{
  "best_metric": 0.963947057723999,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.03795246453816595,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00018976232269082974,
      "grad_norm": 0.3689703047275543,
      "learning_rate": 1e-05,
      "loss": 1.0351,
      "step": 1
    },
    {
      "epoch": 0.00018976232269082974,
      "eval_loss": 1.3198952674865723,
      "eval_runtime": 275.6593,
      "eval_samples_per_second": 32.199,
      "eval_steps_per_second": 8.05,
      "step": 1
    },
    {
      "epoch": 0.00037952464538165947,
      "grad_norm": 0.32817262411117554,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 2
    },
    {
      "epoch": 0.0005692869680724893,
      "grad_norm": 0.4041525423526764,
      "learning_rate": 3e-05,
      "loss": 1.1584,
      "step": 3
    },
    {
      "epoch": 0.0007590492907633189,
      "grad_norm": 0.3785117268562317,
      "learning_rate": 4e-05,
      "loss": 1.0783,
      "step": 4
    },
    {
      "epoch": 0.0009488116134541487,
      "grad_norm": 0.43445149064064026,
      "learning_rate": 5e-05,
      "loss": 1.1869,
      "step": 5
    },
    {
      "epoch": 0.0011385739361449785,
      "grad_norm": 0.3155576288700104,
      "learning_rate": 6e-05,
      "loss": 1.1816,
      "step": 6
    },
    {
      "epoch": 0.001328336258835808,
      "grad_norm": 0.26356881856918335,
      "learning_rate": 7e-05,
      "loss": 1.1151,
      "step": 7
    },
    {
      "epoch": 0.0015180985815266379,
      "grad_norm": 0.16498738527297974,
      "learning_rate": 8e-05,
      "loss": 1.21,
      "step": 8
    },
    {
      "epoch": 0.0017078609042174677,
      "grad_norm": 0.16361583769321442,
      "learning_rate": 9e-05,
      "loss": 1.1318,
      "step": 9
    },
    {
      "epoch": 0.0018976232269082975,
      "grad_norm": 0.15728989243507385,
      "learning_rate": 0.0001,
      "loss": 1.1683,
      "step": 10
    },
    {
      "epoch": 0.0020873855495991273,
      "grad_norm": 0.18534033000469208,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.1149,
      "step": 11
    },
    {
      "epoch": 0.002277147872289957,
      "grad_norm": 0.18808847665786743,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.094,
      "step": 12
    },
    {
      "epoch": 0.0024669101949807864,
      "grad_norm": 0.1968209147453308,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.0957,
      "step": 13
    },
    {
      "epoch": 0.002656672517671616,
      "grad_norm": 0.18778468668460846,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.1423,
      "step": 14
    },
    {
      "epoch": 0.002846434840362446,
      "grad_norm": 0.18522831797599792,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.1087,
      "step": 15
    },
    {
      "epoch": 0.0030361971630532758,
      "grad_norm": 0.1853436827659607,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.1391,
      "step": 16
    },
    {
      "epoch": 0.0032259594857441056,
      "grad_norm": 0.18637384474277496,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.1179,
      "step": 17
    },
    {
      "epoch": 0.0034157218084349353,
      "grad_norm": 0.18694156408309937,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.159,
      "step": 18
    },
    {
      "epoch": 0.003605484131125765,
      "grad_norm": 0.1744527667760849,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.1811,
      "step": 19
    },
    {
      "epoch": 0.003795246453816595,
      "grad_norm": 0.17337507009506226,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.1214,
      "step": 20
    },
    {
      "epoch": 0.003985008776507424,
      "grad_norm": 0.17523472011089325,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1186,
      "step": 21
    },
    {
      "epoch": 0.0041747710991982545,
      "grad_norm": 0.1785983443260193,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.1123,
      "step": 22
    },
    {
      "epoch": 0.004364533421889084,
      "grad_norm": 0.17681241035461426,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.0687,
      "step": 23
    },
    {
      "epoch": 0.004554295744579914,
      "grad_norm": 0.18389420211315155,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.1033,
      "step": 24
    },
    {
      "epoch": 0.0047440580672707434,
      "grad_norm": 0.18924522399902344,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.0684,
      "step": 25
    },
    {
      "epoch": 0.004933820389961573,
      "grad_norm": 0.17484529316425323,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.0511,
      "step": 26
    },
    {
      "epoch": 0.005123582712652403,
      "grad_norm": 0.17750877141952515,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.1304,
      "step": 27
    },
    {
      "epoch": 0.005313345035343232,
      "grad_norm": 0.178945392370224,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.137,
      "step": 28
    },
    {
      "epoch": 0.005503107358034063,
      "grad_norm": 0.1712048500776291,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0915,
      "step": 29
    },
    {
      "epoch": 0.005692869680724892,
      "grad_norm": 0.19296303391456604,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.1328,
      "step": 30
    },
    {
      "epoch": 0.005882632003415722,
      "grad_norm": 0.18090558052062988,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.0345,
      "step": 31
    },
    {
      "epoch": 0.0060723943261065515,
      "grad_norm": 0.1833214908838272,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.0776,
      "step": 32
    },
    {
      "epoch": 0.006262156648797381,
      "grad_norm": 0.17969094216823578,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.1061,
      "step": 33
    },
    {
      "epoch": 0.006451918971488211,
      "grad_norm": 0.18190151453018188,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.0681,
      "step": 34
    },
    {
      "epoch": 0.0066416812941790405,
      "grad_norm": 0.19453197717666626,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.0855,
      "step": 35
    },
    {
      "epoch": 0.006831443616869871,
      "grad_norm": 0.18432414531707764,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.0901,
      "step": 36
    },
    {
      "epoch": 0.0070212059395607,
      "grad_norm": 0.20541809499263763,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.1055,
      "step": 37
    },
    {
      "epoch": 0.00721096826225153,
      "grad_norm": 0.19959476590156555,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.102,
      "step": 38
    },
    {
      "epoch": 0.00740073058494236,
      "grad_norm": 0.1939975470304489,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.1024,
      "step": 39
    },
    {
      "epoch": 0.00759049290763319,
      "grad_norm": 0.19972407817840576,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.9908,
      "step": 40
    },
    {
      "epoch": 0.007780255230324019,
      "grad_norm": 0.2053292840719223,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.0316,
      "step": 41
    },
    {
      "epoch": 0.007970017553014849,
      "grad_norm": 0.21963410079479218,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.1153,
      "step": 42
    },
    {
      "epoch": 0.008159779875705679,
      "grad_norm": 0.2149333655834198,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.1209,
      "step": 43
    },
    {
      "epoch": 0.008349542198396509,
      "grad_norm": 0.2174244076013565,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.1614,
      "step": 44
    },
    {
      "epoch": 0.008539304521087338,
      "grad_norm": 0.23298555612564087,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.0815,
      "step": 45
    },
    {
      "epoch": 0.008729066843778168,
      "grad_norm": 0.26479408144950867,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.1979,
      "step": 46
    },
    {
      "epoch": 0.008918829166468998,
      "grad_norm": 0.24461990594863892,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.0346,
      "step": 47
    },
    {
      "epoch": 0.009108591489159828,
      "grad_norm": 0.2512701451778412,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.0319,
      "step": 48
    },
    {
      "epoch": 0.009298353811850657,
      "grad_norm": 0.2517155110836029,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.055,
      "step": 49
    },
    {
      "epoch": 0.009488116134541487,
      "grad_norm": 0.2827569246292114,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.1684,
      "step": 50
    },
    {
      "epoch": 0.009488116134541487,
      "eval_loss": 1.0552674531936646,
      "eval_runtime": 277.3628,
      "eval_samples_per_second": 32.001,
      "eval_steps_per_second": 8.0,
      "step": 50
    },
    {
      "epoch": 0.009677878457232317,
      "grad_norm": 0.21244028210639954,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.9075,
      "step": 51
    },
    {
      "epoch": 0.009867640779923146,
      "grad_norm": 0.20778271555900574,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.045,
      "step": 52
    },
    {
      "epoch": 0.010057403102613976,
      "grad_norm": 0.19142813980579376,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.0145,
      "step": 53
    },
    {
      "epoch": 0.010247165425304806,
      "grad_norm": 0.1807219535112381,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.0956,
      "step": 54
    },
    {
      "epoch": 0.010436927747995636,
      "grad_norm": 0.16223390400409698,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.0037,
      "step": 55
    },
    {
      "epoch": 0.010626690070686465,
      "grad_norm": 0.15767407417297363,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.9815,
      "step": 56
    },
    {
      "epoch": 0.010816452393377295,
      "grad_norm": 0.14841850101947784,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.004,
      "step": 57
    },
    {
      "epoch": 0.011006214716068125,
      "grad_norm": 0.14706046879291534,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.0287,
      "step": 58
    },
    {
      "epoch": 0.011195977038758954,
      "grad_norm": 0.16504324972629547,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.0654,
      "step": 59
    },
    {
      "epoch": 0.011385739361449784,
      "grad_norm": 0.1560998260974884,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.9596,
      "step": 60
    },
    {
      "epoch": 0.011575501684140614,
      "grad_norm": 0.16232743859291077,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.0206,
      "step": 61
    },
    {
      "epoch": 0.011765264006831444,
      "grad_norm": 0.16730357706546783,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.9955,
      "step": 62
    },
    {
      "epoch": 0.011955026329522273,
      "grad_norm": 0.1669314056634903,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.9638,
      "step": 63
    },
    {
      "epoch": 0.012144788652213103,
      "grad_norm": 0.17288701236248016,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.0123,
      "step": 64
    },
    {
      "epoch": 0.012334550974903933,
      "grad_norm": 0.1711699515581131,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.024,
      "step": 65
    },
    {
      "epoch": 0.012524313297594762,
      "grad_norm": 0.17235831916332245,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.0528,
      "step": 66
    },
    {
      "epoch": 0.012714075620285592,
      "grad_norm": 0.17410272359848022,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.1027,
      "step": 67
    },
    {
      "epoch": 0.012903837942976422,
      "grad_norm": 0.16469921171665192,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.9787,
      "step": 68
    },
    {
      "epoch": 0.013093600265667252,
      "grad_norm": 0.1830262541770935,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.9897,
      "step": 69
    },
    {
      "epoch": 0.013283362588358081,
      "grad_norm": 0.18811868131160736,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.0582,
      "step": 70
    },
    {
      "epoch": 0.013473124911048911,
      "grad_norm": 0.19638700783252716,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.1117,
      "step": 71
    },
    {
      "epoch": 0.013662887233739741,
      "grad_norm": 0.18470486998558044,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.0433,
      "step": 72
    },
    {
      "epoch": 0.01385264955643057,
      "grad_norm": 0.1824895590543747,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.047,
      "step": 73
    },
    {
      "epoch": 0.0140424118791214,
      "grad_norm": 0.18160541355609894,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.0565,
      "step": 74
    },
    {
      "epoch": 0.01423217420181223,
      "grad_norm": 0.1841384768486023,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.9941,
      "step": 75
    },
    {
      "epoch": 0.01442193652450306,
      "grad_norm": 0.19593071937561035,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.0795,
      "step": 76
    },
    {
      "epoch": 0.014611698847193889,
      "grad_norm": 0.19380022585391998,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.0292,
      "step": 77
    },
    {
      "epoch": 0.01480146116988472,
      "grad_norm": 0.1812576800584793,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.0105,
      "step": 78
    },
    {
      "epoch": 0.01499122349257555,
      "grad_norm": 0.2010316103696823,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.0638,
      "step": 79
    },
    {
      "epoch": 0.01518098581526638,
      "grad_norm": 0.19539521634578705,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.942,
      "step": 80
    },
    {
      "epoch": 0.015370748137957208,
      "grad_norm": 0.19889476895332336,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.0528,
      "step": 81
    },
    {
      "epoch": 0.015560510460648038,
      "grad_norm": 0.21053452789783478,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.0796,
      "step": 82
    },
    {
      "epoch": 0.01575027278333887,
      "grad_norm": 0.20258855819702148,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.9893,
      "step": 83
    },
    {
      "epoch": 0.015940035106029697,
      "grad_norm": 0.20124588906764984,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.9893,
      "step": 84
    },
    {
      "epoch": 0.01612979742872053,
      "grad_norm": 0.233975350856781,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.0696,
      "step": 85
    },
    {
      "epoch": 0.016319559751411358,
      "grad_norm": 0.2315712422132492,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.0057,
      "step": 86
    },
    {
      "epoch": 0.016509322074102186,
      "grad_norm": 0.23223906755447388,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.0939,
      "step": 87
    },
    {
      "epoch": 0.016699084396793018,
      "grad_norm": 0.23546229302883148,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.0663,
      "step": 88
    },
    {
      "epoch": 0.016888846719483847,
      "grad_norm": 0.22160938382148743,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.9972,
      "step": 89
    },
    {
      "epoch": 0.017078609042174675,
      "grad_norm": 0.23993346095085144,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.9169,
      "step": 90
    },
    {
      "epoch": 0.017268371364865507,
      "grad_norm": 0.23985272645950317,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.0183,
      "step": 91
    },
    {
      "epoch": 0.017458133687556335,
      "grad_norm": 0.24388989806175232,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.0668,
      "step": 92
    },
    {
      "epoch": 0.017647896010247164,
      "grad_norm": 0.2586608827114105,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.0306,
      "step": 93
    },
    {
      "epoch": 0.017837658332937996,
      "grad_norm": 0.251071035861969,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.9846,
      "step": 94
    },
    {
      "epoch": 0.018027420655628824,
      "grad_norm": 0.2634812593460083,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.0191,
      "step": 95
    },
    {
      "epoch": 0.018217182978319656,
      "grad_norm": 0.26588213443756104,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.988,
      "step": 96
    },
    {
      "epoch": 0.018406945301010485,
      "grad_norm": 0.2804090976715088,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.0653,
      "step": 97
    },
    {
      "epoch": 0.018596707623701313,
      "grad_norm": 0.2770501971244812,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.0198,
      "step": 98
    },
    {
      "epoch": 0.018786469946392145,
      "grad_norm": 0.31081750988960266,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.9818,
      "step": 99
    },
    {
      "epoch": 0.018976232269082974,
      "grad_norm": 0.3638320863246918,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.9592,
      "step": 100
    },
    {
      "epoch": 0.018976232269082974,
      "eval_loss": 0.9981662631034851,
      "eval_runtime": 277.1213,
      "eval_samples_per_second": 32.029,
      "eval_steps_per_second": 8.007,
      "step": 100
    },
    {
      "epoch": 0.019165994591773802,
      "grad_norm": 0.1871665120124817,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.9838,
      "step": 101
    },
    {
      "epoch": 0.019355756914464634,
      "grad_norm": 0.21412956714630127,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.9434,
      "step": 102
    },
    {
      "epoch": 0.019545519237155463,
      "grad_norm": 0.20799927413463593,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.9957,
      "step": 103
    },
    {
      "epoch": 0.01973528155984629,
      "grad_norm": 0.20693427324295044,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.9414,
      "step": 104
    },
    {
      "epoch": 0.019925043882537123,
      "grad_norm": 0.19739599525928497,
      "learning_rate": 5e-05,
      "loss": 0.925,
      "step": 105
    },
    {
      "epoch": 0.02011480620522795,
      "grad_norm": 0.194368377327919,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.9672,
      "step": 106
    },
    {
      "epoch": 0.02030456852791878,
      "grad_norm": 0.198219433426857,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.9094,
      "step": 107
    },
    {
      "epoch": 0.020494330850609612,
      "grad_norm": 0.18252170085906982,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.9321,
      "step": 108
    },
    {
      "epoch": 0.02068409317330044,
      "grad_norm": 0.18987925350666046,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.9764,
      "step": 109
    },
    {
      "epoch": 0.020873855495991273,
      "grad_norm": 0.18091386556625366,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.9794,
      "step": 110
    },
    {
      "epoch": 0.0210636178186821,
      "grad_norm": 0.18169650435447693,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.0014,
      "step": 111
    },
    {
      "epoch": 0.02125338014137293,
      "grad_norm": 0.19523243606090546,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.9512,
      "step": 112
    },
    {
      "epoch": 0.02144314246406376,
      "grad_norm": 0.20139889419078827,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.9744,
      "step": 113
    },
    {
      "epoch": 0.02163290478675459,
      "grad_norm": 0.18810676038265228,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.9548,
      "step": 114
    },
    {
      "epoch": 0.02182266710944542,
      "grad_norm": 0.2029637098312378,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.8991,
      "step": 115
    },
    {
      "epoch": 0.02201242943213625,
      "grad_norm": 0.20979595184326172,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.0381,
      "step": 116
    },
    {
      "epoch": 0.02220219175482708,
      "grad_norm": 0.20881718397140503,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.9862,
      "step": 117
    },
    {
      "epoch": 0.022391954077517907,
      "grad_norm": 0.20691141486167908,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.9603,
      "step": 118
    },
    {
      "epoch": 0.02258171640020874,
      "grad_norm": 0.21193382143974304,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.0092,
      "step": 119
    },
    {
      "epoch": 0.022771478722899568,
      "grad_norm": 0.19551590085029602,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.89,
      "step": 120
    },
    {
      "epoch": 0.022961241045590396,
      "grad_norm": 0.20724761486053467,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.9521,
      "step": 121
    },
    {
      "epoch": 0.02315100336828123,
      "grad_norm": 0.21067377924919128,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.9714,
      "step": 122
    },
    {
      "epoch": 0.023340765690972057,
      "grad_norm": 0.22233997285366058,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.0466,
      "step": 123
    },
    {
      "epoch": 0.02353052801366289,
      "grad_norm": 0.2184775173664093,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.0076,
      "step": 124
    },
    {
      "epoch": 0.023720290336353717,
      "grad_norm": 0.21032242476940155,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.9897,
      "step": 125
    },
    {
      "epoch": 0.023910052659044546,
      "grad_norm": 0.22417564690113068,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.0463,
      "step": 126
    },
    {
      "epoch": 0.024099814981735378,
      "grad_norm": 0.21387426555156708,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.9584,
      "step": 127
    },
    {
      "epoch": 0.024289577304426206,
      "grad_norm": 0.2128717303276062,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.9954,
      "step": 128
    },
    {
      "epoch": 0.024479339627117035,
      "grad_norm": 0.2258055955171585,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.9482,
      "step": 129
    },
    {
      "epoch": 0.024669101949807867,
      "grad_norm": 0.21299861371517181,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.9147,
      "step": 130
    },
    {
      "epoch": 0.024858864272498695,
      "grad_norm": 0.2303430736064911,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.9663,
      "step": 131
    },
    {
      "epoch": 0.025048626595189524,
      "grad_norm": 0.22739309072494507,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.9808,
      "step": 132
    },
    {
      "epoch": 0.025238388917880356,
      "grad_norm": 0.2431771159172058,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.1092,
      "step": 133
    },
    {
      "epoch": 0.025428151240571184,
      "grad_norm": 0.24554990231990814,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.012,
      "step": 134
    },
    {
      "epoch": 0.025617913563262016,
      "grad_norm": 0.24237067997455597,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.0044,
      "step": 135
    },
    {
      "epoch": 0.025807675885952844,
      "grad_norm": 0.24016183614730835,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.0184,
      "step": 136
    },
    {
      "epoch": 0.025997438208643673,
      "grad_norm": 0.24562525749206543,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.9808,
      "step": 137
    },
    {
      "epoch": 0.026187200531334505,
      "grad_norm": 0.2517540156841278,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.96,
      "step": 138
    },
    {
      "epoch": 0.026376962854025333,
      "grad_norm": 0.24588365852832794,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.0525,
      "step": 139
    },
    {
      "epoch": 0.026566725176716162,
      "grad_norm": 0.25156688690185547,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.0025,
      "step": 140
    },
    {
      "epoch": 0.026756487499406994,
      "grad_norm": 0.2560715079307556,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.0106,
      "step": 141
    },
    {
      "epoch": 0.026946249822097822,
      "grad_norm": 0.2707711458206177,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.938,
      "step": 142
    },
    {
      "epoch": 0.02713601214478865,
      "grad_norm": 0.2656857371330261,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.9412,
      "step": 143
    },
    {
      "epoch": 0.027325774467479483,
      "grad_norm": 0.27317842841148376,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.9181,
      "step": 144
    },
    {
      "epoch": 0.02751553679017031,
      "grad_norm": 0.27799221873283386,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.044,
      "step": 145
    },
    {
      "epoch": 0.02770529911286114,
      "grad_norm": 0.2845875024795532,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.0191,
      "step": 146
    },
    {
      "epoch": 0.02789506143555197,
      "grad_norm": 0.2962326109409332,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.0425,
      "step": 147
    },
    {
      "epoch": 0.0280848237582428,
      "grad_norm": 0.31925490498542786,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.9788,
      "step": 148
    },
    {
      "epoch": 0.028274586080933632,
      "grad_norm": 0.32261669635772705,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.9128,
      "step": 149
    },
    {
      "epoch": 0.02846434840362446,
      "grad_norm": 0.3913915157318115,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.0025,
      "step": 150
    },
    {
      "epoch": 0.02846434840362446,
      "eval_loss": 0.9701759219169617,
      "eval_runtime": 277.0435,
      "eval_samples_per_second": 32.038,
      "eval_steps_per_second": 8.01,
      "step": 150
    },
    {
      "epoch": 0.02865411072631529,
      "grad_norm": 0.17973150312900543,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.8605,
      "step": 151
    },
    {
      "epoch": 0.02884387304900612,
      "grad_norm": 0.18076074123382568,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.903,
      "step": 152
    },
    {
      "epoch": 0.02903363537169695,
      "grad_norm": 0.18535369634628296,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.9506,
      "step": 153
    },
    {
      "epoch": 0.029223397694387778,
      "grad_norm": 0.19823555648326874,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.9051,
      "step": 154
    },
    {
      "epoch": 0.02941316001707861,
      "grad_norm": 0.19839805364608765,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.8783,
      "step": 155
    },
    {
      "epoch": 0.02960292233976944,
      "grad_norm": 0.21332909166812897,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.9725,
      "step": 156
    },
    {
      "epoch": 0.029792684662460267,
      "grad_norm": 0.209469735622406,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.9312,
      "step": 157
    },
    {
      "epoch": 0.0299824469851511,
      "grad_norm": 0.2010391503572464,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.9853,
      "step": 158
    },
    {
      "epoch": 0.030172209307841927,
      "grad_norm": 0.20170678198337555,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.9651,
      "step": 159
    },
    {
      "epoch": 0.03036197163053276,
      "grad_norm": 0.2024935632944107,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.9731,
      "step": 160
    },
    {
      "epoch": 0.030551733953223588,
      "grad_norm": 0.20285053551197052,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.9838,
      "step": 161
    },
    {
      "epoch": 0.030741496275914416,
      "grad_norm": 0.20995524525642395,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.9988,
      "step": 162
    },
    {
      "epoch": 0.03093125859860525,
      "grad_norm": 0.20692457258701324,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.9401,
      "step": 163
    },
    {
      "epoch": 0.031121020921296077,
      "grad_norm": 0.20284044742584229,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.038,
      "step": 164
    },
    {
      "epoch": 0.03131078324398691,
      "grad_norm": 0.21334685385227203,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.9762,
      "step": 165
    },
    {
      "epoch": 0.03150054556667774,
      "grad_norm": 0.20292659103870392,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.9274,
      "step": 166
    },
    {
      "epoch": 0.031690307889368566,
      "grad_norm": 0.20538827776908875,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.9339,
      "step": 167
    },
    {
      "epoch": 0.031880070212059394,
      "grad_norm": 0.2079194039106369,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.9661,
      "step": 168
    },
    {
      "epoch": 0.03206983253475022,
      "grad_norm": 0.21250757575035095,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.9922,
      "step": 169
    },
    {
      "epoch": 0.03225959485744106,
      "grad_norm": 0.21290943026542664,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.0206,
      "step": 170
    },
    {
      "epoch": 0.03244935718013189,
      "grad_norm": 0.2162042260169983,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.9769,
      "step": 171
    },
    {
      "epoch": 0.032639119502822715,
      "grad_norm": 0.22110210359096527,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.9048,
      "step": 172
    },
    {
      "epoch": 0.032828881825513544,
      "grad_norm": 0.22040364146232605,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.9719,
      "step": 173
    },
    {
      "epoch": 0.03301864414820437,
      "grad_norm": 0.21894502639770508,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.0065,
      "step": 174
    },
    {
      "epoch": 0.0332084064708952,
      "grad_norm": 0.23096728324890137,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.9631,
      "step": 175
    },
    {
      "epoch": 0.033398168793586036,
      "grad_norm": 0.22498327493667603,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9149,
      "step": 176
    },
    {
      "epoch": 0.033587931116276865,
      "grad_norm": 0.23559530079364777,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.0205,
      "step": 177
    },
    {
      "epoch": 0.03377769343896769,
      "grad_norm": 0.23133589327335358,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.952,
      "step": 178
    },
    {
      "epoch": 0.03396745576165852,
      "grad_norm": 0.2298775464296341,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.9282,
      "step": 179
    },
    {
      "epoch": 0.03415721808434935,
      "grad_norm": 0.23774872720241547,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.9635,
      "step": 180
    },
    {
      "epoch": 0.034346980407040185,
      "grad_norm": 0.24305981397628784,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.0001,
      "step": 181
    },
    {
      "epoch": 0.034536742729731014,
      "grad_norm": 0.22747716307640076,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.9938,
      "step": 182
    },
    {
      "epoch": 0.03472650505242184,
      "grad_norm": 0.2410711795091629,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.0034,
      "step": 183
    },
    {
      "epoch": 0.03491626737511267,
      "grad_norm": 0.2383907437324524,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.0638,
      "step": 184
    },
    {
      "epoch": 0.0351060296978035,
      "grad_norm": 0.2630782127380371,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.0287,
      "step": 185
    },
    {
      "epoch": 0.03529579202049433,
      "grad_norm": 0.24380278587341309,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.8993,
      "step": 186
    },
    {
      "epoch": 0.03548555434318516,
      "grad_norm": 0.25371795892715454,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.9341,
      "step": 187
    },
    {
      "epoch": 0.03567531666587599,
      "grad_norm": 0.26402977108955383,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.0405,
      "step": 188
    },
    {
      "epoch": 0.03586507898856682,
      "grad_norm": 0.28010696172714233,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.0636,
      "step": 189
    },
    {
      "epoch": 0.03605484131125765,
      "grad_norm": 0.26449769735336304,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.9787,
      "step": 190
    },
    {
      "epoch": 0.03624460363394848,
      "grad_norm": 0.2666074335575104,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.9424,
      "step": 191
    },
    {
      "epoch": 0.03643436595663931,
      "grad_norm": 0.2772065997123718,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.0386,
      "step": 192
    },
    {
      "epoch": 0.03662412827933014,
      "grad_norm": 0.2703202962875366,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.9225,
      "step": 193
    },
    {
      "epoch": 0.03681389060202097,
      "grad_norm": 0.2837485671043396,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.9709,
      "step": 194
    },
    {
      "epoch": 0.0370036529247118,
      "grad_norm": 0.28822094202041626,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.0295,
      "step": 195
    },
    {
      "epoch": 0.03719341524740263,
      "grad_norm": 0.31115013360977173,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.9704,
      "step": 196
    },
    {
      "epoch": 0.037383177570093455,
      "grad_norm": 0.3050763010978699,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.0271,
      "step": 197
    },
    {
      "epoch": 0.03757293989278429,
      "grad_norm": 0.33131471276283264,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.0034,
      "step": 198
    },
    {
      "epoch": 0.03776270221547512,
      "grad_norm": 0.33271533250808716,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.9322,
      "step": 199
    },
    {
      "epoch": 0.03795246453816595,
      "grad_norm": 0.4427962005138397,
      "learning_rate": 0.0,
      "loss": 0.9581,
      "step": 200
    },
    {
      "epoch": 0.03795246453816595,
      "eval_loss": 0.963947057723999,
      "eval_runtime": 277.5789,
      "eval_samples_per_second": 31.976,
      "eval_steps_per_second": 7.994,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.675988411632845e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}