 1.4572,
      "step": 2839
    },
    {
      "epoch": 0.6241758241758242,
      "grad_norm": 0.2489821663440469,
      "learning_rate": 2.6520864352712976e-07,
      "loss": 1.4843,
      "step": 2840
    },
    {
      "epoch": 0.6243956043956044,
      "grad_norm": 0.2594474675887204,
      "learning_rate": 2.650075182013523e-07,
      "loss": 1.4407,
      "step": 2841
    },
    {
      "epoch": 0.6246153846153846,
      "grad_norm": 0.25712986434896845,
      "learning_rate": 2.6480645008017706e-07,
      "loss": 1.4515,
      "step": 2842
    },
    {
      "epoch": 0.6248351648351649,
      "grad_norm": 0.2573780357581436,
      "learning_rate": 2.6460543925946023e-07,
      "loss": 1.4536,
      "step": 2843
    },
    {
      "epoch": 0.6250549450549451,
      "grad_norm": 0.30665596993707006,
      "learning_rate": 2.64404485835031e-07,
      "loss": 1.4529,
      "step": 2844
    },
    {
      "epoch": 0.6252747252747253,
      "grad_norm": 0.23967905051755029,
      "learning_rate": 2.6420358990269047e-07,
      "loss": 1.5155,
      "step": 2845
    },
    {
      "epoch": 0.6254945054945055,
      "grad_norm": 0.29993653997854336,
      "learning_rate": 2.640027515582131e-07,
      "loss": 1.424,
      "step": 2846
    },
    {
      "epoch": 0.6257142857142857,
      "grad_norm": 0.24506179692758245,
      "learning_rate": 2.6380197089734514e-07,
      "loss": 1.456,
      "step": 2847
    },
    {
      "epoch": 0.625934065934066,
      "grad_norm": 0.24575536037400184,
      "learning_rate": 2.6360124801580614e-07,
      "loss": 1.4429,
      "step": 2848
    },
    {
      "epoch": 0.6261538461538462,
      "grad_norm": 0.24513992249774819,
      "learning_rate": 2.634005830092874e-07,
      "loss": 1.4885,
      "step": 2849
    },
    {
      "epoch": 0.6263736263736264,
      "grad_norm": 0.2447623203675496,
      "learning_rate": 2.6319997597345303e-07,
      "loss": 1.4452,
      "step": 2850
    },
    {
      "epoch": 0.6265934065934066,
      "grad_norm": 0.24424616248784922,
      "learning_rate": 2.6299942700393944e-07,
      "loss": 1.4282,
      "step": 2851
    },
    {
      "epoch": 0.6268131868131868,
      "grad_norm": 0.2393103125713913,
      "learning_rate": 2.627989361963554e-07,
      "loss": 1.4944,
      "step": 2852
    },
    {
      "epoch": 0.6270329670329671,
      "grad_norm": 0.24372315941926556,
      "learning_rate": 2.6259850364628164e-07,
      "loss": 1.4597,
      "step": 2853
    },
    {
      "epoch": 0.6272527472527473,
      "grad_norm": 0.2576497581670632,
      "learning_rate": 2.623981294492716e-07,
      "loss": 1.4798,
      "step": 2854
    },
    {
      "epoch": 0.6274725274725275,
      "grad_norm": 0.25662975162566215,
      "learning_rate": 2.621978137008504e-07,
      "loss": 1.4821,
      "step": 2855
    },
    {
      "epoch": 0.6276923076923077,
      "grad_norm": 0.24280530676974668,
      "learning_rate": 2.619975564965158e-07,
      "loss": 1.4673,
      "step": 2856
    },
    {
      "epoch": 0.6279120879120879,
      "grad_norm": 0.23964867825847486,
      "learning_rate": 2.6179735793173716e-07,
      "loss": 1.499,
      "step": 2857
    },
    {
      "epoch": 0.6281318681318682,
      "grad_norm": 0.5861739492202969,
      "learning_rate": 2.615972181019562e-07,
      "loss": 1.408,
      "step": 2858
    },
    {
      "epoch": 0.6283516483516484,
      "grad_norm": 0.24986249129666405,
      "learning_rate": 2.613971371025867e-07,
      "loss": 1.4493,
      "step": 2859
    },
    {
      "epoch": 0.6285714285714286,
      "grad_norm": 0.24015338506617812,
      "learning_rate": 2.611971150290141e-07,
      "loss": 1.4288,
      "step": 2860
    },
    {
      "epoch": 0.6287912087912088,
      "grad_norm": 0.2463092478272792,
      "learning_rate": 2.6099715197659596e-07,
      "loss": 1.4322,
      "step": 2861
    },
    {
      "epoch": 0.6290109890109891,
      "grad_norm": 0.2532701143533487,
      "learning_rate": 2.607972480406615e-07,
      "loss": 1.4637,
      "step": 2862
    },
    {
      "epoch": 0.6292307692307693,
      "grad_norm": 0.2552728846307907,
      "learning_rate": 2.6059740331651217e-07,
      "loss": 1.4707,
      "step": 2863
    },
    {
      "epoch": 0.6294505494505495,
      "grad_norm": 0.2553736055829221,
      "learning_rate": 2.603976178994208e-07,
      "loss": 1.4731,
      "step": 2864
    },
    {
      "epoch": 0.6296703296703297,
      "grad_norm": 0.25576625836644434,
      "learning_rate": 2.6019789188463193e-07,
      "loss": 1.4295,
      "step": 2865
    },
    {
      "epoch": 0.6298901098901099,
      "grad_norm": 0.23985643407427887,
      "learning_rate": 2.599982253673621e-07,
      "loss": 1.4582,
      "step": 2866
    },
    {
      "epoch": 0.6301098901098902,
      "grad_norm": 0.25161103928396245,
      "learning_rate": 2.597986184427991e-07,
      "loss": 1.4415,
      "step": 2867
    },
    {
      "epoch": 0.6303296703296704,
      "grad_norm": 0.24200350717316507,
      "learning_rate": 2.595990712061028e-07,
      "loss": 1.5023,
      "step": 2868
    },
    {
      "epoch": 0.6305494505494506,
      "grad_norm": 0.2449181858527255,
      "learning_rate": 2.5939958375240426e-07,
      "loss": 1.4137,
      "step": 2869
    },
    {
      "epoch": 0.6307692307692307,
      "grad_norm": 0.24963028357508124,
      "learning_rate": 2.59200156176806e-07,
      "loss": 1.5189,
      "step": 2870
    },
    {
      "epoch": 0.630989010989011,
      "grad_norm": 0.2544803480741481,
      "learning_rate": 2.5900078857438207e-07,
      "loss": 1.4613,
      "step": 2871
    },
    {
      "epoch": 0.6312087912087913,
      "grad_norm": 0.25710601023148333,
      "learning_rate": 2.5880148104017794e-07,
      "loss": 1.4881,
      "step": 2872
    },
    {
      "epoch": 0.6314285714285715,
      "grad_norm": 0.2660596997956752,
      "learning_rate": 2.586022336692106e-07,
      "loss": 1.4239,
      "step": 2873
    },
    {
      "epoch": 0.6316483516483516,
      "grad_norm": 0.24672583608519805,
      "learning_rate": 2.584030465564683e-07,
      "loss": 1.4013,
      "step": 2874
    },
    {
      "epoch": 0.6318681318681318,
      "grad_norm": 0.25768288683848095,
      "learning_rate": 2.5820391979691025e-07,
      "loss": 1.4666,
      "step": 2875
    },
    {
      "epoch": 0.632087912087912,
      "grad_norm": 0.25865658897524546,
      "learning_rate": 2.580048534854673e-07,
      "loss": 1.529,
      "step": 2876
    },
    {
      "epoch": 0.6323076923076923,
      "grad_norm": 0.24769300741875816,
      "learning_rate": 2.5780584771704116e-07,
      "loss": 1.4807,
      "step": 2877
    },
    {
      "epoch": 0.6325274725274725,
      "grad_norm": 0.2628685955578374,
      "learning_rate": 2.5760690258650496e-07,
      "loss": 1.4682,
      "step": 2878
    },
    {
      "epoch": 0.6327472527472527,
      "grad_norm": 0.2486143115957994,
      "learning_rate": 2.5740801818870267e-07,
      "loss": 1.4692,
      "step": 2879
    },
    {
      "epoch": 0.6329670329670329,
      "grad_norm": 0.30566491823381436,
      "learning_rate": 2.572091946184496e-07,
      "loss": 1.4942,
      "step": 2880
    },
    {
      "epoch": 0.6331868131868131,
      "grad_norm": 0.26502102734665955,
      "learning_rate": 2.570104319705317e-07,
      "loss": 1.4832,
      "step": 2881
    },
    {
      "epoch": 0.6334065934065934,
      "grad_norm": 0.2558474918238458,
      "learning_rate": 2.56811730339706e-07,
      "loss": 1.512,
      "step": 2882
    },
    {
      "epoch": 0.6336263736263736,
      "grad_norm": 0.23618937873946327,
      "learning_rate": 2.5661308982070065e-07,
      "loss": 1.4216,
      "step": 2883
    },
    {
      "epoch": 0.6338461538461538,
      "grad_norm": 0.24607635970302394,
      "learning_rate": 2.564145105082147e-07,
      "loss": 1.4578,
      "step": 2884
    },
    {
      "epoch": 0.634065934065934,
      "grad_norm": 0.37591479413491696,
      "learning_rate": 2.5621599249691746e-07,
      "loss": 1.4868,
      "step": 2885
    },
    {
      "epoch": 0.6342857142857142,
      "grad_norm": 0.25925714314608084,
      "learning_rate": 2.5601753588144953e-07,
      "loss": 1.5097,
      "step": 2886
    },
    {
      "epoch": 0.6345054945054945,
      "grad_norm": 0.25206041921771516,
      "learning_rate": 2.558191407564221e-07,
      "loss": 1.3957,
      "step": 2887
    },
    {
      "epoch": 0.6347252747252747,
      "grad_norm": 0.2388974492827397,
      "learning_rate": 2.5562080721641715e-07,
      "loss": 1.5077,
      "step": 2888
    },
    {
      "epoch": 0.6349450549450549,
      "grad_norm": 0.2352967262259724,
      "learning_rate": 2.5542253535598717e-07,
      "loss": 1.4807,
      "step": 2889
    },
    {
      "epoch": 0.6351648351648351,
      "grad_norm": 0.2446569128586504,
      "learning_rate": 2.552243252696551e-07,
      "loss": 1.4399,
      "step": 2890
    },
    {
      "epoch": 0.6353846153846154,
      "grad_norm": 0.2365234128733907,
      "learning_rate": 2.550261770519148e-07,
      "loss": 1.4485,
      "step": 2891
    },
    {
      "epoch": 0.6356043956043956,
      "grad_norm": 0.24301965126548045,
      "learning_rate": 2.5482809079723024e-07,
      "loss": 1.4666,
      "step": 2892
    },
    {
      "epoch": 0.6358241758241758,
      "grad_norm": 0.24242178670602155,
      "learning_rate": 2.5463006660003637e-07,
      "loss": 1.5414,
      "step": 2893
    },
    {
      "epoch": 0.636043956043956,
      "grad_norm": 0.24207991875232743,
      "learning_rate": 2.5443210455473786e-07,
      "loss": 1.4618,
      "step": 2894
    },
    {
      "epoch": 0.6362637362637362,
      "grad_norm": 0.2567225455233669,
      "learning_rate": 2.542342047557105e-07,
      "loss": 1.4766,
      "step": 2895
    },
    {
      "epoch": 0.6364835164835165,
      "grad_norm": 0.24731626801041443,
      "learning_rate": 2.5403636729729964e-07,
      "loss": 1.4465,
      "step": 2896
    },
    {
      "epoch": 0.6367032967032967,
      "grad_norm": 0.24926976310952406,
      "learning_rate": 2.538385922738215e-07,
      "loss": 1.4898,
      "step": 2897
    },
    {
      "epoch": 0.6369230769230769,
      "grad_norm": 0.23383539155294128,
      "learning_rate": 2.5364087977956243e-07,
      "loss": 1.4101,
      "step": 2898
    },
    {
      "epoch": 0.6371428571428571,
      "grad_norm": 0.251464325919328,
      "learning_rate": 2.534432299087787e-07,
      "loss": 1.4793,
      "step": 2899
    },
    {
      "epoch": 0.6373626373626373,
      "grad_norm": 0.24509098112180402,
      "learning_rate": 2.5324564275569706e-07,
      "loss": 1.4745,
      "step": 2900
    },
    {
      "epoch": 0.6375824175824176,
      "grad_norm": 0.2663807872672918,
      "learning_rate": 2.5304811841451395e-07,
      "loss": 1.4217,
      "step": 2901
    },
    {
      "epoch": 0.6378021978021978,
      "grad_norm": 0.23696124194719534,
      "learning_rate": 2.5285065697939634e-07,
      "loss": 1.4688,
      "step": 2902
    },
    {
      "epoch": 0.638021978021978,
      "grad_norm": 0.2456102335966778,
      "learning_rate": 2.526532585444808e-07,
      "loss": 1.5261,
      "step": 2903
    },
    {
      "epoch": 0.6382417582417582,
      "grad_norm": 0.24972928092226704,
      "learning_rate": 2.524559232038743e-07,
      "loss": 1.4612,
      "step": 2904
    },
    {
      "epoch": 0.6384615384615384,
      "grad_norm": 0.25201685389695233,
      "learning_rate": 2.522586510516534e-07,
      "loss": 1.44,
      "step": 2905
    },
    {
      "epoch": 0.6386813186813187,
      "grad_norm": 0.255514240779577,
      "learning_rate": 2.5206144218186436e-07,
      "loss": 1.5007,
      "step": 2906
    },
    {
      "epoch": 0.6389010989010989,
      "grad_norm": 0.25577694608065626,
      "learning_rate": 2.5186429668852386e-07,
      "loss": 1.421,
      "step": 2907
    },
    {
      "epoch": 0.6391208791208791,
      "grad_norm": 0.25157469910805375,
      "learning_rate": 2.5166721466561795e-07,
      "loss": 1.4542,
      "step": 2908
    },
    {
      "epoch": 0.6393406593406593,
      "grad_norm": 0.25938177309484367,
      "learning_rate": 2.5147019620710245e-07,
      "loss": 1.511,
      "step": 2909
    },
    {
      "epoch": 0.6395604395604395,
      "grad_norm": 0.24075538377136294,
      "learning_rate": 2.512732414069028e-07,
      "loss": 1.455,
      "step": 2910
    },
    {
      "epoch": 0.6397802197802198,
      "grad_norm": 0.259048723673585,
      "learning_rate": 2.5107635035891437e-07,
      "loss": 1.4618,
      "step": 2911
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.25283089613057796,
      "learning_rate": 2.508795231570021e-07,
      "loss": 1.4847,
      "step": 2912
    },
    {
      "epoch": 0.6402197802197802,
      "grad_norm": 0.32406040698220695,
      "learning_rate": 2.5068275989500023e-07,
      "loss": 1.5001,
      "step": 2913
    },
    {
      "epoch": 0.6404395604395604,
      "grad_norm": 0.27582907409618784,
      "learning_rate": 2.504860606667128e-07,
      "loss": 1.5091,
      "step": 2914
    },
    {
      "epoch": 0.6406593406593407,
      "grad_norm": 0.2524924066748553,
      "learning_rate": 2.5028942556591296e-07,
      "loss": 1.4731,
      "step": 2915
    },
    {
      "epoch": 0.6408791208791209,
      "grad_norm": 0.25315946602881495,
      "learning_rate": 2.5009285468634377e-07,
      "loss": 1.4732,
      "step": 2916
    },
    {
      "epoch": 0.6410989010989011,
      "grad_norm": 0.2472126417658668,
      "learning_rate": 2.498963481217174e-07,
      "loss": 1.4972,
      "step": 2917
    },
    {
      "epoch": 0.6413186813186813,
      "grad_norm": 0.26297434701382966,
      "learning_rate": 2.496999059657151e-07,
      "loss": 1.3957,
      "step": 2918
    },
    {
      "epoch": 0.6415384615384615,
      "grad_norm": 0.2540569813398366,
      "learning_rate": 2.4950352831198806e-07,
      "loss": 1.4581,
      "step": 2919
    },
    {
      "epoch": 0.6417582417582418,
      "grad_norm": 0.2592848693101979,
      "learning_rate": 2.493072152541561e-07,
      "loss": 1.5138,
      "step": 2920
    },
    {
      "epoch": 0.641978021978022,
      "grad_norm": 0.3090859085566191,
      "learning_rate": 2.4911096688580873e-07,
      "loss": 1.4268,
      "step": 2921
    },
    {
      "epoch": 0.6421978021978022,
      "grad_norm": 0.23717743116654658,
      "learning_rate": 2.4891478330050427e-07,
      "loss": 1.4456,
      "step": 2922
    },
    {
      "epoch": 0.6424175824175824,
      "grad_norm": 0.25147376891162104,
      "learning_rate": 2.4871866459177035e-07,
      "loss": 1.4794,
      "step": 2923
    },
    {
      "epoch": 0.6426373626373626,
      "grad_norm": 0.2460244860737173,
      "learning_rate": 2.485226108531034e-07,
      "loss": 1.3746,
      "step": 2924
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 0.25053272659868825,
      "learning_rate": 2.483266221779692e-07,
      "loss": 1.4533,
      "step": 2925
    },
    {
      "epoch": 0.6430769230769231,
      "grad_norm": 0.24555094406174524,
      "learning_rate": 2.4813069865980243e-07,
      "loss": 1.4907,
      "step": 2926
    },
    {
      "epoch": 0.6432967032967033,
      "grad_norm": 0.2508731121060695,
      "learning_rate": 2.479348403920067e-07,
      "loss": 1.4715,
      "step": 2927
    },
    {
      "epoch": 0.6435164835164835,
      "grad_norm": 0.2694951088347539,
      "learning_rate": 2.4773904746795437e-07,
      "loss": 1.4947,
      "step": 2928
    },
    {
      "epoch": 0.6437362637362637,
      "grad_norm": 0.2503538129609662,
      "learning_rate": 2.475433199809869e-07,
      "loss": 1.4644,
      "step": 2929
    },
    {
      "epoch": 0.643956043956044,
      "grad_norm": 0.24790099011997746,
      "learning_rate": 2.4734765802441417e-07,
      "loss": 1.4962,
      "step": 2930
    },
    {
      "epoch": 0.6441758241758242,
      "grad_norm": 0.2535367449362057,
      "learning_rate": 2.4715206169151537e-07,
      "loss": 1.5242,
      "step": 2931
    },
    {
      "epoch": 0.6443956043956044,
      "grad_norm": 0.24765899273957198,
      "learning_rate": 2.4695653107553787e-07,
      "loss": 1.3793,
      "step": 2932
    },
    {
      "epoch": 0.6446153846153846,
      "grad_norm": 0.2568970873517436,
      "learning_rate": 2.467610662696979e-07,
      "loss": 1.397,
      "step": 2933
    },
    {
      "epoch": 0.6448351648351648,
      "grad_norm": 0.24359852655681025,
      "learning_rate": 2.4656566736718055e-07,
      "loss": 1.4776,
      "step": 2934
    },
    {
      "epoch": 0.6450549450549451,
      "grad_norm": 0.2521438960939031,
      "learning_rate": 2.463703344611391e-07,
      "loss": 1.5206,
      "step": 2935
    },
    {
      "epoch": 0.6452747252747253,
      "grad_norm": 0.2533804782220817,
      "learning_rate": 2.4617506764469583e-07,
      "loss": 1.431,
      "step": 2936
    },
    {
      "epoch": 0.6454945054945055,
      "grad_norm": 0.24411108105400223,
      "learning_rate": 2.4597986701094094e-07,
      "loss": 1.4873,
      "step": 2937
    },
    {
      "epoch": 0.6457142857142857,
      "grad_norm": 0.2749327049552055,
      "learning_rate": 2.457847326529337e-07,
      "loss": 1.4897,
      "step": 2938
    },
    {
      "epoch": 0.6459340659340659,
      "grad_norm": 0.2556229087342906,
      "learning_rate": 2.455896646637011e-07,
      "loss": 1.4577,
      "step": 2939
    },
    {
      "epoch": 0.6461538461538462,
      "grad_norm": 0.2429621105813027,
      "learning_rate": 2.45394663136239e-07,
      "loss": 1.4333,
      "step": 2940
    },
    {
      "epoch": 0.6463736263736264,
      "grad_norm": 0.28905077326746476,
      "learning_rate": 2.4519972816351147e-07,
      "loss": 1.4174,
      "step": 2941
    },
    {
      "epoch": 0.6465934065934066,
      "grad_norm": 0.2692161782606739,
      "learning_rate": 2.450048598384509e-07,
      "loss": 1.464,
      "step": 2942
    },
    {
      "epoch": 0.6468131868131868,
      "grad_norm": 0.2599451377641455,
      "learning_rate": 2.4481005825395777e-07,
      "loss": 1.4712,
      "step": 2943
    },
    {
      "epoch": 0.6470329670329671,
      "grad_norm": 0.25847217362042313,
      "learning_rate": 2.446153235029006e-07,
      "loss": 1.5089,
      "step": 2944
    },
    {
      "epoch": 0.6472527472527473,
      "grad_norm": 0.25025000456975677,
      "learning_rate": 2.4442065567811653e-07,
      "loss": 1.4325,
      "step": 2945
    },
    {
      "epoch": 0.6474725274725275,
      "grad_norm": 0.3062559101992876,
      "learning_rate": 2.4422605487241036e-07,
      "loss": 1.4408,
      "step": 2946
    },
    {
      "epoch": 0.6476923076923077,
      "grad_norm": 0.2544567029041009,
      "learning_rate": 2.440315211785551e-07,
      "loss": 1.5209,
      "step": 2947
    },
    {
      "epoch": 0.6479120879120879,
      "grad_norm": 0.2517821459407482,
      "learning_rate": 2.438370546892916e-07,
      "loss": 1.4349,
      "step": 2948
    },
    {
      "epoch": 0.6481318681318682,
      "grad_norm": 0.25776996022413107,
      "learning_rate": 2.4364265549732907e-07,
      "loss": 1.453,
      "step": 2949
    },
    {
      "epoch": 0.6483516483516484,
      "grad_norm": 0.2627200763149241,
      "learning_rate": 2.434483236953441e-07,
      "loss": 1.4635,
      "step": 2950
    },
    {
      "epoch": 0.6485714285714286,
      "grad_norm": 0.24327742804716065,
      "learning_rate": 2.4325405937598185e-07,
      "loss": 1.4517,
      "step": 2951
    },
    {
      "epoch": 0.6487912087912088,
      "grad_norm": 0.2504827525899091,
      "learning_rate": 2.430598626318546e-07,
      "loss": 1.4397,
      "step": 2952
    },
    {
      "epoch": 0.649010989010989,
      "grad_norm": 0.2425369524665089,
      "learning_rate": 2.4286573355554274e-07,
      "loss": 1.4879,
      "step": 2953
    },
    {
      "epoch": 0.6492307692307693,
      "grad_norm": 0.25325488583168404,
      "learning_rate": 2.4267167223959435e-07,
      "loss": 1.4791,
      "step": 2954
    },
    {
      "epoch": 0.6494505494505495,
      "grad_norm": 0.25234062890244585,
      "learning_rate": 2.424776787765254e-07,
      "loss": 1.5006,
      "step": 2955
    },
    {
      "epoch": 0.6496703296703297,
      "grad_norm": 0.24206717094628213,
      "learning_rate": 2.42283753258819e-07,
      "loss": 1.446,
      "step": 2956
    },
    {
      "epoch": 0.6498901098901099,
      "grad_norm": 0.24610605739098584,
      "learning_rate": 2.420898957789267e-07,
      "loss": 1.4242,
      "step": 2957
    },
    {
      "epoch": 0.6501098901098901,
      "grad_norm": 0.27691333257297623,
      "learning_rate": 2.4189610642926674e-07,
      "loss": 1.4453,
      "step": 2958
    },
    {
      "epoch": 0.6503296703296704,
      "grad_norm": 0.24930504957336627,
      "learning_rate": 2.4170238530222535e-07,
      "loss": 1.4866,
      "step": 2959
    },
    {
      "epoch": 0.6505494505494506,
      "grad_norm": 0.255335178655049,
      "learning_rate": 2.4150873249015626e-07,
      "loss": 1.5051,
      "step": 2960
    },
    {
      "epoch": 0.6507692307692308,
      "grad_norm": 0.24921237889027203,
      "learning_rate": 2.413151480853805e-07,
      "loss": 1.4839,
      "step": 2961
    },
    {
      "epoch": 0.650989010989011,
      "grad_norm": 0.23625646421617658,
      "learning_rate": 2.4112163218018643e-07,
      "loss": 1.4502,
      "step": 2962
    },
    {
      "epoch": 0.6512087912087912,
      "grad_norm": 0.26607537504914247,
      "learning_rate": 2.409281848668298e-07,
      "loss": 1.527,
      "step": 2963
    },
    {
      "epoch": 0.6514285714285715,
      "grad_norm": 0.37042601653975316,
      "learning_rate": 2.407348062375338e-07,
      "loss": 1.4136,
      "step": 2964
    },
    {
      "epoch": 0.6516483516483517,
      "grad_norm": 0.24518878874398756,
      "learning_rate": 2.405414963844886e-07,
      "loss": 1.5283,
      "step": 2965
    },
    {
      "epoch": 0.6518681318681319,
      "grad_norm": 0.2598143109317973,
      "learning_rate": 2.40348255399852e-07,
      "loss": 1.4527,
      "step": 2966
    },
    {
      "epoch": 0.6520879120879121,
      "grad_norm": 0.25024726976257977,
      "learning_rate": 2.401550833757486e-07,
      "loss": 1.4782,
      "step": 2967
    },
    {
      "epoch": 0.6523076923076923,
      "grad_norm": 0.26792275526618126,
      "learning_rate": 2.3996198040427027e-07,
      "loss": 1.4535,
      "step": 2968
    },
    {
      "epoch": 0.6525274725274726,
      "grad_norm": 0.3656491138960403,
      "learning_rate": 2.397689465774757e-07,
      "loss": 1.4651,
      "step": 2969
    },
    {
      "epoch": 0.6527472527472528,
      "grad_norm": 0.24484044654345521,
      "learning_rate": 2.3957598198739124e-07,
      "loss": 1.4874,
      "step": 2970
    },
    {
      "epoch": 0.652967032967033,
      "grad_norm": 0.25598445315108104,
      "learning_rate": 2.3938308672600956e-07,
      "loss": 1.4653,
      "step": 2971
    },
    {
      "epoch": 0.6531868131868132,
      "grad_norm": 0.25291595516474,
      "learning_rate": 2.391902608852906e-07,
      "loss": 1.438,
      "step": 2972
    },
    {
      "epoch": 0.6534065934065935,
      "grad_norm": 0.2573687623226722,
      "learning_rate": 2.3899750455716123e-07,
      "loss": 1.4504,
      "step": 2973
    },
    {
      "epoch": 0.6536263736263737,
      "grad_norm": 0.2640884074388109,
      "learning_rate": 2.388048178335152e-07,
      "loss": 1.4663,
      "step": 2974
    },
    {
      "epoch": 0.6538461538461539,
      "grad_norm": 0.24004062990968525,
      "learning_rate": 2.386122008062129e-07,
      "loss": 1.396,
      "step": 2975
    },
    {
      "epoch": 0.654065934065934,
      "grad_norm": 0.25327911391656,
      "learning_rate": 2.3841965356708157e-07,
      "loss": 1.5331,
      "step": 2976
    },
    {
      "epoch": 0.6542857142857142,
      "grad_norm": 0.24644399792474514,
      "learning_rate": 2.3822717620791528e-07,
      "loss": 1.442,
      "step": 2977
    },
    {
      "epoch": 0.6545054945054946,
      "grad_norm": 0.24948516403498913,
      "learning_rate": 2.3803476882047442e-07,
      "loss": 1.4755,
      "step": 2978
    },
    {
      "epoch": 0.6547252747252748,
      "grad_norm": 0.24073357890270214,
      "learning_rate": 2.3784243149648666e-07,
      "loss": 1.4767,
      "step": 2979
    },
    {
      "epoch": 0.654945054945055,
      "grad_norm": 0.283618267596697,
      "learning_rate": 2.376501643276457e-07,
      "loss": 1.4586,
      "step": 2980
    },
    {
      "epoch": 0.6551648351648351,
      "grad_norm": 0.2468768745864606,
      "learning_rate": 2.3745796740561218e-07,
      "loss": 1.4543,
      "step": 2981
    },
    {
      "epoch": 0.6553846153846153,
      "grad_norm": 0.2568937663514108,
      "learning_rate": 2.3726584082201295e-07,
      "loss": 1.492,
      "step": 2982
    },
    {
      "epoch": 0.6556043956043957,
      "grad_norm": 0.2618632259781446,
      "learning_rate": 2.3707378466844135e-07,
      "loss": 1.4513,
      "step": 2983
    },
    {
      "epoch": 0.6558241758241758,
      "grad_norm": 0.25599881617870923,
      "learning_rate": 2.3688179903645752e-07,
      "loss": 1.5482,
      "step": 2984
    },
    {
      "epoch": 0.656043956043956,
      "grad_norm": 0.2571918931916502,
      "learning_rate": 2.366898840175875e-07,
      "loss": 1.4192,
      "step": 2985
    },
    {
      "epoch": 0.6562637362637362,
      "grad_norm": 0.2412963502691558,
      "learning_rate": 2.3649803970332405e-07,
      "loss": 1.5063,
      "step": 2986
    },
    {
      "epoch": 0.6564835164835164,
      "grad_norm": 0.23989128783862962,
      "learning_rate": 2.3630626618512578e-07,
      "loss": 1.4124,
      "step": 2987
    },
    {
      "epoch": 0.6567032967032967,
      "grad_norm": 0.24540190318494906,
      "learning_rate": 2.3611456355441795e-07,
      "loss": 1.4372,
      "step": 2988
    },
    {
      "epoch": 0.6569230769230769,
      "grad_norm": 0.2498617140330551,
      "learning_rate": 2.3592293190259204e-07,
      "loss": 1.4829,
      "step": 2989
    },
    {
      "epoch": 0.6571428571428571,
      "grad_norm": 0.2523377042802845,
      "learning_rate": 2.3573137132100543e-07,
      "loss": 1.4521,
      "step": 2990
    },
    {
      "epoch": 0.6573626373626373,
      "grad_norm": 0.24784159630196054,
      "learning_rate": 2.3553988190098174e-07,
      "loss": 1.5054,
      "step": 2991
    },
    {
      "epoch": 0.6575824175824175,
      "grad_norm": 0.2473577182913839,
      "learning_rate": 2.3534846373381062e-07,
      "loss": 1.554,
      "step": 2992
    },
    {
      "epoch": 0.6578021978021978,
      "grad_norm": 0.2465708419356356,
      "learning_rate": 2.351571169107477e-07,
      "loss": 1.4286,
      "step": 2993
    },
    {
      "epoch": 0.658021978021978,
      "grad_norm": 0.2570133814109051,
      "learning_rate": 2.349658415230149e-07,
      "loss": 1.469,
      "step": 2994
    },
    {
      "epoch": 0.6582417582417582,
      "grad_norm": 0.24891790017703455,
      "learning_rate": 2.3477463766179964e-07,
      "loss": 1.4992,
      "step": 2995
    },
    {
      "epoch": 0.6584615384615384,
      "grad_norm": 0.24462414762608164,
      "learning_rate": 2.3458350541825577e-07,
      "loss": 1.4019,
      "step": 2996
    },
    {
      "epoch": 0.6586813186813186,
      "grad_norm": 0.2545435603975285,
      "learning_rate": 2.3439244488350255e-07,
      "loss": 1.4369,
      "step": 2997
    },
    {
      "epoch": 0.6589010989010989,
      "grad_norm": 0.24277545462796876,
      "learning_rate": 2.3420145614862501e-07,
      "loss": 1.4799,
      "step": 2998
    },
    {
      "epoch": 0.6591208791208791,
      "grad_norm": 0.25252291174055863,
      "learning_rate": 2.3401053930467438e-07,
      "loss": 1.4394,
      "step": 2999
    },
    {
      "epoch": 0.6593406593406593,
      "grad_norm": 0.2578725036009162,
      "learning_rate": 2.3381969444266735e-07,
      "loss": 1.4569,
      "step": 3000
    },
    {
      "epoch": 0.6595604395604395,
      "grad_norm": 0.24158085775251778,
      "learning_rate": 2.3362892165358633e-07,
      "loss": 1.3909,
      "step": 3001
    },
    {
      "epoch": 0.6597802197802198,
      "grad_norm": 0.25588298250955477,
      "learning_rate": 2.3343822102837913e-07,
      "loss": 1.5275,
      "step": 3002
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24381431471343762,
      "learning_rate": 2.3324759265795963e-07,
      "loss": 1.4082,
      "step": 3003
    },
    {
      "epoch": 0.6602197802197802,
      "grad_norm": 0.2536951003775051,
      "learning_rate": 2.330570366332071e-07,
      "loss": 1.4535,
      "step": 3004
    },
    {
      "epoch": 0.6604395604395604,
      "grad_norm": 0.26507366301011936,
      "learning_rate": 2.328665530449661e-07,
      "loss": 1.4707,
      "step": 3005
    },
    {
      "epoch": 0.6606593406593406,
      "grad_norm": 0.24127812506832325,
      "learning_rate": 2.3267614198404682e-07,
      "loss": 1.4423,
      "step": 3006
    },
    {
      "epoch": 0.6608791208791209,
      "grad_norm": 0.2557160869294038,
      "learning_rate": 2.3248580354122476e-07,
      "loss": 1.4679,
      "step": 3007
    },
    {
      "epoch": 0.6610989010989011,
      "grad_norm": 0.24916186623065067,
      "learning_rate": 2.3229553780724108e-07,
      "loss": 1.4679,
      "step": 3008
    },
    {
      "epoch": 0.6613186813186813,
      "grad_norm": 0.25841460090363655,
      "learning_rate": 2.3210534487280205e-07,
      "loss": 1.4534,
      "step": 3009
    },
    {
      "epoch": 0.6615384615384615,
      "grad_norm": 0.2376517574484778,
      "learning_rate": 2.3191522482857908e-07,
      "loss": 1.4277,
      "step": 3010
    },
    {
      "epoch": 0.6617582417582417,
      "grad_norm": 0.29047109814999295,
      "learning_rate": 2.3172517776520933e-07,
      "loss": 1.5039,
      "step": 3011
    },
    {
      "epoch": 0.661978021978022,
      "grad_norm": 0.2854690411185694,
      "learning_rate": 2.3153520377329458e-07,
      "loss": 1.4527,
      "step": 3012
    },
    {
      "epoch": 0.6621978021978022,
      "grad_norm": 0.25168645703946924,
      "learning_rate": 2.3134530294340232e-07,
      "loss": 1.4274,
      "step": 3013
    },
    {
      "epoch": 0.6624175824175824,
      "grad_norm": 0.24783582545489324,
      "learning_rate": 2.3115547536606478e-07,
      "loss": 1.5132,
      "step": 3014
    },
    {
      "epoch": 0.6626373626373626,
      "grad_norm": 0.24510804262015817,
      "learning_rate": 2.3096572113177937e-07,
      "loss": 1.5068,
      "step": 3015
    },
    {
      "epoch": 0.6628571428571428,
      "grad_norm": 0.2555103275469262,
      "learning_rate": 2.307760403310086e-07,
      "loss": 1.4884,
      "step": 3016
    },
    {
      "epoch": 0.6630769230769231,
      "grad_norm": 0.24574180794453893,
      "learning_rate": 2.3058643305417975e-07,
      "loss": 1.4446,
      "step": 3017
    },
    {
      "epoch": 0.6632967032967033,
      "grad_norm": 0.2447077574083961,
      "learning_rate": 2.3039689939168542e-07,
      "loss": 1.5006,
      "step": 3018
    },
    {
      "epoch": 0.6635164835164835,
      "grad_norm": 0.34544343390585736,
      "learning_rate": 2.3020743943388296e-07,
      "loss": 1.511,
      "step": 3019
    },
    {
      "epoch": 0.6637362637362637,
      "grad_norm": 0.250237284650321,
      "learning_rate": 2.3001805327109445e-07,
      "loss": 1.4755,
      "step": 3020
    },
    {
      "epoch": 0.6639560439560439,
      "grad_norm": 0.24431944270074715,
      "learning_rate": 2.2982874099360686e-07,
      "loss": 1.5218,
      "step": 3021
    },
    {
      "epoch": 0.6641758241758242,
      "grad_norm": 0.24619630632580405,
      "learning_rate": 2.296395026916719e-07,
      "loss": 1.5287,
      "step": 3022
    },
    {
      "epoch": 0.6643956043956044,
      "grad_norm": 0.2421460507560475,
      "learning_rate": 2.2945033845550628e-07,
      "loss": 1.4595,
      "step": 3023
    },
    {
      "epoch": 0.6646153846153846,
      "grad_norm": 0.2829390316329465,
      "learning_rate": 2.29261248375291e-07,
      "loss": 1.4412,
      "step": 3024
    },
    {
      "epoch": 0.6648351648351648,
      "grad_norm": 0.25538296977530495,
      "learning_rate": 2.2907223254117183e-07,
      "loss": 1.3967,
      "step": 3025
    },
    {
      "epoch": 0.665054945054945,
      "grad_norm": 0.3169874856464177,
      "learning_rate": 2.2888329104325956e-07,
      "loss": 1.5162,
      "step": 3026
    },
    {
      "epoch": 0.6652747252747253,
      "grad_norm": 0.24591002130587566,
      "learning_rate": 2.2869442397162882e-07,
      "loss": 1.436,
      "step": 3027
    },
    {
      "epoch": 0.6654945054945055,
      "grad_norm": 1.2438378204165965,
      "learning_rate": 2.2850563141631935e-07,
      "loss": 1.5281,
      "step": 3028
    },
    {
      "epoch": 0.6657142857142857,
      "grad_norm": 0.23552149623893676,
      "learning_rate": 2.283169134673351e-07,
      "loss": 1.4341,
      "step": 3029
    },
    {
      "epoch": 0.6659340659340659,
      "grad_norm": 0.43744069766685717,
      "learning_rate": 2.2812827021464457e-07,
      "loss": 1.4989,
      "step": 3030
    },
    {
      "epoch": 0.6661538461538462,
      "grad_norm": 0.27395660410861833,
      "learning_rate": 2.2793970174818025e-07,
      "loss": 1.4418,
      "step": 3031
    },
    {
      "epoch": 0.6663736263736264,
      "grad_norm": 0.2605035534891694,
      "learning_rate": 2.2775120815783972e-07,
      "loss": 1.4687,
      "step": 3032
    },
    {
      "epoch": 0.6665934065934066,
      "grad_norm": 0.29914621715832324,
      "learning_rate": 2.275627895334842e-07,
      "loss": 1.4947,
      "step": 3033
    },
    {
      "epoch": 0.6668131868131868,
      "grad_norm": 0.24979565628141243,
      "learning_rate": 2.273744459649396e-07,
      "loss": 1.4918,
      "step": 3034
    },
    {
      "epoch": 0.667032967032967,
      "grad_norm": 0.24666021686537823,
      "learning_rate": 2.2718617754199578e-07,
      "loss": 1.427,
      "step": 3035
    },
    {
      "epoch": 0.6672527472527473,
      "grad_norm": 0.26473206776587377,
      "learning_rate": 2.269979843544067e-07,
      "loss": 1.3772,
      "step": 3036
    },
    {
      "epoch": 0.6674725274725275,
      "grad_norm": 0.24762613360330796,
      "learning_rate": 2.268098664918909e-07,
      "loss": 1.4845,
      "step": 3037
    },
    {
      "epoch": 0.6676923076923077,
      "grad_norm": 0.2755099801753309,
      "learning_rate": 2.2662182404413068e-07,
      "loss": 1.5255,
      "step": 3038
    },
    {
      "epoch": 0.6679120879120879,
      "grad_norm": 0.24234455204365682,
      "learning_rate": 2.2643385710077234e-07,
      "loss": 1.5065,
      "step": 3039
    },
    {
      "epoch": 0.6681318681318681,
      "grad_norm": 0.24716690624443044,
      "learning_rate": 2.2624596575142615e-07,
      "loss": 1.43,
      "step": 3040
    },
    {
      "epoch": 0.6683516483516484,
      "grad_norm": 0.25925718573281836,
      "learning_rate": 2.2605815008566662e-07,
      "loss": 1.5404,
      "step": 3041
    },
    {
      "epoch": 0.6685714285714286,
      "grad_norm": 0.24592395223821256,
      "learning_rate": 2.2587041019303217e-07,
      "loss": 1.5795,
      "step": 3042
    },
    {
      "epoch": 0.6687912087912088,
      "grad_norm": 0.4111149414258364,
      "learning_rate": 2.2568274616302478e-07,
      "loss": 1.4284,
      "step": 3043
    },
    {
      "epoch": 0.669010989010989,
      "grad_norm": 0.2681271119853714,
      "learning_rate": 2.254951580851105e-07,
      "loss": 1.5025,
      "step": 3044
    },
    {
      "epoch": 0.6692307692307692,
      "grad_norm": 0.2564820006540057,
      "learning_rate": 2.2530764604871915e-07,
      "loss": 1.4669,
      "step": 3045
    },
    {
      "epoch": 0.6694505494505495,
      "grad_norm": 0.23993998136149258,
      "learning_rate": 2.2512021014324398e-07,
      "loss": 1.4381,
      "step": 3046
    },
    {
      "epoch": 0.6696703296703297,
      "grad_norm": 0.24597328410825253,
      "learning_rate": 2.2493285045804263e-07,
      "loss": 1.4429,
      "step": 3047
    },
    {
      "epoch": 0.6698901098901099,
      "grad_norm": 0.2442446102577392,
      "learning_rate": 2.2474556708243567e-07,
      "loss": 1.4278,
      "step": 3048
    },
    {
      "epoch": 0.6701098901098901,
      "grad_norm": 0.2660171934620519,
      "learning_rate": 2.245583601057079e-07,
      "loss": 1.4914,
      "step": 3049
    },
    {
      "epoch": 0.6703296703296703,
      "grad_norm": 0.24590583067691427,
      "learning_rate": 2.2437122961710723e-07,
      "loss": 1.4517,
      "step": 3050
    },
    {
      "epoch": 0.6705494505494506,
      "grad_norm": 0.27129946221444673,
      "learning_rate": 2.2418417570584525e-07,
      "loss": 1.524,
      "step": 3051
    },
    {
      "epoch": 0.6707692307692308,
      "grad_norm": 0.25371149060523107,
      "learning_rate": 2.2399719846109736e-07,
      "loss": 1.5097,
      "step": 3052
    },
    {
      "epoch": 0.670989010989011,
      "grad_norm": 0.25645313108582296,
      "learning_rate": 2.238102979720019e-07,
      "loss": 1.4682,
      "step": 3053
    },
    {
      "epoch": 0.6712087912087912,
      "grad_norm": 0.25008491242513803,
      "learning_rate": 2.2362347432766087e-07,
      "loss": 1.4428,
      "step": 3054
    },
    {
      "epoch": 0.6714285714285714,
      "grad_norm": 0.24918179303638485,
      "learning_rate": 2.2343672761713958e-07,
      "loss": 1.4284,
      "step": 3055
    },
    {
      "epoch": 0.6716483516483517,
      "grad_norm": 0.2641300455781301,
      "learning_rate": 2.2325005792946676e-07,
      "loss": 1.4456,
      "step": 3056
    },
    {
      "epoch": 0.6718681318681319,
      "grad_norm": 0.25355982281227657,
      "learning_rate": 2.2306346535363454e-07,
      "loss": 1.4435,
      "step": 3057
    },
    {
      "epoch": 0.6720879120879121,
      "grad_norm": 0.23953670821996706,
      "learning_rate": 2.2287694997859788e-07,
      "loss": 1.4761,
      "step": 3058
    },
    {
      "epoch": 0.6723076923076923,
      "grad_norm": 0.26255369200191203,
      "learning_rate": 2.2269051189327533e-07,
      "loss": 1.4265,
      "step": 3059
    },
    {
      "epoch": 0.6725274725274726,
      "grad_norm": 0.30347224252128724,
      "learning_rate": 2.2250415118654814e-07,
      "loss": 1.4215,
      "step": 3060
    },
    {
      "epoch": 0.6727472527472528,
      "grad_norm": 0.23973036656762103,
      "learning_rate": 2.2231786794726128e-07,
      "loss": 1.4776,
      "step": 3061
    },
    {
      "epoch": 0.672967032967033,
      "grad_norm": 0.2696345520772091,
      "learning_rate": 2.2213166226422238e-07,
      "loss": 1.452,
      "step": 3062
    },
    {
      "epoch": 0.6731868131868132,
      "grad_norm": 0.2364206023120924,
      "learning_rate": 2.2194553422620203e-07,
      "loss": 1.4749,
      "step": 3063
    },
    {
      "epoch": 0.6734065934065934,
      "grad_norm": 0.24853105640171716,
      "learning_rate": 2.2175948392193416e-07,
      "loss": 1.4835,
      "step": 3064
    },
    {
      "epoch": 0.6736263736263737,
      "grad_norm": 0.2512307002156439,
      "learning_rate": 2.215735114401153e-07,
      "loss": 1.4624,
      "step": 3065
    },
    {
      "epoch": 0.6738461538461539,
      "grad_norm": 0.25526373880125336,
      "learning_rate": 2.2138761686940516e-07,
      "loss": 1.5062,
      "step": 3066
    },
    {
      "epoch": 0.6740659340659341,
      "grad_norm": 0.24535367297223465,
      "learning_rate": 2.212018002984261e-07,
      "loss": 1.4196,
      "step": 3067
    },
    {
      "epoch": 0.6742857142857143,
      "grad_norm": 0.2574168943162774,
      "learning_rate": 2.2101606181576338e-07,
      "loss": 1.4473,
      "step": 3068
    },
    {
      "epoch": 0.6745054945054945,
      "grad_norm": 0.26110707797614324,
      "learning_rate": 2.2083040150996495e-07,
      "loss": 1.534,
      "step": 3069
    },
    {
      "epoch": 0.6747252747252748,
      "grad_norm": 0.2525884240847623,
      "learning_rate": 2.206448194695415e-07,
      "loss": 1.4394,
      "step": 3070
    },
    {
      "epoch": 0.674945054945055,
      "grad_norm": 0.24358300659593868,
      "learning_rate": 2.2045931578296653e-07,
      "loss": 1.5019,
      "step": 3071
    },
    {
      "epoch": 0.6751648351648352,
      "grad_norm": 0.2434315489452012,
      "learning_rate": 2.2027389053867627e-07,
      "loss": 1.4753,
      "step": 3072
    },
    {
      "epoch": 0.6753846153846154,
      "grad_norm": 0.3123842926702313,
      "learning_rate": 2.2008854382506927e-07,
      "loss": 1.4943,
      "step": 3073
    },
    {
      "epoch": 0.6756043956043956,
      "grad_norm": 0.24783645902626378,
      "learning_rate": 2.199032757305068e-07,
      "loss": 1.4517,
      "step": 3074
    },
    {
      "epoch": 0.6758241758241759,
      "grad_norm": 0.2461502951028342,
      "learning_rate": 2.1971808634331246e-07,
      "loss": 1.4843,
      "step": 3075
    },
    {
      "epoch": 0.6760439560439561,
      "grad_norm": 0.24477843238407343,
      "learning_rate": 2.1953297575177276e-07,
      "loss": 1.5018,
      "step": 3076
    },
    {
      "epoch": 0.6762637362637363,
      "grad_norm": 0.24202176622759503,
      "learning_rate": 2.193479440441363e-07,
      "loss": 1.3793,
      "step": 3077
    },
    {
      "epoch": 0.6764835164835165,
      "grad_norm": 0.2553382699920082,
      "learning_rate": 2.1916299130861392e-07,
      "loss": 1.5,
      "step": 3078
    },
    {
      "epoch": 0.6767032967032967,
      "grad_norm": 0.2590305026210981,
      "learning_rate": 2.1897811763337939e-07,
      "loss": 1.49,
      "step": 3079
    },
    {
      "epoch": 0.676923076923077,
      "grad_norm": 0.26940706711649237,
      "learning_rate": 2.1879332310656815e-07,
      "loss": 1.5435,
      "step": 3080
    },
    {
      "epoch": 0.6771428571428572,
      "grad_norm": 0.23610551384248166,
      "learning_rate": 2.1860860781627842e-07,
      "loss": 1.4784,
      "step": 3081
    },
    {
      "epoch": 0.6773626373626374,
      "grad_norm": 0.25544121629166344,
      "learning_rate": 2.1842397185057037e-07,
      "loss": 1.4369,
      "step": 3082
    },
    {
      "epoch": 0.6775824175824176,
      "grad_norm": 0.25572194309840396,
      "learning_rate": 2.1823941529746632e-07,
      "loss": 1.4368,
      "step": 3083
    },
    {
      "epoch": 0.6778021978021977,
      "grad_norm": 0.2508399245102378,
      "learning_rate": 2.180549382449509e-07,
      "loss": 1.5001,
      "step": 3084
    },
    {
      "epoch": 0.6780219780219781,
      "grad_norm": 0.2445312473823318,
      "learning_rate": 2.1787054078097058e-07,
      "loss": 1.4513,
      "step": 3085
    },
    {
      "epoch": 0.6782417582417583,
      "grad_norm": 0.2506420509392925,
      "learning_rate": 2.1768622299343422e-07,
      "loss": 1.4933,
      "step": 3086
    },
    {
      "epoch": 0.6784615384615384,
      "grad_norm": 0.25456041414152586,
      "learning_rate": 2.1750198497021261e-07,
      "loss": 1.4729,
      "step": 3087
    },
    {
      "epoch": 0.6786813186813186,
      "grad_norm": 0.25097074076004683,
      "learning_rate": 2.1731782679913836e-07,
      "loss": 1.4494,
      "step": 3088
    },
    {
      "epoch": 0.678901098901099,
      "grad_norm": 0.24085724506774758,
      "learning_rate": 2.17133748568006e-07,
      "loss": 1.4898,
      "step": 3089
    },
    {
      "epoch": 0.6791208791208792,
      "grad_norm": 0.24075355484106975,
      "learning_rate": 2.1694975036457223e-07,
      "loss": 1.4266,
      "step": 3090
    },
    {
      "epoch": 0.6793406593406593,
      "grad_norm": 0.25466765182508455,
      "learning_rate": 2.167658322765553e-07,
      "loss": 1.4302,
      "step": 3091
    },
    {
      "epoch": 0.6795604395604395,
      "grad_norm": 0.24993986538291837,
      "learning_rate": 2.1658199439163543e-07,
      "loss": 1.4778,
      "step": 3092
    },
    {
      "epoch": 0.6797802197802197,
      "grad_norm": 0.24157887696162691,
      "learning_rate": 2.1639823679745443e-07,
      "loss": 1.4915,
      "step": 3093
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.25949850421442,
      "learning_rate": 2.16214559581616e-07,
      "loss": 1.459,
      "step": 3094
    },
    {
      "epoch": 0.6802197802197802,
      "grad_norm": 0.23773255396933846,
      "learning_rate": 2.1603096283168567e-07,
      "loss": 1.4366,
      "step": 3095
    },
    {
      "epoch": 0.6804395604395604,
      "grad_norm": 0.24714077246063373,
      "learning_rate": 2.1584744663519024e-07,
      "loss": 1.475,
      "step": 3096
    },
    {
      "epoch": 0.6806593406593406,
      "grad_norm": 0.24648925557271095,
      "learning_rate": 2.156640110796183e-07,
      "loss": 1.4609,
      "step": 3097
    },
    {
      "epoch": 0.6808791208791208,
      "grad_norm": 0.2594198583463061,
      "learning_rate": 2.1548065625242005e-07,
      "loss": 1.514,
      "step": 3098
    },
    {
      "epoch": 0.6810989010989011,
      "grad_norm": 0.29286364110074287,
      "learning_rate": 2.152973822410069e-07,
      "loss": 1.4118,
      "step": 3099
    },
    {
      "epoch": 0.6813186813186813,
      "grad_norm": 0.2423648161204909,
      "learning_rate": 2.1511418913275227e-07,
      "loss": 1.4327,
      "step": 3100
    },
    {
      "epoch": 0.6815384615384615,
      "grad_norm": 0.23993786329091552,
      "learning_rate": 2.1493107701499052e-07,
      "loss": 1.5077,
      "step": 3101
    },
    {
      "epoch": 0.6817582417582417,
      "grad_norm": 0.3411989665514392,
      "learning_rate": 2.1474804597501768e-07,
      "loss": 1.4289,
      "step": 3102
    },
    {
      "epoch": 0.6819780219780219,
      "grad_norm": 0.25536913747651535,
      "learning_rate": 2.1456509610009107e-07,
      "loss": 1.3766,
      "step": 3103
    },
    {
      "epoch": 0.6821978021978022,
      "grad_norm": 0.2532263164782981,
      "learning_rate": 2.14382227477429e-07,
      "loss": 1.4468,
      "step": 3104
    },
    {
      "epoch": 0.6824175824175824,
      "grad_norm": 0.24029046044235383,
      "learning_rate": 2.1419944019421166e-07,
      "loss": 1.5588,
      "step": 3105
    },
    {
      "epoch": 0.6826373626373626,
      "grad_norm": 0.23969928997050355,
      "learning_rate": 2.1401673433758002e-07,
      "loss": 1.4366,
      "step": 3106
    },
    {
      "epoch": 0.6828571428571428,
      "grad_norm": 0.2585854322539309,
      "learning_rate": 2.138341099946363e-07,
      "loss": 1.4755,
      "step": 3107
    },
    {
      "epoch": 0.683076923076923,
      "grad_norm": 0.2442193789475446,
      "learning_rate": 2.136515672524438e-07,
      "loss": 1.5113,
      "step": 3108
    },
    {
      "epoch": 0.6832967032967033,
      "grad_norm": 0.27456272290932193,
      "learning_rate": 2.1346910619802707e-07,
      "loss": 1.4706,
      "step": 3109
    },
    {
      "epoch": 0.6835164835164835,
      "grad_norm": 0.24814010870811984,
      "learning_rate": 2.1328672691837182e-07,
      "loss": 1.4417,
      "step": 3110
    },
    {
      "epoch": 0.6837362637362637,
      "grad_norm": 0.23617749416609635,
      "learning_rate": 2.1310442950042455e-07,
      "loss": 1.4359,
      "step": 3111
    },
    {
      "epoch": 0.6839560439560439,
      "grad_norm": 0.24831668058757458,
      "learning_rate": 2.1292221403109274e-07,
      "loss": 1.4452,
      "step": 3112
    },
    {
      "epoch": 0.6841758241758241,
      "grad_norm": 0.25205386687745585,
      "learning_rate": 2.127400805972448e-07,
      "loss": 1.457,
      "step": 3113
    },
    {
      "epoch": 0.6843956043956044,
      "grad_norm": 0.2577114555677272,
      "learning_rate": 2.1255802928571008e-07,
      "loss": 1.4333,
      "step": 3114
    },
    {
      "epoch": 0.6846153846153846,
      "grad_norm": 0.24382561172249276,
      "learning_rate": 2.1237606018327903e-07,
      "loss": 1.5009,
      "step": 3115
    },
    {
      "epoch": 0.6848351648351648,
      "grad_norm": 0.2847541174438912,
      "learning_rate": 2.1219417337670233e-07,
      "loss": 1.4895,
      "step": 3116
    },
    {
      "epoch": 0.685054945054945,
      "grad_norm": 0.27838387006735205,
      "learning_rate": 2.1201236895269212e-07,
      "loss": 1.4323,
      "step": 3117
    },
    {
      "epoch": 0.6852747252747253,
      "grad_norm": 0.2759133069945561,
      "learning_rate": 2.1183064699792062e-07,
      "loss": 1.5442,
      "step": 3118
    },
    {
      "epoch": 0.6854945054945055,
      "grad_norm": 0.2528145717777085,
      "learning_rate": 2.1164900759902124e-07,
      "loss": 1.3942,
      "step": 3119
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 0.25217428292549016,
      "learning_rate": 2.1146745084258773e-07,
      "loss": 1.4769,
      "step": 3120
    },
    {
      "epoch": 0.6859340659340659,
      "grad_norm": 0.24806848979131116,
      "learning_rate": 2.1128597681517453e-07,
      "loss": 1.4779,
      "step": 3121
    },
    {
      "epoch": 0.6861538461538461,
      "grad_norm": 0.25201626753038225,
      "learning_rate": 2.1110458560329664e-07,
      "loss": 1.4908,
      "step": 3122
    },
    {
      "epoch": 0.6863736263736264,
      "grad_norm": 0.34458576673155794,
      "learning_rate": 2.1092327729342943e-07,
      "loss": 1.486,
      "step": 3123
    },
    {
      "epoch": 0.6865934065934066,
      "grad_norm": 0.2519743257027312,
      "learning_rate": 2.1074205197200905e-07,
      "loss": 1.5106,
      "step": 3124
    },
    {
      "epoch": 0.6868131868131868,
      "grad_norm": 0.24061342136295233,
      "learning_rate": 2.105609097254321e-07,
      "loss": 1.4173,
      "step": 3125
    },
    {
      "epoch": 0.687032967032967,
      "grad_norm": 0.2434514127876388,
      "learning_rate": 2.1037985064005515e-07,
      "loss": 1.5073,
      "step": 3126
    },
    {
      "epoch": 0.6872527472527472,
      "grad_norm": 0.24435642631768628,
      "learning_rate": 2.101988748021955e-07,
      "loss": 1.4492,
      "step": 3127
    },
    {
      "epoch": 0.6874725274725275,
      "grad_norm": 0.24286132884219838,
      "learning_rate": 2.1001798229813049e-07,
      "loss": 1.4344,
      "step": 3128
    },
    {
      "epoch": 0.6876923076923077,
      "grad_norm": 0.26170499726834145,
      "learning_rate": 2.0983717321409808e-07,
      "loss": 1.4172,
      "step": 3129
    },
    {
      "epoch": 0.6879120879120879,
      "grad_norm": 0.23227327559356964,
      "learning_rate": 2.096564476362963e-07,
      "loss": 1.4432,
      "step": 3130
    },
    {
      "epoch": 0.6881318681318681,
      "grad_norm": 0.27840101443572146,
      "learning_rate": 2.09475805650883e-07,
      "loss": 1.3829,
      "step": 3131
    },
    {
      "epoch": 0.6883516483516483,
      "grad_norm": 0.24187472135404398,
      "learning_rate": 2.0929524734397695e-07,
      "loss": 1.4699,
      "step": 3132
    },
    {
      "epoch": 0.6885714285714286,
      "grad_norm": 0.2527146797146545,
      "learning_rate": 2.0911477280165625e-07,
      "loss": 1.4175,
      "step": 3133
    },
    {
      "epoch": 0.6887912087912088,
      "grad_norm": 0.2609213836624885,
      "learning_rate": 2.0893438210995972e-07,
      "loss": 1.4425,
      "step": 3134
    },
    {
      "epoch": 0.689010989010989,
      "grad_norm": 0.24496251146873171,
      "learning_rate": 2.0875407535488563e-07,
      "loss": 1.403,
      "step": 3135
    },
    {
      "epoch": 0.6892307692307692,
      "grad_norm": 0.250920865990223,
      "learning_rate": 2.0857385262239268e-07,
      "loss": 1.5082,
      "step": 3136
    },
    {
      "epoch": 0.6894505494505494,
      "grad_norm": 0.24564768858946562,
      "learning_rate": 2.0839371399839924e-07,
      "loss": 1.4635,
      "step": 3137
    },
    {
      "epoch": 0.6896703296703297,
      "grad_norm": 0.24187048441058168,
      "learning_rate": 2.0821365956878357e-07,
      "loss": 1.4062,
      "step": 3138
    },
    {
      "epoch": 0.6898901098901099,
      "grad_norm": 0.24735457834632849,
      "learning_rate": 2.0803368941938404e-07,
      "loss": 1.5286,
      "step": 3139
    },
    {
      "epoch": 0.6901098901098901,
      "grad_norm": 0.25833072189664436,
      "learning_rate": 2.0785380363599882e-07,
      "loss": 1.4721,
      "step": 3140
    },
    {
      "epoch": 0.6903296703296703,
      "grad_norm": 0.24399907985969946,
      "learning_rate": 2.0767400230438554e-07,
      "loss": 1.4639,
      "step": 3141
    },
    {
      "epoch": 0.6905494505494505,
      "grad_norm": 0.2517968074441635,
      "learning_rate": 2.0749428551026188e-07,
      "loss": 1.3812,
      "step": 3142
    },
    {
      "epoch": 0.6907692307692308,
      "grad_norm": 0.24466899540903284,
      "learning_rate": 2.0731465333930481e-07,
      "loss": 1.4336,
      "step": 3143
    },
    {
      "epoch": 0.690989010989011,
      "grad_norm": 0.2750572732307114,
      "learning_rate": 2.0713510587715174e-07,
      "loss": 1.4307,
      "step": 3144
    },
    {
      "epoch": 0.6912087912087912,
      "grad_norm": 0.2413244511686587,
      "learning_rate": 2.0695564320939895e-07,
      "loss": 1.5065,
      "step": 3145
    },
    {
      "epoch": 0.6914285714285714,
      "grad_norm": 0.24815369423796488,
      "learning_rate": 2.0677626542160242e-07,
      "loss": 1.4443,
      "step": 3146
    },
    {
      "epoch": 0.6916483516483517,
      "grad_norm": 0.2646080787831905,
      "learning_rate": 2.06596972599278e-07,
      "loss": 1.5057,
      "step": 3147
    },
    {
      "epoch": 0.6918681318681319,
      "grad_norm": 0.2425595954547546,
      "learning_rate": 2.0641776482790086e-07,
      "loss": 1.4369,
      "step": 3148
    },
    {
      "epoch": 0.6920879120879121,
      "grad_norm": 0.2527111061408441,
      "learning_rate": 2.0623864219290548e-07,
      "loss": 1.4684,
      "step": 3149
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 0.25041609236608553,
      "learning_rate": 2.0605960477968594e-07,
      "loss": 1.4906,
      "step": 3150
    },
    {
      "epoch": 0.6925274725274725,
      "grad_norm": 0.2485469344657923,
      "learning_rate": 2.0588065267359564e-07,
      "loss": 1.4239,
      "step": 3151
    },
    {
      "epoch": 0.6927472527472528,
      "grad_norm": 0.26629831155357364,
      "learning_rate": 2.0570178595994708e-07,
      "loss": 1.4996,
      "step": 3152
    },
    {
      "epoch": 0.692967032967033,
      "grad_norm": 0.23641471142457823,
      "learning_rate": 2.055230047240125e-07,
      "loss": 1.482,
      "step": 3153
    },
    {
      "epoch": 0.6931868131868132,
      "grad_norm": 0.3667053908025628,
      "learning_rate": 2.05344309051023e-07,
      "loss": 1.4338,
      "step": 3154
    },
    {
      "epoch": 0.6934065934065934,
      "grad_norm": 0.24496616092781123,
      "learning_rate": 2.0516569902616923e-07,
      "loss": 1.4402,
      "step": 3155
    },
    {
      "epoch": 0.6936263736263736,
      "grad_norm": 0.25579099163610536,
      "learning_rate": 2.049871747346007e-07,
      "loss": 1.5221,
      "step": 3156
    },
    {
      "epoch": 0.6938461538461539,
      "grad_norm": 0.24817376173997746,
      "learning_rate": 2.0480873626142614e-07,
      "loss": 1.428,
      "step": 3157
    },
    {
      "epoch": 0.6940659340659341,
      "grad_norm": 0.40343921679542905,
      "learning_rate": 2.0463038369171346e-07,
      "loss": 1.5188,
      "step": 3158
    },
    {
      "epoch": 0.6942857142857143,
      "grad_norm": 0.24999018687260025,
      "learning_rate": 2.044521171104895e-07,
      "loss": 1.4645,
      "step": 3159
    },
    {
      "epoch": 0.6945054945054945,
      "grad_norm": 0.24075434555395248,
      "learning_rate": 2.0427393660274025e-07,
      "loss": 1.5099,
      "step": 3160
    },
    {
      "epoch": 0.6947252747252747,
      "grad_norm": 0.2533563131813355,
      "learning_rate": 2.0409584225341038e-07,
      "loss": 1.4723,
      "step": 3161
    },
    {
      "epoch": 0.694945054945055,
      "grad_norm": 0.24333315308734801,
      "learning_rate": 2.0391783414740392e-07,
      "loss": 1.4673,
      "step": 3162
    },
    {
      "epoch": 0.6951648351648352,
      "grad_norm": 0.24180432171033167,
      "learning_rate": 2.0373991236958331e-07,
      "loss": 1.4257,
      "step": 3163
    },
    {
      "epoch": 0.6953846153846154,
      "grad_norm": 0.25739434860090893,
      "learning_rate": 2.035620770047703e-07,
      "loss": 1.4347,
      "step": 3164
    },
    {
      "epoch": 0.6956043956043956,
      "grad_norm": 0.3074683582796377,
      "learning_rate": 2.0338432813774507e-07,
      "loss": 1.4102,
      "step": 3165
    },
    {
      "epoch": 0.6958241758241758,
      "grad_norm": 0.23928778515501578,
      "learning_rate": 2.0320666585324678e-07,
      "loss": 1.4264,
      "step": 3166
    },
    {
      "epoch": 0.6960439560439561,
      "grad_norm": 0.2475993981417254,
      "learning_rate": 2.0302909023597305e-07,
      "loss": 1.4806,
      "step": 3167
    },
    {
      "epoch": 0.6962637362637363,
      "grad_norm": 0.23846983202090427,
      "learning_rate": 2.028516013705807e-07,
      "loss": 1.4911,
      "step": 3168
    },
    {
      "epoch": 0.6964835164835165,
      "grad_norm": 0.25884262787185724,
      "learning_rate": 2.0267419934168447e-07,
      "loss": 1.481,
      "step": 3169
    },
    {
      "epoch": 0.6967032967032967,
      "grad_norm": 0.33077262673832125,
      "learning_rate": 2.0249688423385854e-07,
      "loss": 1.4652,
      "step": 3170
    },
    {
      "epoch": 0.696923076923077,
      "grad_norm": 0.26993218056623297,
      "learning_rate": 2.023196561316348e-07,
      "loss": 1.422,
      "step": 3171
    },
    {
      "epoch": 0.6971428571428572,
      "grad_norm": 0.24238165964977457,
      "learning_rate": 2.0214251511950433e-07,
      "loss": 1.4782,
      "step": 3172
    },
    {
      "epoch": 0.6973626373626374,
      "grad_norm": 0.25281497035301775,
      "learning_rate": 2.0196546128191636e-07,
      "loss": 1.4714,
      "step": 3173
    },
    {
      "epoch": 0.6975824175824176,
      "grad_norm": 0.23806553817481058,
      "learning_rate": 2.017884947032787e-07,
      "loss": 1.4722,
      "step": 3174
    },
    {
      "epoch": 0.6978021978021978,
      "grad_norm": 0.2621306314371981,
      "learning_rate": 2.0161161546795733e-07,
      "loss": 1.5219,
      "step": 3175
    },
    {
      "epoch": 0.6980219780219781,
      "grad_norm": 0.2510491892136605,
      "learning_rate": 2.0143482366027676e-07,
      "loss": 1.5622,
      "step": 3176
    },
    {
      "epoch": 0.6982417582417583,
      "grad_norm": 0.24834781690274416,
      "learning_rate": 2.0125811936451996e-07,
      "loss": 1.4336,
      "step": 3177
    },
    {
      "epoch": 0.6984615384615385,
      "grad_norm": 0.2404310866256325,
      "learning_rate": 2.0108150266492785e-07,
      "loss": 1.3799,
      "step": 3178
    },
    {
      "epoch": 0.6986813186813187,
      "grad_norm": 0.24294667141443566,
      "learning_rate": 2.009049736457e-07,
      "loss": 1.4843,
      "step": 3179
    },
    {
      "epoch": 0.6989010989010989,
      "grad_norm": 0.24572665064557875,
      "learning_rate": 2.0072853239099384e-07,
      "loss": 1.4512,
      "step": 3180
    },
    {
      "epoch": 0.6991208791208792,
      "grad_norm": 0.2695810921788662,
      "learning_rate": 2.0055217898492498e-07,
      "loss": 1.484,
      "step": 3181
    },
    {
      "epoch": 0.6993406593406594,
      "grad_norm": 5.373659850927592,
      "learning_rate": 2.0037591351156742e-07,
      "loss": 1.4724,
      "step": 3182
    },
    {
      "epoch": 0.6995604395604396,
      "grad_norm": 0.24471537583157932,
      "learning_rate": 2.0019973605495293e-07,
      "loss": 1.4605,
      "step": 3183
    },
    {
      "epoch": 0.6997802197802198,
      "grad_norm": 0.2356772956889882,
      "learning_rate": 2.000236466990714e-07,
      "loss": 1.4751,
      "step": 3184
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2542023525404883,
      "learning_rate": 1.9984764552787103e-07,
      "loss": 1.4768,
      "step": 3185
    },
    {
      "epoch": 0.7002197802197803,
      "grad_norm": 0.2693191620754359,
      "learning_rate": 1.996717326252573e-07,
      "loss": 1.4816,
      "step": 3186
    },
    {
      "epoch": 0.7004395604395605,
      "grad_norm": 0.24633212397883253,
      "learning_rate": 1.9949590807509449e-07,
      "loss": 1.5305,
      "step": 3187
    },
    {
      "epoch": 0.7006593406593407,
      "grad_norm": 0.24945030856824707,
      "learning_rate": 1.9932017196120401e-07,
      "loss": 1.5288,
      "step": 3188
    },
    {
      "epoch": 0.7008791208791209,
      "grad_norm": 0.25189285712767506,
      "learning_rate": 1.9914452436736552e-07,
      "loss": 1.4602,
      "step": 3189
    },
    {
      "epoch": 0.701098901098901,
      "grad_norm": 0.2523811608389368,
      "learning_rate": 1.9896896537731632e-07,
      "loss": 1.3812,
      "step": 3190
    },
    {
      "epoch": 0.7013186813186814,
      "grad_norm": 0.24492554039973877,
      "learning_rate": 1.987934950747513e-07,
      "loss": 1.514,
      "step": 3191
    },
    {
      "epoch": 0.7015384615384616,
      "grad_norm": 0.7800093696725251,
      "learning_rate": 1.9861811354332364e-07,
      "loss": 1.4468,
      "step": 3192
    },
    {
      "epoch": 0.7017582417582418,
      "grad_norm": 0.2547839687640688,
      "learning_rate": 1.9844282086664354e-07,
      "loss": 1.4991,
      "step": 3193
    },
    {
      "epoch": 0.701978021978022,
      "grad_norm": 0.25480512592699334,
      "learning_rate": 1.9826761712827927e-07,
      "loss": 1.4161,
      "step": 3194
    },
    {
      "epoch": 0.7021978021978021,
      "grad_norm": 0.24873346798027937,
      "learning_rate": 1.9809250241175666e-07,
      "loss": 1.4332,
      "step": 3195
    },
    {
      "epoch": 0.7024175824175825,
      "grad_norm": 0.2536388255891338,
      "learning_rate": 1.9791747680055873e-07,
      "loss": 1.5154,
      "step": 3196
    },
    {
      "epoch": 0.7026373626373627,
      "grad_norm": 0.24990942350035344,
      "learning_rate": 1.9774254037812657e-07,
      "loss": 1.4742,
      "step": 3197
    },
    {
      "epoch": 0.7028571428571428,
      "grad_norm": 0.24481102545247088,
      "learning_rate": 1.9756769322785836e-07,
      "loss": 1.4048,
      "step": 3198
    },
    {
      "epoch": 0.703076923076923,
      "grad_norm": 0.26701590931205094,
      "learning_rate": 1.9739293543310976e-07,
      "loss": 1.4619,
      "step": 3199
    },
    {
      "epoch": 0.7032967032967034,
      "grad_norm": 0.24977058514541342,
      "learning_rate": 1.9721826707719403e-07,
      "loss": 1.4717,
      "step": 3200
    },
    {
      "epoch": 0.7035164835164835,
      "grad_norm": 0.2504927141253575,
      "learning_rate": 1.9704368824338156e-07,
      "loss": 1.5187,
      "step": 3201
    },
    {
      "epoch": 0.7037362637362637,
      "grad_norm": 0.24157986300437764,
      "learning_rate": 1.968691990149003e-07,
      "loss": 1.5005,
      "step": 3202
    },
    {
      "epoch": 0.7039560439560439,
      "grad_norm": 0.24422773100869952,
      "learning_rate": 1.9669479947493536e-07,
      "loss": 1.5058,
      "step": 3203
    },
    {
      "epoch": 0.7041758241758241,
      "grad_norm": 0.2636538207992709,
      "learning_rate": 1.96520489706629e-07,
      "loss": 1.5121,
      "step": 3204
    },
    {
      "epoch": 0.7043956043956044,
      "grad_norm": 0.2508846912155803,
      "learning_rate": 1.9634626979308063e-07,
      "loss": 1.5071,
      "step": 3205
    },
    {
      "epoch": 0.7046153846153846,
      "grad_norm": 0.24934544618780202,
      "learning_rate": 1.9617213981734727e-07,
      "loss": 1.4475,
      "step": 3206
    },
    {
      "epoch": 0.7048351648351648,
      "grad_norm": 0.2834563128653475,
      "learning_rate": 1.959980998624425e-07,
      "loss": 1.4617,
      "step": 3207
    },
    {
      "epoch": 0.705054945054945,
      "grad_norm": 0.24364443488014928,
      "learning_rate": 1.958241500113373e-07,
      "loss": 1.4504,
      "step": 3208
    },
    {
      "epoch": 0.7052747252747252,
      "grad_norm": 0.24498371324935125,
      "learning_rate": 1.9565029034695976e-07,
      "loss": 1.4348,
      "step": 3209
    },
    {
      "epoch": 0.7054945054945055,
      "grad_norm": 0.24224691125853023,
      "learning_rate": 1.9547652095219466e-07,
      "loss": 1.4163,
      "step": 3210
    },
    {
      "epoch": 0.7057142857142857,
      "grad_norm": 0.24619339270884996,
      "learning_rate": 1.953028419098841e-07,
      "loss": 1.4669,
      "step": 3211
    },
    {
      "epoch": 0.7059340659340659,
      "grad_norm": 0.2432722588425196,
      "learning_rate": 1.951292533028269e-07,
      "loss": 1.4541,
      "step": 3212
    },
    {
      "epoch": 0.7061538461538461,
      "grad_norm": 0.2469630623778641,
      "learning_rate": 1.9495575521377875e-07,
      "loss": 1.464,
      "step": 3213
    },
    {
      "epoch": 0.7063736263736263,
      "grad_norm": 0.25267432543083657,
      "learning_rate": 1.9478234772545232e-07,
      "loss": 1.4606,
      "step": 3214
    },
    {
      "epoch": 0.7065934065934066,
      "grad_norm": 0.2605139268164092,
      "learning_rate": 1.9460903092051678e-07,
      "loss": 1.5297,
      "step": 3215
    },
    {
      "epoch": 0.7068131868131868,
      "grad_norm": 0.24803353950597865,
      "learning_rate": 1.944358048815985e-07,
      "loss": 1.4619,
      "step": 3216
    },
    {
      "epoch": 0.707032967032967,
      "grad_norm": 0.25667653637217674,
      "learning_rate": 1.9426266969128044e-07,
      "loss": 1.4205,
      "step": 3217
    },
    {
      "epoch": 0.7072527472527472,
      "grad_norm": 0.23818064685961463,
      "learning_rate": 1.9408962543210208e-07,
      "loss": 1.4516,
      "step": 3218
    },
    {
      "epoch": 0.7074725274725274,
      "grad_norm": 0.25442353898695175,
      "learning_rate": 1.9391667218655967e-07,
      "loss": 1.4197,
      "step": 3219
    },
    {
      "epoch": 0.7076923076923077,
      "grad_norm": 0.24937386993306204,
      "learning_rate": 1.93743810037106e-07,
      "loss": 1.4257,
      "step": 3220
    },
    {
      "epoch": 0.7079120879120879,
      "grad_norm": 0.2691867333141872,
      "learning_rate": 1.9357103906615056e-07,
      "loss": 1.4567,
      "step": 3221
    },
    {
      "epoch": 0.7081318681318681,
      "grad_norm": 0.31630300982249876,
      "learning_rate": 1.9339835935605933e-07,
      "loss": 1.4658,
      "step": 3222
    },
    {
      "epoch": 0.7083516483516483,
      "grad_norm": 0.23962892426204127,
      "learning_rate": 1.932257709891546e-07,
      "loss": 1.4942,
      "step": 3223
    },
    {
      "epoch": 0.7085714285714285,
      "grad_norm": 0.41089038044383347,
      "learning_rate": 1.9305327404771547e-07,
      "loss": 1.4822,
      "step": 3224
    },
    {
      "epoch": 0.7087912087912088,
      "grad_norm": 0.24171709799232335,
      "learning_rate": 1.9288086861397704e-07,
      "loss": 1.4951,
      "step": 3225
    },
    {
      "epoch": 0.709010989010989,
      "grad_norm": 0.25245628249702773,
      "learning_rate": 1.9270855477013123e-07,
      "loss": 1.4492,
      "step": 3226
    },
    {
      "epoch": 0.7092307692307692,
      "grad_norm": 0.26284560030310655,
      "learning_rate": 1.925363325983259e-07,
      "loss": 1.4917,
      "step": 3227
    },
    {
      "epoch": 0.7094505494505494,
      "grad_norm": 0.2463516949941388,
      "learning_rate": 1.9236420218066538e-07,
      "loss": 1.4079,
      "step": 3228
    },
    {
      "epoch": 0.7096703296703297,
      "grad_norm": 0.263921226755463,
      "learning_rate": 1.9219216359921018e-07,
      "loss": 1.4824,
      "step": 3229
    },
    {
      "epoch": 0.7098901098901099,
      "grad_norm": 0.24651817675689766,
      "learning_rate": 1.9202021693597735e-07,
      "loss": 1.4845,
      "step": 3230
    },
    {
      "epoch": 0.7101098901098901,
      "grad_norm": 0.24910206911441665,
      "learning_rate": 1.9184836227293953e-07,
      "loss": 1.4456,
      "step": 3231
    },
    {
      "epoch": 0.7103296703296703,
      "grad_norm": 0.2505853015151859,
      "learning_rate": 1.9167659969202608e-07,
      "loss": 1.4469,
      "step": 3232
    },
    {
      "epoch": 0.7105494505494505,
      "grad_norm": 0.2544684201809747,
      "learning_rate": 1.9150492927512218e-07,
      "loss": 1.4046,
      "step": 3233
    },
    {
      "epoch": 0.7107692307692308,
      "grad_norm": 0.2465100264312349,
      "learning_rate": 1.9133335110406892e-07,
      "loss": 1.4689,
      "step": 3234
    },
    {
      "epoch": 0.710989010989011,
      "grad_norm": 0.2426853563471961,
      "learning_rate": 1.9116186526066384e-07,
      "loss": 1.5355,
      "step": 3235
    },
    {
      "epoch": 0.7112087912087912,
      "grad_norm": 0.24803171720248252,
      "learning_rate": 1.9099047182666014e-07,
      "loss": 1.4461,
      "step": 3236
    },
    {
      "epoch": 0.7114285714285714,
      "grad_norm": 0.2780107540933846,
      "learning_rate": 1.90819170883767e-07,
      "loss": 1.4616,
      "step": 3237
    },
    {
      "epoch": 0.7116483516483516,
      "grad_norm": 0.27470184243471857,
      "learning_rate": 1.9064796251364956e-07,
      "loss": 1.5052,
      "step": 3238
    },
    {
      "epoch": 0.7118681318681319,
      "grad_norm": 0.25169211517097295,
      "learning_rate": 1.904768467979288e-07,
      "loss": 1.4661,
      "step": 3239
    },
    {
      "epoch": 0.7120879120879121,
      "grad_norm": 0.25545898180429133,
      "learning_rate": 1.903058238181817e-07,
      "loss": 1.5073,
      "step": 3240
    },
    {
      "epoch": 0.7123076923076923,
      "grad_norm": 0.34123367359757256,
      "learning_rate": 1.9013489365594084e-07,
      "loss": 1.4718,
      "step": 3241
    },
    {
      "epoch": 0.7125274725274725,
      "grad_norm": 0.24916513219850814,
      "learning_rate": 1.8996405639269446e-07,
      "loss": 1.4685,
      "step": 3242
    },
    {
      "epoch": 0.7127472527472527,
      "grad_norm": 0.287545194200866,
      "learning_rate": 1.8979331210988678e-07,
      "loss": 1.4795,
      "step": 3243
    },
    {
      "epoch": 0.712967032967033,
      "grad_norm": 0.2412527323583216,
      "learning_rate": 1.8962266088891736e-07,
      "loss": 1.4452,
      "step": 3244
    },
    {
      "epoch": 0.7131868131868132,
      "grad_norm": 0.24399314333194297,
      "learning_rate": 1.8945210281114183e-07,
      "loss": 1.4066,
      "step": 3245
    },
    {
      "epoch": 0.7134065934065934,
      "grad_norm": 0.23808242277076305,
      "learning_rate": 1.8928163795787096e-07,
      "loss": 1.4539,
      "step": 3246
    },
    {
      "epoch": 0.7136263736263736,
      "grad_norm": 0.253134156732432,
      "learning_rate": 1.891112664103715e-07,
      "loss": 1.4865,
      "step": 3247
    },
    {
      "epoch": 0.7138461538461538,
      "grad_norm": 0.2461659703341645,
      "learning_rate": 1.8894098824986546e-07,
      "loss": 1.4467,
      "step": 3248
    },
    {
      "epoch": 0.7140659340659341,
      "grad_norm": 0.2689641006317081,
      "learning_rate": 1.887708035575302e-07,
      "loss": 1.4629,
      "step": 3249
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.23989033150090913,
      "learning_rate": 1.8860071241449896e-07,
      "loss": 1.4704,
      "step": 3250
    },
    {
      "epoch": 0.7145054945054945,
      "grad_norm": 0.26677787884863696,
      "learning_rate": 1.8843071490185996e-07,
      "loss": 1.4694,
      "step": 3251
    },
    {
      "epoch": 0.7147252747252747,
      "grad_norm": 0.23844986824915337,
      "learning_rate": 1.8826081110065702e-07,
      "loss": 1.4545,
      "step": 3252
    },
    {
      "epoch": 0.7149450549450549,
      "grad_norm": 0.31345042350401403,
      "learning_rate": 1.880910010918891e-07,
      "loss": 1.4333,
      "step": 3253
    },
    {
      "epoch": 0.7151648351648352,
      "grad_norm": 0.2555583725561702,
      "learning_rate": 1.8792128495651053e-07,
      "loss": 1.4538,
      "step": 3254
    },
    {
      "epoch": 0.7153846153846154,
      "grad_norm": 0.24811479341230933,
      "learning_rate": 1.877516627754312e-07,
      "loss": 1.5049,
      "step": 3255
    },
    {
      "epoch": 0.7156043956043956,
      "grad_norm": 0.24447189966855204,
      "learning_rate": 1.8758213462951568e-07,
      "loss": 1.487,
      "step": 3256
    },
    {
      "epoch": 0.7158241758241758,
      "grad_norm": 0.24154809271473499,
      "learning_rate": 1.87412700599584e-07,
      "loss": 1.5263,
      "step": 3257
    },
    {
      "epoch": 0.7160439560439561,
      "grad_norm": 0.24063646509027245,
      "learning_rate": 1.872433607664112e-07,
      "loss": 1.4076,
      "step": 3258
    },
    {
      "epoch": 0.7162637362637363,
      "grad_norm": 0.2498250801060971,
      "learning_rate": 1.8707411521072768e-07,
      "loss": 1.4191,
      "step": 3259
    },
    {
      "epoch": 0.7164835164835165,
      "grad_norm": 0.25350785794448677,
      "learning_rate": 1.8690496401321855e-07,
      "loss": 1.4553,
      "step": 3260
    },
    {
      "epoch": 0.7167032967032967,
      "grad_norm": 0.2407961693858166,
      "learning_rate": 1.8673590725452402e-07,
      "loss": 1.447,
      "step": 3261
    },
    {
      "epoch": 0.7169230769230769,
      "grad_norm": 0.24641495109494188,
      "learning_rate": 1.865669450152396e-07,
      "loss": 1.4332,
      "step": 3262
    },
    {
      "epoch": 0.7171428571428572,
      "grad_norm": 0.24588475335334106,
      "learning_rate": 1.8639807737591512e-07,
      "loss": 1.4299,
      "step": 3263
    },
    {
      "epoch": 0.7173626373626374,
      "grad_norm": 0.2624717335867807,
      "learning_rate": 1.8622930441705606e-07,
      "loss": 1.5054,
      "step": 3264
    },
    {
      "epoch": 0.7175824175824176,
      "grad_norm": 0.27990391708380047,
      "learning_rate": 1.860606262191222e-07,
      "loss": 1.461,
      "step": 3265
    },
    {
      "epoch": 0.7178021978021978,
      "grad_norm": 0.2595034808065735,
      "learning_rate": 1.8589204286252833e-07,
      "loss": 1.4493,
      "step": 3266
    },
    {
      "epoch": 0.718021978021978,
      "grad_norm": 0.24735307720367514,
      "learning_rate": 1.8572355442764399e-07,
      "loss": 1.4956,
      "step": 3267
    },
    {
      "epoch": 0.7182417582417583,
      "grad_norm": 0.2610167548710171,
      "learning_rate": 1.855551609947934e-07,
      "loss": 1.4545,
      "step": 3268
    },
    {
      "epoch": 0.7184615384615385,
      "grad_norm": 0.24621173602374435,
      "learning_rate": 1.8538686264425571e-07,
      "loss": 1.4619,
      "step": 3269
    },
    {
      "epoch": 0.7186813186813187,
      "grad_norm": 0.2736164815114475,
      "learning_rate": 1.852186594562647e-07,
      "loss": 1.4773,
      "step": 3270
    },
    {
      "epoch": 0.7189010989010989,
      "grad_norm": 0.25480669335083556,
      "learning_rate": 1.8505055151100872e-07,
      "loss": 1.5123,
      "step": 3271
    },
    {
      "epoch": 0.7191208791208791,
      "grad_norm": 0.24798224721324247,
      "learning_rate": 1.848825388886304e-07,
      "loss": 1.5256,
      "step": 3272
    },
    {
      "epoch": 0.7193406593406594,
      "grad_norm": 0.25733916159075726,
      "learning_rate": 1.847146216692274e-07,
      "loss": 1.4671,
      "step": 3273
    },
    {
      "epoch": 0.7195604395604396,
      "grad_norm": 0.24376604468862545,
      "learning_rate": 1.8454679993285176e-07,
      "loss": 1.4894,
      "step": 3274
    },
    {
      "epoch": 0.7197802197802198,
      "grad_norm": 0.24600997729651705,
      "learning_rate": 1.8437907375950997e-07,
      "loss": 1.5186,
      "step": 3275
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2488959774772532,
      "learning_rate": 1.8421144322916273e-07,
      "loss": 1.4553,
      "step": 3276
    },
    {
      "epoch": 0.7202197802197802,
      "grad_norm": 0.24938165662904727,
      "learning_rate": 1.8404390842172562e-07,
      "loss": 1.449,
      "step": 3277
    },
    {
      "epoch": 0.7204395604395605,
      "grad_norm": 0.32252047564284203,
      "learning_rate": 1.8387646941706818e-07,
      "loss": 1.4665,
      "step": 3278
    },
    {
      "epoch": 0.7206593406593407,
      "grad_norm": 0.2525303889820265,
      "learning_rate": 1.8370912629501456e-07,
      "loss": 1.432,
      "step": 3279
    },
    {
      "epoch": 0.7208791208791209,
      "grad_norm": 0.23499246121617587,
      "learning_rate": 1.835418791353429e-07,
      "loss": 1.3934,
      "step": 3280
    },
    {
      "epoch": 0.7210989010989011,
      "grad_norm": 0.24906913844441317,
      "learning_rate": 1.8337472801778592e-07,
      "loss": 1.4236,
      "step": 3281
    },
    {
      "epoch": 0.7213186813186813,
      "grad_norm": 0.24678602622920362,
      "learning_rate": 1.8320767302203028e-07,
      "loss": 1.4371,
      "step": 3282
    },
    {
      "epoch": 0.7215384615384616,
      "grad_norm": 0.3113074610143378,
      "learning_rate": 1.8304071422771679e-07,
      "loss": 1.4679,
      "step": 3283
    },
    {
      "epoch": 0.7217582417582418,
      "grad_norm": 0.26082664228768687,
      "learning_rate": 1.828738517144407e-07,
      "loss": 1.4577,
      "step": 3284
    },
    {
      "epoch": 0.721978021978022,
      "grad_norm": 0.24495495503093082,
      "learning_rate": 1.8270708556175134e-07,
      "loss": 1.4281,
      "step": 3285
    },
    {
      "epoch": 0.7221978021978022,
      "grad_norm": 0.2582850342960555,
      "learning_rate": 1.8254041584915174e-07,
      "loss": 1.4721,
      "step": 3286
    },
    {
      "epoch": 0.7224175824175825,
      "grad_norm": 0.241793919075375,
      "learning_rate": 1.8237384265609912e-07,
      "loss": 1.4833,
      "step": 3287
    },
    {
      "epoch": 0.7226373626373627,
      "grad_norm": 0.2501095050095039,
      "learning_rate": 1.8220736606200497e-07,
      "loss": 1.4916,
      "step": 3288
    },
    {
      "epoch": 0.7228571428571429,
      "grad_norm": 0.26555359339606643,
      "learning_rate": 1.8204098614623432e-07,
      "loss": 1.4344,
      "step": 3289
    },
    {
      "epoch": 0.7230769230769231,
      "grad_norm": 0.24798733506828943,
      "learning_rate": 1.8187470298810632e-07,
      "loss": 1.4475,
      "step": 3290
    },
    {
      "epoch": 0.7232967032967033,
      "grad_norm": 0.25674027566194535,
      "learning_rate": 1.8170851666689384e-07,
      "loss": 1.4907,
      "step": 3291
    },
    {
      "epoch": 0.7235164835164836,
      "grad_norm": 0.25240289939628696,
      "learning_rate": 1.8154242726182383e-07,
      "loss": 1.4166,
      "step": 3292
    },
    {
      "epoch": 0.7237362637362638,
      "grad_norm": 0.25343258434035865,
      "learning_rate": 1.8137643485207693e-07,
      "loss": 1.4977,
      "step": 3293
    },
    {
      "epoch": 0.723956043956044,
      "grad_norm": 0.25973109361538593,
      "learning_rate": 1.8121053951678747e-07,
      "loss": 1.4771,
      "step": 3294
    },
    {
      "epoch": 0.7241758241758242,
      "grad_norm": 0.24549839642838567,
      "learning_rate": 1.8104474133504357e-07,
      "loss": 1.505,
      "step": 3295
    },
    {
      "epoch": 0.7243956043956044,
      "grad_norm": 0.2621369442543476,
      "learning_rate": 1.8087904038588693e-07,
      "loss": 1.4428,
      "step": 3296
    },
    {
      "epoch": 0.7246153846153847,
      "grad_norm": 0.24132826419125739,
      "learning_rate": 1.8071343674831296e-07,
      "loss": 1.4258,
      "step": 3297
    },
    {
      "epoch": 0.7248351648351649,
      "grad_norm": 0.24003552269956913,
      "learning_rate": 1.8054793050127078e-07,
      "loss": 1.4788,
      "step": 3298
    },
    {
      "epoch": 0.725054945054945,
      "grad_norm": 0.2544725057915693,
      "learning_rate": 1.8038252172366285e-07,
      "loss": 1.4748,
      "step": 3299
    },
    {
      "epoch": 0.7252747252747253,
      "grad_norm": 0.25386484420295485,
      "learning_rate": 1.8021721049434555e-07,
      "loss": 1.4935,
      "step": 3300
    },
    {
      "epoch": 0.7254945054945054,
      "grad_norm": 0.2435757370023608,
      "learning_rate": 1.800519968921284e-07,
      "loss": 1.4653,
      "step": 3301
    },
    {
      "epoch": 0.7257142857142858,
      "grad_norm": 0.27279085070738307,
      "learning_rate": 1.798868809957743e-07,
      "loss": 1.4265,
      "step": 3302
    },
    {
      "epoch": 0.725934065934066,
      "grad_norm": 0.26117927967675364,
      "learning_rate": 1.797218628840001e-07,
      "loss": 1.4879,
      "step": 3303
    },
    {
      "epoch": 0.7261538461538461,
      "grad_norm": 0.2473784528154045,
      "learning_rate": 1.7955694263547543e-07,
      "loss": 1.5087,
      "step": 3304
    },
    {
      "epoch": 0.7263736263736263,
      "grad_norm": 0.2681737124631539,
      "learning_rate": 1.7939212032882368e-07,
      "loss": 1.468,
      "step": 3305
    },
    {
      "epoch": 0.7265934065934065,
      "grad_norm": 0.2527039926528341,
      "learning_rate": 1.7922739604262114e-07,
      "loss": 1.4972,
      "step": 3306
    },
    {
      "epoch": 0.7268131868131869,
      "grad_norm": 0.2623933772391432,
      "learning_rate": 1.7906276985539777e-07,
      "loss": 1.4191,
      "step": 3307
    },
    {
      "epoch": 0.727032967032967,
      "grad_norm": 0.24264095108199527,
      "learning_rate": 1.7889824184563687e-07,
      "loss": 1.4532,
      "step": 3308
    },
    {
      "epoch": 0.7272527472527472,
      "grad_norm": 0.2585626294135146,
      "learning_rate": 1.787338120917744e-07,
      "loss": 1.5527,
      "step": 3309
    },
    {
      "epoch": 0.7274725274725274,
      "grad_norm": 0.2441638996498909,
      "learning_rate": 1.7856948067219985e-07,
      "loss": 1.4354,
      "step": 3310
    },
    {
      "epoch": 0.7276923076923076,
      "grad_norm": 0.24615369620341251,
      "learning_rate": 1.7840524766525566e-07,
      "loss": 1.4313,
      "step": 3311
    },
    {
      "epoch": 0.727912087912088,
      "grad_norm": 0.24637571327402003,
      "learning_rate": 1.7824111314923744e-07,
      "loss": 1.4676,
      "step": 3312
    },
    {
      "epoch": 0.7281318681318681,
      "grad_norm": 0.2741600516349602,
      "learning_rate": 1.7807707720239397e-07,
      "loss": 1.4246,
      "step": 3313
    },
    {
      "epoch": 0.7283516483516483,
      "grad_norm": 0.24482069265418338,
      "learning_rate": 1.7791313990292673e-07,
      "loss": 1.454,
      "step": 3314
    },
    {
      "epoch": 0.7285714285714285,
      "grad_norm": 0.25074974889991747,
      "learning_rate": 1.7774930132899053e-07,
      "loss": 1.4506,
      "step": 3315
    },
    {
      "epoch": 0.7287912087912088,
      "grad_norm": 0.2444664324215279,
      "learning_rate": 1.7758556155869278e-07,
      "loss": 1.4875,
      "step": 3316
    },
    {
      "epoch": 0.729010989010989,
      "grad_norm": 0.2477842161344449,
      "learning_rate": 1.7742192067009405e-07,
      "loss": 1.4896,
      "step": 3317
    },
    {
      "epoch": 0.7292307692307692,
      "grad_norm": 0.2554882555390125,
      "learning_rate": 1.772583787412077e-07,
      "loss": 1.4367,
      "step": 3318
    },
    {
      "epoch": 0.7294505494505494,
      "grad_norm": 0.2588058927240242,
      "learning_rate": 1.770949358499998e-07,
      "loss": 1.4709,
      "step": 3319
    },
    {
      "epoch": 0.7296703296703296,
      "grad_norm": 0.24129779231256632,
      "learning_rate": 1.769315920743892e-07,
      "loss": 1.4206,
      "step": 3320
    },
    {
      "epoch": 0.7298901098901099,
      "grad_norm": 0.23705292466120687,
      "learning_rate": 1.7676834749224764e-07,
      "loss": 1.4891,
      "step": 3321
    },
    {
      "epoch": 0.7301098901098901,
      "grad_norm": 0.24185560692826077,
      "learning_rate": 1.7660520218139944e-07,
      "loss": 1.483,
      "step": 3322
    },
    {
      "epoch": 0.7303296703296703,
      "grad_norm": 0.2402788640388191,
      "learning_rate": 1.7644215621962188e-07,
      "loss": 1.4908,
      "step": 3323
    },
    {
      "epoch": 0.7305494505494505,
      "grad_norm": 0.24367930657678058,
      "learning_rate": 1.7627920968464443e-07,
      "loss": 1.393,
      "step": 3324
    },
    {
      "epoch": 0.7307692307692307,
      "grad_norm": 0.29088630455111597,
      "learning_rate": 1.761163626541496e-07,
      "loss": 1.4339,
      "step": 3325
    },
    {
      "epoch": 0.730989010989011,
      "grad_norm": 0.24828865102198766,
      "learning_rate": 1.7595361520577196e-07,
      "loss": 1.5337,
      "step": 3326
    },
    {
      "epoch": 0.7312087912087912,
      "grad_norm": 0.23701206977489214,
      "learning_rate": 1.7579096741709917e-07,
      "loss": 1.4834,
      "step": 3327
    },
    {
      "epoch": 0.7314285714285714,
      "grad_norm": 0.25408991367745354,
      "learning_rate": 1.75628419365671e-07,
      "loss": 1.5045,
      "step": 3328
    },
    {
      "epoch": 0.7316483516483516,
      "grad_norm": 0.3536005652886286,
      "learning_rate": 1.7546597112897966e-07,
      "loss": 1.4548,
      "step": 3329
    },
    {
      "epoch": 0.7318681318681318,
      "grad_norm": 0.2491959389437149,
      "learning_rate": 1.7530362278447014e-07,
      "loss": 1.4917,
      "step": 3330
    },
    {
      "epoch": 0.7320879120879121,
      "grad_norm": 0.2516841603708008,
      "learning_rate": 1.7514137440953927e-07,
      "loss": 1.4304,
      "step": 3331
    },
    {
      "epoch": 0.7323076923076923,
      "grad_norm": 0.3148516129988252,
      "learning_rate": 1.7497922608153681e-07,
      "loss": 1.5283,
      "step": 3332
    },
    {
      "epoch": 0.7325274725274725,
      "grad_norm": 0.24297610594375227,
      "learning_rate": 1.7481717787776442e-07,
      "loss": 1.4926,
      "step": 3333
    },
    {
      "epoch": 0.7327472527472527,
      "grad_norm": 0.24651349476658602,
      "learning_rate": 1.74655229875476e-07,
      "loss": 1.4798,
      "step": 3334
    },
    {
      "epoch": 0.7329670329670329,
      "grad_norm": 0.24363110015164513,
      "learning_rate": 1.74493382151878e-07,
      "loss": 1.4844,
      "step": 3335
    },
    {
      "epoch": 0.7331868131868132,
      "grad_norm": 0.24571049245301715,
      "learning_rate": 1.743316347841286e-07,
      "loss": 1.4756,
      "step": 3336
    },
    {
      "epoch": 0.7334065934065934,
      "grad_norm": 0.30127997312691257,
      "learning_rate": 1.7416998784933863e-07,
      "loss": 1.4834,
      "step": 3337
    },
    {
      "epoch": 0.7336263736263736,
      "grad_norm": 0.24090887689107635,
      "learning_rate": 1.7400844142457091e-07,
      "loss": 1.457,
      "step": 3338
    },
    {
      "epoch": 0.7338461538461538,
      "grad_norm": 0.2506063848987723,
      "learning_rate": 1.738469955868401e-07,
      "loss": 1.462,
      "step": 3339
    },
    {
      "epoch": 0.734065934065934,
      "grad_norm": 0.26223452210156184,
      "learning_rate": 1.7368565041311317e-07,
      "loss": 1.5086,
      "step": 3340
    },
    {
      "epoch": 0.7342857142857143,
      "grad_norm": 0.2511184231142407,
      "learning_rate": 1.7352440598030876e-07,
      "loss": 1.5082,
      "step": 3341
    },
    {
      "epoch": 0.7345054945054945,
      "grad_norm": 0.2746536075154917,
      "learning_rate": 1.7336326236529798e-07,
      "loss": 1.4161,
      "step": 3342
    },
    {
      "epoch": 0.7347252747252747,
      "grad_norm": 0.2634029891295733,
      "learning_rate": 1.7320221964490353e-07,
      "loss": 1.4605,
      "step": 3343
    },
    {
      "epoch": 0.7349450549450549,
      "grad_norm": 0.232507129441169,
      "learning_rate": 1.730412778959e-07,
      "loss": 1.3648,
      "step": 3344
    },
    {
      "epoch": 0.7351648351648352,
      "grad_norm": 0.23945792218875195,
      "learning_rate": 1.7288043719501395e-07,
      "loss": 1.4142,
      "step": 3345
    },
    {
      "epoch": 0.7353846153846154,
      "grad_norm": 0.24304528671522965,
      "learning_rate": 1.72719697618924e-07,
      "loss": 1.448,
      "step": 3346
    },
    {
      "epoch": 0.7356043956043956,
      "grad_norm": 0.24245404420714622,
      "learning_rate": 1.7255905924426006e-07,
      "loss": 1.4324,
      "step": 3347
    },
    {
      "epoch": 0.7358241758241758,
      "grad_norm": 0.27729188749805217,
      "learning_rate": 1.7239852214760411e-07,
      "loss": 1.4664,
      "step": 3348
    },
    {
      "epoch": 0.736043956043956,
      "grad_norm": 0.251772724522178,
      "learning_rate": 1.7223808640548984e-07,
      "loss": 1.5219,
      "step": 3349
    },
    {
      "epoch": 0.7362637362637363,
      "grad_norm": 0.2583832288599042,
      "learning_rate": 1.7207775209440237e-07,
      "loss": 1.477,
      "step": 3350
    },
    {
      "epoch": 0.7364835164835165,
      "grad_norm": 0.2496458226077847,
      "learning_rate": 1.7191751929077894e-07,
      "loss": 1.4877,
      "step": 3351
    },
    {
      "epoch": 0.7367032967032967,
      "grad_norm": 0.2506660296488448,
      "learning_rate": 1.7175738807100788e-07,
      "loss": 1.4495,
      "step": 3352
    },
    {
      "epoch": 0.7369230769230769,
      "grad_norm": 0.2461831193132943,
      "learning_rate": 1.7159735851142952e-07,
      "loss": 1.4803,
      "step": 3353
    },
    {
      "epoch": 0.7371428571428571,
      "grad_norm": 0.3196415414113024,
      "learning_rate": 1.714374306883355e-07,
      "loss": 1.4885,
      "step": 3354
    },
    {
      "epoch": 0.7373626373626374,
      "grad_norm": 0.24597553253439683,
      "learning_rate": 1.7127760467796877e-07,
      "loss": 1.4544,
      "step": 3355
    },
    {
      "epoch": 0.7375824175824176,
      "grad_norm": 0.25955877802464666,
      "learning_rate": 1.7111788055652426e-07,
      "loss": 1.4241,
      "step": 3356
    },
    {
      "epoch": 0.7378021978021978,
      "grad_norm": 0.3085525455010409,
      "learning_rate": 1.7095825840014792e-07,
      "loss": 1.446,
      "step": 3357
    },
    {
      "epoch": 0.738021978021978,
      "grad_norm": 0.24592311041502127,
      "learning_rate": 1.7079873828493717e-07,
      "loss": 1.4611,
      "step": 3358
    },
    {
      "epoch": 0.7382417582417582,
      "grad_norm": 0.24841376538151683,
      "learning_rate": 1.7063932028694074e-07,
      "loss": 1.4753,
      "step": 3359
    },
    {
      "epoch": 0.7384615384615385,
      "grad_norm": 0.2713539897488017,
      "learning_rate": 1.7048000448215886e-07,
      "loss": 1.4949,
      "step": 3360
    },
    {
      "epoch": 0.7386813186813187,
      "grad_norm": 0.24746011975549406,
      "learning_rate": 1.70320790946543e-07,
      "loss": 1.5289,
      "step": 3361
    },
    {
      "epoch": 0.7389010989010989,
      "grad_norm": 0.23917557293883,
      "learning_rate": 1.701616797559957e-07,
      "loss": 1.4783,
      "step": 3362
    },
    {
      "epoch": 0.7391208791208791,
      "grad_norm": 0.235060805618452,
      "learning_rate": 1.7000267098637086e-07,
      "loss": 1.45,
      "step": 3363
    },
    {
      "epoch": 0.7393406593406593,
      "grad_norm": 0.2500362528853272,
      "learning_rate": 1.698437647134735e-07,
      "loss": 1.4598,
      "step": 3364
    },
    {
      "epoch": 0.7395604395604396,
      "grad_norm": 0.23903024188249605,
      "learning_rate": 1.696849610130596e-07,
      "loss": 1.4895,
      "step": 3365
    },
    {
      "epoch": 0.7397802197802198,
      "grad_norm": 0.2683642338537404,
      "learning_rate": 1.6952625996083677e-07,
      "loss": 1.5428,
      "step": 3366
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24651225838741586,
      "learning_rate": 1.6936766163246304e-07,
      "loss": 1.4814,
      "step": 3367
    },
    {
      "epoch": 0.7402197802197802,
      "grad_norm": 0.24757725955012502,
      "learning_rate": 1.69209166103548e-07,
      "loss": 1.4704,
      "step": 3368
    },
    {
      "epoch": 0.7404395604395604,
      "grad_norm": 0.24913547631859492,
      "learning_rate": 1.6905077344965188e-07,
      "loss": 1.4637,
      "step": 3369
    },
    {
      "epoch": 0.7406593406593407,
      "grad_norm": 0.24699608292536634,
      "learning_rate": 1.688924837462859e-07,
      "loss": 1.4833,
      "step": 3370
    },
    {
      "epoch": 0.7408791208791209,
      "grad_norm": 0.25031556402939964,
      "learning_rate": 1.6873429706891243e-07,
      "loss": 1.4265,
      "step": 3371
    },
    {
      "epoch": 0.7410989010989011,
      "grad_norm": 0.2472417163494106,
      "learning_rate": 1.6857621349294455e-07,
      "loss": 1.4265,
      "step": 3372
    },
    {
      "epoch": 0.7413186813186813,
      "grad_norm": 0.24054463349031055,
      "learning_rate": 1.684182330937462e-07,
      "loss": 1.4242,
      "step": 3373
    },
    {
      "epoch": 0.7415384615384616,
      "grad_norm": 0.25049026957787496,
      "learning_rate": 1.6826035594663202e-07,
      "loss": 1.4479,
      "step": 3374
    },
    {
      "epoch": 0.7417582417582418,
      "grad_norm": 0.3432479096196123,
      "learning_rate": 1.6810258212686767e-07,
      "loss": 1.4528,
      "step": 3375
    },
    {
      "epoch": 0.741978021978022,
      "grad_norm": 0.43450980079736995,
      "learning_rate": 1.6794491170966954e-07,
      "loss": 1.4056,
      "step": 3376
    },
    {
      "epoch": 0.7421978021978022,
      "grad_norm": 0.2524522000964982,
      "learning_rate": 1.6778734477020451e-07,
      "loss": 1.4629,
      "step": 3377
    },
    {
      "epoch": 0.7424175824175824,
      "grad_norm": 0.28158437297803574,
      "learning_rate": 1.6762988138359027e-07,
      "loss": 1.4661,
      "step": 3378
    },
    {
      "epoch": 0.7426373626373627,
      "grad_norm": 0.2438021232880787,
      "learning_rate": 1.6747252162489492e-07,
      "loss": 1.4857,
      "step": 3379
    },
    {
      "epoch": 0.7428571428571429,
      "grad_norm": 0.24487210679330199,
      "learning_rate": 1.6731526556913763e-07,
      "loss": 1.4632,
      "step": 3380
    },
    {
      "epoch": 0.7430769230769231,
      "grad_norm": 0.2375924533319818,
      "learning_rate": 1.6715811329128773e-07,
      "loss": 1.4816,
      "step": 3381
    },
    {
      "epoch": 0.7432967032967033,
      "grad_norm": 0.25019045466570355,
      "learning_rate": 1.6700106486626512e-07,
      "loss": 1.4484,
      "step": 3382
    },
    {
      "epoch": 0.7435164835164835,
      "grad_norm": 0.248879208187517,
      "learning_rate": 1.6684412036894041e-07,
      "loss": 1.4562,
      "step": 3383
    },
    {
      "epoch": 0.7437362637362638,
      "grad_norm": 0.2533580101247259,
      "learning_rate": 1.6668727987413432e-07,
      "loss": 1.4878,
      "step": 3384
    },
    {
      "epoch": 0.743956043956044,
      "grad_norm": 0.23959289705268724,
      "learning_rate": 1.6653054345661842e-07,
      "loss": 1.4892,
      "step": 3385
    },
    {
      "epoch": 0.7441758241758242,
      "grad_norm": 0.243596496979219,
      "learning_rate": 1.6637391119111432e-07,
      "loss": 1.4503,
      "step": 3386
    },
    {
      "epoch": 0.7443956043956044,
      "grad_norm": 0.25527158467697086,
      "learning_rate": 1.6621738315229406e-07,
      "loss": 1.49,
      "step": 3387
    },
    {
      "epoch": 0.7446153846153846,
      "grad_norm": 0.24111902811215588,
      "learning_rate": 1.6606095941478005e-07,
      "loss": 1.4166,
      "step": 3388
    },
    {
      "epoch": 0.7448351648351649,
      "grad_norm": 0.259317487676881,
      "learning_rate": 1.6590464005314483e-07,
      "loss": 1.4944,
      "step": 3389
    },
    {
      "epoch": 0.7450549450549451,
      "grad_norm": 5.822964868824899,
      "learning_rate": 1.6574842514191139e-07,
      "loss": 1.4883,
      "step": 3390
    },
    {
      "epoch": 0.7452747252747253,
      "grad_norm": 0.24580329806796744,
      "learning_rate": 1.655923147555529e-07,
      "loss": 1.4484,
      "step": 3391
    },
    {
      "epoch": 0.7454945054945055,
      "grad_norm": 0.31420614026945887,
      "learning_rate": 1.654363089684926e-07,
      "loss": 1.4186,
      "step": 3392
    },
    {
      "epoch": 0.7457142857142857,
      "grad_norm": 0.24336437785507056,
      "learning_rate": 1.6528040785510382e-07,
      "loss": 1.4657,
      "step": 3393
    },
    {
      "epoch": 0.745934065934066,
      "grad_norm": 0.24724684614368012,
      "learning_rate": 1.6512461148971e-07,
      "loss": 1.5491,
      "step": 3394
    },
    {
      "epoch": 0.7461538461538462,
      "grad_norm": 0.2586628625291868,
      "learning_rate": 1.6496891994658489e-07,
      "loss": 1.5383,
      "step": 3395
    },
    {
      "epoch": 0.7463736263736264,
      "grad_norm": 0.24906770603882725,
      "learning_rate": 1.6481333329995197e-07,
      "loss": 1.4988,
      "step": 3396
    },
    {
      "epoch": 0.7465934065934066,
      "grad_norm": 0.255615007808465,
      "learning_rate": 1.6465785162398475e-07,
      "loss": 1.5019,
      "step": 3397
    },
    {
      "epoch": 0.7468131868131868,
      "grad_norm": 0.2510150542694438,
      "learning_rate": 1.6450247499280692e-07,
      "loss": 1.4584,
      "step": 3398
    },
    {
      "epoch": 0.7470329670329671,
      "grad_norm": 0.25952766439093616,
      "learning_rate": 1.6434720348049183e-07,
      "loss": 1.4725,
      "step": 3399
    },
    {
      "epoch": 0.7472527472527473,
      "grad_norm": 0.2694863141977391,
      "learning_rate": 1.6419203716106294e-07,
      "loss": 1.4251,
      "step": 3400
    },
    {
      "epoch": 0.7474725274725275,
      "grad_norm": 0.27898197115846796,
      "learning_rate": 1.6403697610849337e-07,
      "loss": 1.4128,
      "step": 3401
    },
    {
      "epoch": 0.7476923076923077,
      "grad_norm": 0.3037927593183233,
      "learning_rate": 1.6388202039670614e-07,
      "loss": 1.4882,
      "step": 3402
    },
    {
      "epoch": 0.747912087912088,
      "grad_norm": 0.2493692644682538,
      "learning_rate": 1.6372717009957397e-07,
      "loss": 1.5281,
      "step": 3403
    },
    {
      "epoch": 0.7481318681318682,
      "grad_norm": 0.23686739433759432,
      "learning_rate": 1.635724252909196e-07,
      "loss": 1.5028,
      "step": 3404
    },
    {
      "epoch": 0.7483516483516484,
      "grad_norm": 0.2515016455629176,
      "learning_rate": 1.6341778604451498e-07,
      "loss": 1.4553,
      "step": 3405
    },
    {
      "epoch": 0.7485714285714286,
      "grad_norm": 0.25018838268203325,
      "learning_rate": 1.6326325243408235e-07,
      "loss": 1.477,
      "step": 3406
    },
    {
      "epoch": 0.7487912087912088,
      "grad_norm": 0.3274730094658662,
      "learning_rate": 1.6310882453329314e-07,
      "loss": 1.425,
      "step": 3407
    },
    {
      "epoch": 0.7490109890109891,
      "grad_norm": 0.24745487892889775,
      "learning_rate": 1.6295450241576836e-07,
      "loss": 1.4169,
      "step": 3408
    },
    {
      "epoch": 0.7492307692307693,
      "grad_norm": 0.24659486656899832,
      "learning_rate": 1.6280028615507903e-07,
      "loss": 1.459,
      "step": 3409
    },
    {
      "epoch": 0.7494505494505495,
      "grad_norm": 0.23935687943994277,
      "learning_rate": 1.626461758247453e-07,
      "loss": 1.4627,
      "step": 3410
    },
    {
      "epoch": 0.7496703296703296,
      "grad_norm": 0.24158440391563418,
      "learning_rate": 1.6249217149823694e-07,
      "loss": 1.4688,
      "step": 3411
    },
    {
      "epoch": 0.7498901098901098,
      "grad_norm": 0.2563783065798018,
      "learning_rate": 1.6233827324897304e-07,
      "loss": 1.5046,
      "step": 3412
    },
    {
      "epoch": 0.7501098901098902,
      "grad_norm": 0.24403422210786213,
      "learning_rate": 1.621844811503225e-07,
      "loss": 1.5023,
      "step": 3413
    },
    {
      "epoch": 0.7503296703296704,
      "grad_norm": 0.2449530891458622,
      "learning_rate": 1.6203079527560321e-07,
      "loss": 1.4354,
      "step": 3414
    },
    {
      "epoch": 0.7505494505494505,
      "grad_norm": 0.24346123409489204,
      "learning_rate": 1.618772156980827e-07,
      "loss": 1.4633,
      "step": 3415
    },
    {
      "epoch": 0.7507692307692307,
      "grad_norm": 0.2586465298697483,
      "learning_rate": 1.6172374249097774e-07,
      "loss": 1.4764,
      "step": 3416
    },
    {
      "epoch": 0.7509890109890109,
      "grad_norm": 0.26093534928694234,
      "learning_rate": 1.6157037572745423e-07,
      "loss": 1.4304,
      "step": 3417
    },
    {
      "epoch": 0.7512087912087912,
      "grad_norm": 0.24764290586163387,
      "learning_rate": 1.6141711548062745e-07,
      "loss": 1.4488,
      "step": 3418
    },
    {
      "epoch": 0.7514285714285714,
      "grad_norm": 0.2540610271643636,
      "learning_rate": 1.6126396182356206e-07,
      "loss": 1.4131,
      "step": 3419
    },
    {
      "epoch": 0.7516483516483516,
      "grad_norm": 0.23576589677329232,
      "learning_rate": 1.6111091482927167e-07,
      "loss": 1.4519,
      "step": 3420
    },
    {
      "epoch": 0.7518681318681318,
      "grad_norm": 0.24835772431958147,
      "learning_rate": 1.6095797457071904e-07,
      "loss": 1.4788,
      "step": 3421
    },
    {
      "epoch": 0.752087912087912,
      "grad_norm": 0.2574199723783817,
      "learning_rate": 1.6080514112081625e-07,
      "loss": 1.4617,
      "step": 3422
    },
    {
      "epoch": 0.7523076923076923,
      "grad_norm": 0.2476977553732144,
      "learning_rate": 1.6065241455242424e-07,
      "loss": 1.5329,
      "step": 3423
    },
    {
      "epoch": 0.7525274725274725,
      "grad_norm": 0.2602043876638605,
      "learning_rate": 1.604997949383532e-07,
      "loss": 1.4821,
      "step": 3424
    },
    {
      "epoch": 0.7527472527472527,
      "grad_norm": 0.2760198955284463,
      "learning_rate": 1.6034728235136223e-07,
      "loss": 1.4769,
      "step": 3425
    },
    {
      "epoch": 0.7529670329670329,
      "grad_norm": 0.4087552429938026,
      "learning_rate": 1.601948768641593e-07,
      "loss": 1.4092,
      "step": 3426
    },
    {
      "epoch": 0.7531868131868132,
      "grad_norm": 0.24167683016679026,
      "learning_rate": 1.6004257854940138e-07,
      "loss": 1.4417,
      "step": 3427
    },
    {
      "epoch": 0.7534065934065934,
      "grad_norm": 0.2645530578878678,
      "learning_rate": 1.5989038747969457e-07,
      "loss": 1.4437,
      "step": 3428
    },
    {
      "epoch": 0.7536263736263736,
      "grad_norm": 0.2536499249107328,
      "learning_rate": 1.5973830372759342e-07,
      "loss": 1.5188,
      "step": 3429
    },
    {
      "epoch": 0.7538461538461538,
      "grad_norm": 0.25005127243745934,
      "learning_rate": 1.595863273656019e-07,
      "loss": 1.5286,
      "step": 3430
    },
    {
      "epoch": 0.754065934065934,
      "grad_norm": 0.250870371649544,
      "learning_rate": 1.5943445846617214e-07,
      "loss": 1.4469,
      "step": 3431
    },
    {
      "epoch": 0.7542857142857143,
      "grad_norm": 0.24240831220869535,
      "learning_rate": 1.5928269710170544e-07,
      "loss": 1.5119,
      "step": 3432
    },
    {
      "epoch": 0.7545054945054945,
      "grad_norm": 0.25008920693077885,
      "learning_rate": 1.5913104334455182e-07,
      "loss": 1.4287,
      "step": 3433
    },
    {
      "epoch": 0.7547252747252747,
      "grad_norm": 0.27701456778254857,
      "learning_rate": 1.5897949726700984e-07,
      "loss": 1.4594,
      "step": 3434
    },
    {
      "epoch": 0.7549450549450549,
      "grad_norm": 0.2636801355696161,
      "learning_rate": 1.5882805894132682e-07,
      "loss": 1.4497,
      "step": 3435
    },
    {
      "epoch": 0.7551648351648351,
      "grad_norm": 0.25439861863278274,
      "learning_rate": 1.586767284396986e-07,
      "loss": 1.4414,
      "step": 3436
    },
    {
      "epoch": 0.7553846153846154,
      "grad_norm": 0.24747572642399232,
      "learning_rate": 1.5852550583426973e-07,
      "loss": 1.4624,
      "step": 3437
    },
    {
      "epoch": 0.7556043956043956,
      "grad_norm": 0.3514579941733955,
      "learning_rate": 1.583743911971335e-07,
      "loss": 1.4549,
      "step": 3438
    },
    {
      "epoch": 0.7558241758241758,
      "grad_norm": 0.24662463724587547,
      "learning_rate": 1.5822338460033142e-07,
      "loss": 1.4864,
      "step": 3439
    },
    {
      "epoch": 0.756043956043956,
      "grad_norm": 0.2568050695942072,
      "learning_rate": 1.5807248611585354e-07,
      "loss": 1.4916,
      "step": 3440
    },
    {
      "epoch": 0.7562637362637362,
      "grad_norm": 0.26561380930110146,
      "learning_rate": 1.5792169581563847e-07,
      "loss": 1.467,
      "step": 3441
    },
    {
      "epoch": 0.7564835164835165,
      "grad_norm": 0.24657508351391255,
      "learning_rate": 1.5777101377157312e-07,
      "loss": 1.4409,
      "step": 3442
    },
    {
      "epoch": 0.7567032967032967,
      "grad_norm": 0.250822224928817,
      "learning_rate": 1.5762044005549306e-07,
      "loss": 1.4361,
      "step": 3443
    },
    {
      "epoch": 0.7569230769230769,
      "grad_norm": 0.2720373682723322,
      "learning_rate": 1.5746997473918184e-07,
      "loss": 1.4137,
      "step": 3444
    },
    {
      "epoch": 0.7571428571428571,
      "grad_norm": 0.2535506633968077,
      "learning_rate": 1.5731961789437168e-07,
      "loss": 1.5107,
      "step": 3445
    },
    {
      "epoch": 0.7573626373626373,
      "grad_norm": 0.24674063795003515,
      "learning_rate": 1.5716936959274292e-07,
      "loss": 1.5388,
      "step": 3446
    },
    {
      "epoch": 0.7575824175824176,
      "grad_norm": 0.35941569129434386,
      "learning_rate": 1.5701922990592402e-07,
      "loss": 1.4322,
      "step": 3447
    },
    {
      "epoch": 0.7578021978021978,
      "grad_norm": 0.24791786193462975,
      "learning_rate": 1.56869198905492e-07,
      "loss": 1.5379,
      "step": 3448
    },
    {
      "epoch": 0.758021978021978,
      "grad_norm": 0.25728322482594124,
      "learning_rate": 1.5671927666297175e-07,
      "loss": 1.4963,
      "step": 3449
    },
    {
      "epoch": 0.7582417582417582,
      "grad_norm": 0.2596196568363513,
      "learning_rate": 1.565694632498365e-07,
      "loss": 1.4646,
      "step": 3450
    },
    {
      "epoch": 0.7584615384615384,
      "grad_norm": 0.24793286961565922,
      "learning_rate": 1.564197587375074e-07,
      "loss": 1.497,
      "step": 3451
    },
    {
      "epoch": 0.7586813186813187,
      "grad_norm": 0.2345649016314522,
      "learning_rate": 1.5627016319735388e-07,
      "loss": 1.443,
      "step": 3452
    },
    {
      "epoch": 0.7589010989010989,
      "grad_norm": 0.2536264010182477,
      "learning_rate": 1.5612067670069351e-07,
      "loss": 1.4511,
      "step": 3453
    },
    {
      "epoch": 0.7591208791208791,
      "grad_norm": 0.2506823976400012,
      "learning_rate": 1.5597129931879156e-07,
      "loss": 1.5272,
      "step": 3454
    },
    {
      "epoch": 0.7593406593406593,
      "grad_norm": 0.2754415001707102,
      "learning_rate": 1.558220311228615e-07,
      "loss": 1.5006,
      "step": 3455
    },
    {
      "epoch": 0.7595604395604396,
      "grad_norm": 0.2389276474612393,
      "learning_rate": 1.5567287218406453e-07,
      "loss": 1.4615,
      "step": 3456
    },
    {
      "epoch": 0.7597802197802198,
      "grad_norm": 0.2571718798103615,
      "learning_rate": 1.5552382257351018e-07,
      "loss": 1.4553,
      "step": 3457
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24655091055769965,
      "learning_rate": 1.5537488236225542e-07,
      "loss": 1.4649,
      "step": 3458
    },
    {
      "epoch": 0.7602197802197802,
      "grad_norm": 0.24302876484845864,
      "learning_rate": 1.5522605162130522e-07,
      "loss": 1.4474,
      "step": 3459
    },
    {
      "epoch": 0.7604395604395604,
      "grad_norm": 0.3185694212030723,
      "learning_rate": 1.5507733042161257e-07,
      "loss": 1.4567,
      "step": 3460
    },
    {
      "epoch": 0.7606593406593407,
      "grad_norm": 0.2696515727530926,
      "learning_rate": 1.549287188340778e-07,
      "loss": 1.4754,
      "step": 3461
    },
    {
      "epoch": 0.7608791208791209,
      "grad_norm": 0.2673114015606836,
      "learning_rate": 1.547802169295495e-07,
      "loss": 1.5218,
      "step": 3462
    },
    {
      "epoch": 0.7610989010989011,
      "grad_norm": 0.23610970983581625,
      "learning_rate": 1.5463182477882358e-07,
      "loss": 1.4978,
      "step": 3463
    },
    {
      "epoch": 0.7613186813186813,
      "grad_norm": 0.2730698187830436,
      "learning_rate": 1.544835424526438e-07,
      "loss": 1.4913,
      "step": 3464
    },
    {
      "epoch": 0.7615384615384615,
      "grad_norm": 0.25254305724161885,
      "learning_rate": 1.5433537002170144e-07,
      "loss": 1.507,
      "step": 3465
    },
    {
      "epoch": 0.7617582417582418,
      "grad_norm": 0.25786668272947233,
      "learning_rate": 1.541873075566354e-07,
      "loss": 1.4795,
      "step": 3466
    },
    {
      "epoch": 0.761978021978022,
      "grad_norm": 0.2637700902156982,
      "learning_rate": 1.5403935512803235e-07,
      "loss": 1.5609,
      "step": 3467
    },
    {
      "epoch": 0.7621978021978022,
      "grad_norm": 0.25065676260386205,
      "learning_rate": 1.538915128064265e-07,
      "loss": 1.4574,
      "step": 3468
    },
    {
      "epoch": 0.7624175824175824,
      "grad_norm": 0.25594108668524185,
      "learning_rate": 1.5374378066229924e-07,
      "loss": 1.4698,
      "step": 3469
    },
    {
      "epoch": 0.7626373626373626,
      "grad_norm": 0.2610440771977634,
      "learning_rate": 1.5359615876607972e-07,
      "loss": 1.4879,
      "step": 3470
    },
    {
      "epoch": 0.7628571428571429,
      "grad_norm": 0.25432582156452654,
      "learning_rate": 1.534486471881443e-07,
      "loss": 1.4248,
      "step": 3471
    },
    {
      "epoch": 0.7630769230769231,
      "grad_norm": 0.26147745724715327,
      "learning_rate": 1.5330124599881714e-07,
      "loss": 1.4193,
      "step": 3472
    },
    {
      "epoch": 0.7632967032967033,
      "grad_norm": 0.2428130190593821,
      "learning_rate": 1.531539552683694e-07,
      "loss": 1.4906,
      "step": 3473
    },
    {
      "epoch": 0.7635164835164835,
      "grad_norm": 0.2728233192533336,
      "learning_rate": 1.5300677506701962e-07,
      "loss": 1.4934,
      "step": 3474
    },
    {
      "epoch": 0.7637362637362637,
      "grad_norm": 0.2518794128485732,
      "learning_rate": 1.5285970546493395e-07,
      "loss": 1.4364,
      "step": 3475
    },
    {
      "epoch": 0.763956043956044,
      "grad_norm": 0.24789489049670058,
      "learning_rate": 1.5271274653222537e-07,
      "loss": 1.5031,
      "step": 3476
    },
    {
      "epoch": 0.7641758241758242,
      "grad_norm": 0.2520873445398746,
      "learning_rate": 1.5256589833895455e-07,
      "loss": 1.4572,
      "step": 3477
    },
    {
      "epoch": 0.7643956043956044,
      "grad_norm": 0.25574820315312446,
      "learning_rate": 1.5241916095512905e-07,
      "loss": 1.5286,
      "step": 3478
    },
    {
      "epoch": 0.7646153846153846,
      "grad_norm": 0.2414238185517599,
      "learning_rate": 1.522725344507037e-07,
      "loss": 1.464,
      "step": 3479
    },
    {
      "epoch": 0.7648351648351648,
      "grad_norm": 0.24641018639956744,
      "learning_rate": 1.521260188955804e-07,
      "loss": 1.4514,
      "step": 3480
    },
    {
      "epoch": 0.7650549450549451,
      "grad_norm": 0.24799631724452226,
      "learning_rate": 1.5197961435960828e-07,
      "loss": 1.4701,
      "step": 3481
    },
    {
      "epoch": 0.7652747252747253,
      "grad_norm": 0.23412073254313034,
      "learning_rate": 1.5183332091258343e-07,
      "loss": 1.4642,
      "step": 3482
    },
    {
      "epoch": 0.7654945054945055,
      "grad_norm": 0.25836737209752714,
      "learning_rate": 1.5168713862424925e-07,
      "loss": 1.3991,
      "step": 3483
    },
    {
      "epoch": 0.7657142857142857,
      "grad_norm": 0.4351468390241667,
      "learning_rate": 1.5154106756429578e-07,
      "loss": 1.4956,
      "step": 3484
    },
    {
      "epoch": 0.765934065934066,
      "grad_norm": 0.2524831931161781,
      "learning_rate": 1.513951078023601e-07,
      "loss": 1.4345,
      "step": 3485
    },
    {
      "epoch": 0.7661538461538462,
      "grad_norm": 0.2675055175327928,
      "learning_rate": 1.5124925940802654e-07,
      "loss": 1.4711,
      "step": 3486
    },
    {
      "epoch": 0.7663736263736264,
      "grad_norm": 0.24495792608121375,
      "learning_rate": 1.51103522450826e-07,
      "loss": 1.4758,
      "step": 3487
    },
    {
      "epoch": 0.7665934065934066,
      "grad_norm": 0.2522590634894788,
      "learning_rate": 1.5095789700023637e-07,
      "loss": 1.4539,
      "step": 3488
    },
    {
      "epoch": 0.7668131868131868,
      "grad_norm": 0.25286682540893707,
      "learning_rate": 1.508123831256823e-07,
      "loss": 1.4253,
      "step": 3489
    },
    {
      "epoch": 0.7670329670329671,
      "grad_norm": 0.23925401562789175,
      "learning_rate": 1.5066698089653542e-07,
      "loss": 1.4544,
      "step": 3490
    },
    {
      "epoch": 0.7672527472527473,
      "grad_norm": 0.25340387447587676,
      "learning_rate": 1.5052169038211415e-07,
      "loss": 1.4535,
      "step": 3491
    },
    {
      "epoch": 0.7674725274725275,
      "grad_norm": 0.24347089266421007,
      "learning_rate": 1.503765116516834e-07,
      "loss": 1.4148,
      "step": 3492
    },
    {
      "epoch": 0.7676923076923077,
      "grad_norm": 0.23447059043429536,
      "learning_rate": 1.5023144477445505e-07,
      "loss": 1.452,
      "step": 3493
    },
    {
      "epoch": 0.7679120879120879,
      "grad_norm": 0.2514153456877283,
      "learning_rate": 1.500864898195874e-07,
      "loss": 1.4588,
      "step": 3494
    },
    {
      "epoch": 0.7681318681318682,
      "grad_norm": 0.2592288169580409,
      "learning_rate": 1.4994164685618558e-07,
      "loss": 1.4572,
      "step": 3495
    },
    {
      "epoch": 0.7683516483516484,
      "grad_norm": 0.25138327897053964,
      "learning_rate": 1.4979691595330134e-07,
      "loss": 1.4905,
      "step": 3496
    },
    {
      "epoch": 0.7685714285714286,
      "grad_norm": 0.24374688066574657,
      "learning_rate": 1.4965229717993287e-07,
      "loss": 1.3699,
      "step": 3497
    },
    {
      "epoch": 0.7687912087912088,
      "grad_norm": 0.23783679830155766,
      "learning_rate": 1.495077906050251e-07,
      "loss": 1.3991,
      "step": 3498
    },
    {
      "epoch": 0.769010989010989,
      "grad_norm": 0.2876140351160437,
      "learning_rate": 1.4936339629746926e-07,
      "loss": 1.4186,
      "step": 3499
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.2563523347998719,
      "learning_rate": 1.4921911432610313e-07,
      "loss": 1.5102,
      "step": 3500
    },
    {
      "epoch": 0.7694505494505495,
      "grad_norm": 0.24859372497912346,
      "learning_rate": 1.4907494475971111e-07,
      "loss": 1.4803,
      "step": 3501
    },
    {
      "epoch": 0.7696703296703297,
      "grad_norm": 0.2518190377848507,
      "learning_rate": 1.4893088766702374e-07,
      "loss": 1.4816,
      "step": 3502
    },
    {
      "epoch": 0.7698901098901099,
      "grad_norm": 0.2460201616848149,
      "learning_rate": 1.4878694311671812e-07,
      "loss": 1.4837,
      "step": 3503
    },
    {
      "epoch": 0.7701098901098901,
      "grad_norm": 0.3125759802208812,
      "learning_rate": 1.4864311117741748e-07,
      "loss": 1.4507,
      "step": 3504
    },
    {
      "epoch": 0.7703296703296704,
      "grad_norm": 0.23312905930707495,
      "learning_rate": 1.4849939191769167e-07,
      "loss": 1.505,
      "step": 3505
    },
    {
      "epoch": 0.7705494505494506,
      "grad_norm": 0.24369591311036734,
      "learning_rate": 1.4835578540605674e-07,
      "loss": 1.4383,
      "step": 3506
    },
    {
      "epoch": 0.7707692307692308,
      "grad_norm": 0.24574200737848517,
      "learning_rate": 1.4821229171097481e-07,
      "loss": 1.4348,
      "step": 3507
    },
    {
      "epoch": 0.770989010989011,
      "grad_norm": 0.25948030707170733,
      "learning_rate": 1.480689109008544e-07,
      "loss": 1.4927,
      "step": 3508
    },
    {
      "epoch": 0.7712087912087912,
      "grad_norm": 0.24646101665815814,
      "learning_rate": 1.4792564304405008e-07,
      "loss": 1.4658,
      "step": 3509
    },
    {
      "epoch": 0.7714285714285715,
      "grad_norm": 0.25183416888382154,
      "learning_rate": 1.4778248820886256e-07,
      "loss": 1.4892,
      "step": 3510
    },
    {
      "epoch": 0.7716483516483517,
      "grad_norm": 0.26214956967582087,
      "learning_rate": 1.4763944646353896e-07,
      "loss": 1.4403,
      "step": 3511
    },
    {
      "epoch": 0.7718681318681319,
      "grad_norm": 0.24312505162483267,
      "learning_rate": 1.4749651787627206e-07,
      "loss": 1.4669,
      "step": 3512
    },
    {
      "epoch": 0.772087912087912,
      "grad_norm": 0.27184280319665866,
      "learning_rate": 1.4735370251520106e-07,
      "loss": 1.4458,
      "step": 3513
    },
    {
      "epoch": 0.7723076923076924,
      "grad_norm": 0.2528196050262525,
      "learning_rate": 1.4721100044841088e-07,
      "loss": 1.4496,
      "step": 3514
    },
    {
      "epoch": 0.7725274725274726,
      "grad_norm": 0.24368904732271113,
      "learning_rate": 1.470684117439328e-07,
      "loss": 1.4656,
      "step": 3515
    },
    {
      "epoch": 0.7727472527472528,
      "grad_norm": 0.2545030683078165,
      "learning_rate": 1.469259364697436e-07,
      "loss": 1.4787,
      "step": 3516
    },
    {
      "epoch": 0.772967032967033,
      "grad_norm": 0.2505895633602117,
      "learning_rate": 1.4678357469376635e-07,
      "loss": 1.4548,
      "step": 3517
    },
    {
      "epoch": 0.7731868131868131,
      "grad_norm": 0.2467224646872098,
      "learning_rate": 1.466413264838698e-07,
      "loss": 1.4153,
      "step": 3518
    },
    {
      "epoch": 0.7734065934065935,
      "grad_norm": 0.25185570838648613,
      "learning_rate": 1.464991919078686e-07,
      "loss": 1.4617,
      "step": 3519
    },
    {
      "epoch": 0.7736263736263737,
      "grad_norm": 0.2534599780017951,
      "learning_rate": 1.463571710335233e-07,
      "loss": 1.4327,
      "step": 3520
    },
    {
      "epoch": 0.7738461538461539,
      "grad_norm": 0.25144705278300167,
      "learning_rate": 1.4621526392854032e-07,
      "loss": 1.5118,
      "step": 3521
    },
    {
      "epoch": 0.774065934065934,
      "grad_norm": 0.26670095797124216,
      "learning_rate": 1.4607347066057164e-07,
      "loss": 1.439,
      "step": 3522
    },
    {
      "epoch": 0.7742857142857142,
      "grad_norm": 0.2413791122321761,
      "learning_rate": 1.4593179129721504e-07,
      "loss": 1.478,
      "step": 3523
    },
    {
      "epoch": 0.7745054945054946,
      "grad_norm": 0.25088589122680605,
      "learning_rate": 1.4579022590601395e-07,
      "loss": 1.4302,
      "step": 3524
    },
    {
      "epoch": 0.7747252747252747,
      "grad_norm": 0.26053424455776625,
      "learning_rate": 1.456487745544577e-07,
      "loss": 1.4311,
      "step": 3525
    },
    {
      "epoch": 0.774945054945055,
      "grad_norm": 0.25485413495907105,
      "learning_rate": 1.45507437309981e-07,
      "loss": 1.5281,
      "step": 3526
    },
    {
      "epoch": 0.7751648351648351,
      "grad_norm": 0.2659615784811842,
      "learning_rate": 1.4536621423996413e-07,
      "loss": 1.4063,
      "step": 3527
    },
    {
      "epoch": 0.7753846153846153,
      "grad_norm": 0.23943128667617608,
      "learning_rate": 1.452251054117332e-07,
      "loss": 1.4649,
      "step": 3528
    },
    {
      "epoch": 0.7756043956043956,
      "grad_norm": 0.25526777049689364,
      "learning_rate": 1.4508411089255965e-07,
      "loss": 1.455,
      "step": 3529
    },
    {
      "epoch": 0.7758241758241758,
      "grad_norm": 0.2558591110689858,
      "learning_rate": 1.4494323074966062e-07,
      "loss": 1.4661,
      "step": 3530
    },
    {
      "epoch": 0.776043956043956,
      "grad_norm": 0.25419760736204466,
      "learning_rate": 1.4480246505019842e-07,
      "loss": 1.4577,
      "step": 3531
    },
    {
      "epoch": 0.7762637362637362,
      "grad_norm": 0.25903526768170637,
      "learning_rate": 1.4466181386128103e-07,
      "loss": 1.4744,
      "step": 3532
    },
    {
      "epoch": 0.7764835164835164,
      "grad_norm": 0.24828421408389714,
      "learning_rate": 1.4452127724996183e-07,
      "loss": 1.479,
      "step": 3533
    },
    {
      "epoch": 0.7767032967032967,
      "grad_norm": 0.26867652179704915,
      "learning_rate": 1.4438085528323935e-07,
      "loss": 1.5028,
      "step": 3534
    },
    {
      "epoch": 0.7769230769230769,
      "grad_norm": 0.25106634632110353,
      "learning_rate": 1.4424054802805773e-07,
      "loss": 1.4358,
      "step": 3535
    },
    {
      "epoch": 0.7771428571428571,
      "grad_norm": 0.25370969667004695,
      "learning_rate": 1.441003555513065e-07,
      "loss": 1.5298,
      "step": 3536
    },
    {
      "epoch": 0.7773626373626373,
      "grad_norm": 0.2598201877282118,
      "learning_rate": 1.4396027791982015e-07,
      "loss": 1.4975,
      "step": 3537
    },
    {
      "epoch": 0.7775824175824175,
      "grad_norm": 0.2553860475676706,
      "learning_rate": 1.438203152003786e-07,
      "loss": 1.5021,
      "step": 3538
    },
    {
      "epoch": 0.7778021978021978,
      "grad_norm": 0.45663122512373583,
      "learning_rate": 1.436804674597069e-07,
      "loss": 1.4772,
      "step": 3539
    },
    {
      "epoch": 0.778021978021978,
      "grad_norm": 0.2552533439140228,
      "learning_rate": 1.4354073476447547e-07,
      "loss": 1.4803,
      "step": 3540
    },
    {
      "epoch": 0.7782417582417582,
      "grad_norm": 0.2479704540195358,
      "learning_rate": 1.434011171812997e-07,
      "loss": 1.5289,
      "step": 3541
    },
    {
      "epoch": 0.7784615384615384,
      "grad_norm": 0.32542840344395213,
      "learning_rate": 1.4326161477674004e-07,
      "loss": 1.5266,
      "step": 3542
    },
    {
      "epoch": 0.7786813186813187,
      "grad_norm": 0.23565302383184014,
      "learning_rate": 1.431222276173023e-07,
      "loss": 1.4597,
      "step": 3543
    },
    {
      "epoch": 0.7789010989010989,
      "grad_norm": 0.3867720198894138,
      "learning_rate": 1.429829557694372e-07,
      "loss": 1.424,
      "step": 3544
    },
    {
      "epoch": 0.7791208791208791,
      "grad_norm": 0.2597633759561389,
      "learning_rate": 1.4284379929954057e-07,
      "loss": 1.4179,
      "step": 3545
    },
    {
      "epoch": 0.7793406593406593,
      "grad_norm": 0.30002084123346406,
      "learning_rate": 1.4270475827395293e-07,
      "loss": 1.4433,
      "step": 3546
    },
    {
      "epoch": 0.7795604395604395,
      "grad_norm": 0.25167569917621246,
      "learning_rate": 1.4256583275896016e-07,
      "loss": 1.4922,
      "step": 3547
    },
    {
      "epoch": 0.7797802197802198,
      "grad_norm": 0.24780269228434898,
      "learning_rate": 1.4242702282079278e-07,
      "loss": 1.4483,
      "step": 3548
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23782809847618502,
      "learning_rate": 1.4228832852562644e-07,
      "loss": 1.5192,
      "step": 3549
    },
    {
      "epoch": 0.7802197802197802,
      "grad_norm": 0.25015173000447616,
      "learning_rate": 1.421497499395814e-07,
      "loss": 1.4726,
      "step": 3550
    },
    {
      "epoch": 0.7804395604395604,
      "grad_norm": 0.25407649625995454,
      "learning_rate": 1.4201128712872312e-07,
      "loss": 1.4719,
      "step": 3551
    },
    {
      "epoch": 0.7806593406593406,
      "grad_norm": 0.24941603461830075,
      "learning_rate": 1.4187294015906157e-07,
      "loss": 1.5027,
      "step": 3552
    },
    {
      "epoch": 0.7808791208791209,
      "grad_norm": 0.23807140103975025,
      "learning_rate": 1.417347090965514e-07,
      "loss": 1.4064,
      "step": 3553
    },
    {
      "epoch": 0.7810989010989011,
      "grad_norm": 0.24191487472968665,
      "learning_rate": 1.4159659400709242e-07,
      "loss": 1.4771,
      "step": 3554
    },
    {
      "epoch": 0.7813186813186813,
      "grad_norm": 0.25806951471456535,
      "learning_rate": 1.414585949565289e-07,
      "loss": 1.4518,
      "step": 3555
    },
    {
      "epoch": 0.7815384615384615,
      "grad_norm": 0.26102687643833583,
      "learning_rate": 1.4132071201064965e-07,
      "loss": 1.4944,
      "step": 3556
    },
    {
      "epoch": 0.7817582417582417,
      "grad_norm": 0.23708380309095017,
      "learning_rate": 1.4118294523518827e-07,
      "loss": 1.4228,
      "step": 3557
    },
    {
      "epoch": 0.781978021978022,
      "grad_norm": 0.2471786128723923,
      "learning_rate": 1.4104529469582307e-07,
      "loss": 1.4799,
      "step": 3558
    },
    {
      "epoch": 0.7821978021978022,
      "grad_norm": 0.24500193059953368,
      "learning_rate": 1.4090776045817697e-07,
      "loss": 1.473,
      "step": 3559
    },
    {
      "epoch": 0.7824175824175824,
      "grad_norm": 0.25561153064545017,
      "learning_rate": 1.407703425878172e-07,
      "loss": 1.4672,
      "step": 3560
    },
    {
      "epoch": 0.7826373626373626,
      "grad_norm": 0.25072978364339477,
      "learning_rate": 1.406330411502557e-07,
      "loss": 1.3961,
      "step": 3561
    },
    {
      "epoch": 0.7828571428571428,
      "grad_norm": 0.2503499710724997,
      "learning_rate": 1.4049585621094887e-07,
      "loss": 1.4371,
      "step": 3562
    },
    {
      "epoch": 0.7830769230769231,
      "grad_norm": 0.25685892700179125,
      "learning_rate": 1.4035878783529743e-07,
      "loss": 1.5071,
      "step": 3563
    },
    {
      "epoch": 0.7832967032967033,
      "grad_norm": 0.24035617397967948,
      "learning_rate": 1.4022183608864683e-07,
      "loss": 1.462,
      "step": 3564
    },
    {
      "epoch": 0.7835164835164835,
      "grad_norm": 0.2558478125326631,
      "learning_rate": 1.4008500103628658e-07,
      "loss": 1.4597,
      "step": 3565
    },
    {
      "epoch": 0.7837362637362637,
      "grad_norm": 0.24857759745695338,
      "learning_rate": 1.3994828274345092e-07,
      "loss": 1.4438,
      "step": 3566
    },
    {
      "epoch": 0.7839560439560439,
      "grad_norm": 0.25543980960816365,
      "learning_rate": 1.3981168127531812e-07,
      "loss": 1.4608,
      "step": 3567
    },
    {
      "epoch": 0.7841758241758242,
      "grad_norm": 0.29150447172254046,
      "learning_rate": 1.3967519669701077e-07,
      "loss": 1.5015,
      "step": 3568
    },
    {
      "epoch": 0.7843956043956044,
      "grad_norm": 0.24935243330679485,
      "learning_rate": 1.3953882907359602e-07,
      "loss": 1.4433,
      "step": 3569
    },
    {
      "epoch": 0.7846153846153846,
      "grad_norm": 0.2514029301929969,
      "learning_rate": 1.3940257847008497e-07,
      "loss": 1.4389,
      "step": 3570
    },
    {
      "epoch": 0.7848351648351648,
      "grad_norm": 0.25925198028207674,
      "learning_rate": 1.392664449514331e-07,
      "loss": 1.4709,
      "step": 3571
    },
    {
      "epoch": 0.7850549450549451,
      "grad_norm": 0.2723531342337533,
      "learning_rate": 1.3913042858253984e-07,
      "loss": 1.448,
      "step": 3572
    },
    {
      "epoch": 0.7852747252747253,
      "grad_norm": 0.24850649632846272,
      "learning_rate": 1.3899452942824903e-07,
      "loss": 1.4978,
      "step": 3573
    },
    {
      "epoch": 0.7854945054945055,
      "grad_norm": 0.2689628321688233,
      "learning_rate": 1.3885874755334867e-07,
      "loss": 1.5149,
      "step": 3574
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 0.24957677048899574,
      "learning_rate": 1.3872308302257062e-07,
      "loss": 1.4822,
      "step": 3575
    },
    {
      "epoch": 0.7859340659340659,
      "grad_norm": 0.24824761818350033,
      "learning_rate": 1.385875359005909e-07,
      "loss": 1.4686,
      "step": 3576
    },
    {
      "epoch": 0.7861538461538462,
      "grad_norm": 0.24653686749157777,
      "learning_rate": 1.384521062520294e-07,
      "loss": 1.4865,
      "step": 3577
    },
    {
      "epoch": 0.7863736263736264,
      "grad_norm": 0.24079307159403177,
      "learning_rate": 1.3831679414145045e-07,
      "loss": 1.5198,
      "step": 3578
    },
    {
      "epoch": 0.7865934065934066,
      "grad_norm": 0.2635919590559695,
      "learning_rate": 1.3818159963336183e-07,
      "loss": 1.4742,
      "step": 3579
    },
    {
      "epoch": 0.7868131868131868,
      "grad_norm": 0.2751772003342524,
      "learning_rate": 1.3804652279221552e-07,
      "loss": 1.4842,
      "step": 3580
    },
    {
      "epoch": 0.787032967032967,
      "grad_norm": 0.2431755192493663,
      "learning_rate": 1.3791156368240744e-07,
      "loss": 1.4473,
      "step": 3581
    },
    {
      "epoch": 0.7872527472527473,
      "grad_norm": 0.25128470024797633,
      "learning_rate": 1.3777672236827717e-07,
      "loss": 1.5358,
      "step": 3582
    },
    {
      "epoch": 0.7874725274725275,
      "grad_norm": 0.25448077071732245,
      "learning_rate": 1.3764199891410846e-07,
      "loss": 1.4844,
      "step": 3583
    },
    {
      "epoch": 0.7876923076923077,
      "grad_norm": 1.0570616444023322,
      "learning_rate": 1.3750739338412858e-07,
      "loss": 1.4446,
      "step": 3584
    },
    {
      "epoch": 0.7879120879120879,
      "grad_norm": 0.23944048780271676,
      "learning_rate": 1.3737290584250865e-07,
      "loss": 1.464,
      "step": 3585
    },
    {
      "epoch": 0.7881318681318681,
      "grad_norm": 0.24227352066455082,
      "learning_rate": 1.372385363533636e-07,
      "loss": 1.4377,
      "step": 3586
    },
    {
      "epoch": 0.7883516483516484,
      "grad_norm": 0.2537578306005774,
      "learning_rate": 1.37104284980752e-07,
      "loss": 1.495,
      "step": 3587
    },
    {
      "epoch": 0.7885714285714286,
      "grad_norm": 0.2620923166657437,
      "learning_rate": 1.3697015178867622e-07,
      "loss": 1.4803,
      "step": 3588
    },
    {
      "epoch": 0.7887912087912088,
      "grad_norm": 0.24691711029696287,
      "learning_rate": 1.3683613684108231e-07,
      "loss": 1.4204,
      "step": 3589
    },
    {
      "epoch": 0.789010989010989,
      "grad_norm": 0.24829109602928742,
      "learning_rate": 1.3670224020185976e-07,
      "loss": 1.4059,
      "step": 3590
    },
    {
      "epoch": 0.7892307692307692,
      "grad_norm": 0.2591759361401097,
      "learning_rate": 1.3656846193484182e-07,
      "loss": 1.5086,
      "step": 3591
    },
    {
      "epoch": 0.7894505494505495,
      "grad_norm": 0.25291516492263455,
      "learning_rate": 1.3643480210380517e-07,
      "loss": 1.4726,
      "step": 3592
    },
    {
      "epoch": 0.7896703296703297,
      "grad_norm": 0.6249091491515365,
      "learning_rate": 1.3630126077247027e-07,
      "loss": 1.4475,
      "step": 3593
    },
    {
      "epoch": 0.7898901098901099,
      "grad_norm": 0.23734670216213488,
      "learning_rate": 1.361678380045008e-07,
      "loss": 1.487,
      "step": 3594
    },
    {
      "epoch": 0.7901098901098901,
      "grad_norm": 0.3959081821447881,
      "learning_rate": 1.3603453386350402e-07,
      "loss": 1.4392,
      "step": 3595
    },
    {
      "epoch": 0.7903296703296703,
      "grad_norm": 0.24372053188157605,
      "learning_rate": 1.3590134841303083e-07,
      "loss": 1.4699,
      "step": 3596
    },
    {
      "epoch": 0.7905494505494506,
      "grad_norm": 0.24807847698118313,
      "learning_rate": 1.357682817165752e-07,
      "loss": 1.4987,
      "step": 3597
    },
    {
      "epoch": 0.7907692307692308,
      "grad_norm": 0.2497931550368373,
      "learning_rate": 1.3563533383757485e-07,
      "loss": 1.4807,
      "step": 3598
    },
    {
      "epoch": 0.790989010989011,
      "grad_norm": 0.25284357241917227,
      "learning_rate": 1.3550250483941054e-07,
      "loss": 1.517,
      "step": 3599
    },
    {
      "epoch": 0.7912087912087912,
      "grad_norm": 0.42253741591653593,
      "learning_rate": 1.3536979478540658e-07,
      "loss": 1.4666,
      "step": 3600
    },
    {
      "epoch": 0.7914285714285715,
      "grad_norm": 0.2532715902024033,
      "learning_rate": 1.352372037388303e-07,
      "loss": 1.4652,
      "step": 3601
    },
    {
      "epoch": 0.7916483516483517,
      "grad_norm": 0.25926879674229014,
      "learning_rate": 1.3510473176289269e-07,
      "loss": 1.4912,
      "step": 3602
    },
    {
      "epoch": 0.7918681318681319,
      "grad_norm": 0.26097516478868577,
      "learning_rate": 1.3497237892074762e-07,
      "loss": 1.5148,
      "step": 3603
    },
    {
      "epoch": 0.7920879120879121,
      "grad_norm": 0.254543043371435,
      "learning_rate": 1.3484014527549244e-07,
      "loss": 1.5257,
      "step": 3604
    },
    {
      "epoch": 0.7923076923076923,
      "grad_norm": 0.2457397860178966,
      "learning_rate": 1.3470803089016746e-07,
      "loss": 1.4766,
      "step": 3605
    },
    {
      "epoch": 0.7925274725274726,
      "grad_norm": 0.4144715230446353,
      "learning_rate": 1.3457603582775611e-07,
      "loss": 1.5179,
      "step": 3606
    },
    {
      "epoch": 0.7927472527472528,
      "grad_norm": 0.247610218313042,
      "learning_rate": 1.3444416015118525e-07,
      "loss": 1.5228,
      "step": 3607
    },
    {
      "epoch": 0.792967032967033,
      "grad_norm": 0.25391350551724173,
      "learning_rate": 1.3431240392332454e-07,
      "loss": 1.5207,
      "step": 3608
    },
    {
      "epoch": 0.7931868131868132,
      "grad_norm": 0.25184527811361523,
      "learning_rate": 1.3418076720698675e-07,
      "loss": 1.4727,
      "step": 3609
    },
    {
      "epoch": 0.7934065934065934,
      "grad_norm": 0.24622976983362402,
      "learning_rate": 1.3404925006492758e-07,
      "loss": 1.4735,
      "step": 3610
    },
    {
      "epoch": 0.7936263736263737,
      "grad_norm": 0.25390713793587044,
      "learning_rate": 1.3391785255984596e-07,
      "loss": 1.4771,
      "step": 3611
    },
    {
      "epoch": 0.7938461538461539,
      "grad_norm": 0.38109727986093606,
      "learning_rate": 1.337865747543838e-07,
      "loss": 1.4558,
      "step": 3612
    },
    {
      "epoch": 0.7940659340659341,
      "grad_norm": 0.2641541808340066,
      "learning_rate": 1.336554167111256e-07,
      "loss": 1.4885,
      "step": 3613
    },
    {
      "epoch": 0.7942857142857143,
      "grad_norm": 0.27480749669000204,
      "learning_rate": 1.3352437849259907e-07,
      "loss": 1.4257,
      "step": 3614
    },
    {
      "epoch": 0.7945054945054945,
      "grad_norm": 0.2401052566258336,
      "learning_rate": 1.3339346016127466e-07,
      "loss": 1.4629,
      "step": 3615
    },
    {
      "epoch": 0.7947252747252748,
      "grad_norm": 0.2564716645167739,
      "learning_rate": 1.3326266177956567e-07,
      "loss": 1.3667,
      "step": 3616
    },
    {
      "epoch": 0.794945054945055,
      "grad_norm": 0.2644254892320189,
      "learning_rate": 1.331319834098284e-07,
      "loss": 1.4852,
      "step": 3617
    },
    {
      "epoch": 0.7951648351648352,
      "grad_norm": 0.23510699478496122,
      "learning_rate": 1.3300142511436154e-07,
      "loss": 1.4449,
      "step": 3618
    },
    {
      "epoch": 0.7953846153846154,
      "grad_norm": 0.24680782985585087,
      "learning_rate": 1.3287098695540706e-07,
      "loss": 1.4294,
      "step": 3619
    },
    {
      "epoch": 0.7956043956043956,
      "grad_norm": 0.23889386781148805,
      "learning_rate": 1.3274066899514926e-07,
      "loss": 1.47,
      "step": 3620
    },
    {
      "epoch": 0.7958241758241759,
      "grad_norm": 0.25622950788305665,
      "learning_rate": 1.3261047129571515e-07,
      "loss": 1.5286,
      "step": 3621
    },
    {
      "epoch": 0.7960439560439561,
      "grad_norm": 0.25262759088704845,
      "learning_rate": 1.3248039391917472e-07,
      "loss": 1.4826,
      "step": 3622
    },
    {
      "epoch": 0.7962637362637363,
      "grad_norm": 0.25194531805796505,
      "learning_rate": 1.323504369275403e-07,
      "loss": 1.4836,
      "step": 3623
    },
    {
      "epoch": 0.7964835164835165,
      "grad_norm": 0.25171370581313124,
      "learning_rate": 1.3222060038276684e-07,
      "loss": 1.4469,
      "step": 3624
    },
    {
      "epoch": 0.7967032967032966,
      "grad_norm": 0.24556699791763342,
      "learning_rate": 1.3209088434675204e-07,
      "loss": 1.4828,
      "step": 3625
    },
    {
      "epoch": 0.796923076923077,
      "grad_norm": 0.2521356125431271,
      "learning_rate": 1.3196128888133607e-07,
      "loss": 1.4993,
      "step": 3626
    },
    {
      "epoch": 0.7971428571428572,
      "grad_norm": 0.2553718391138483,
      "learning_rate": 1.3183181404830155e-07,
      "loss": 1.495,
      "step": 3627
    },
    {
      "epoch": 0.7973626373626373,
      "grad_norm": 0.25627697102082586,
      "learning_rate": 1.3170245990937372e-07,
      "loss": 1.4891,
      "step": 3628
    },
    {
      "epoch": 0.7975824175824175,
      "grad_norm": 0.25474200716456835,
      "learning_rate": 1.3157322652622018e-07,
      "loss": 1.451,
      "step": 3629
    },
    {
      "epoch": 0.7978021978021979,
      "grad_norm": 0.2780876430660836,
      "learning_rate": 1.3144411396045093e-07,
      "loss": 1.4847,
      "step": 3630
    },
    {
      "epoch": 0.798021978021978,
      "grad_norm": 0.24647331582578247,
      "learning_rate": 1.313151222736186e-07,
      "loss": 1.5069,
      "step": 3631
    },
    {
      "epoch": 0.7982417582417582,
      "grad_norm": 0.2439146931874856,
      "learning_rate": 1.3118625152721788e-07,
      "loss": 1.4613,
      "step": 3632
    },
    {
      "epoch": 0.7984615384615384,
      "grad_norm": 0.248085550020872,
      "learning_rate": 1.3105750178268592e-07,
      "loss": 1.4373,
      "step": 3633
    },
    {
      "epoch": 0.7986813186813186,
      "grad_norm": 0.25150351290048434,
      "learning_rate": 1.3092887310140235e-07,
      "loss": 1.4231,
      "step": 3634
    },
    {
      "epoch": 0.798901098901099,
      "grad_norm": 0.24013690468453455,
      "learning_rate": 1.3080036554468878e-07,
      "loss": 1.4962,
      "step": 3635
    },
    {
      "epoch": 0.7991208791208791,
      "grad_norm": 0.25278653321754047,
      "learning_rate": 1.3067197917380946e-07,
      "loss": 1.4942,
      "step": 3636
    },
    {
      "epoch": 0.7993406593406593,
      "grad_norm": 0.2569306306755292,
      "learning_rate": 1.3054371404997055e-07,
      "loss": 1.4595,
      "step": 3637
    },
    {
      "epoch": 0.7995604395604395,
      "grad_norm": 0.2573964718132754,
      "learning_rate": 1.3041557023432052e-07,
      "loss": 1.4758,
      "step": 3638
    },
    {
      "epoch": 0.7997802197802197,
      "grad_norm": 0.24776350556113183,
      "learning_rate": 1.3028754778794997e-07,
      "loss": 1.4398,
      "step": 3639
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.24989334557464338,
      "learning_rate": 1.301596467718916e-07,
      "loss": 1.4161,
      "step": 3640
    }
  ],
  "logging_steps": 1,
  "max_steps": 4550,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 910,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.6845111312777216e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}