{
  "best_global_step": 1125,
  "best_metric": 0.0719488188624382,
  "best_model_checkpoint": "outputs_3/checkpoint-1125",
  "epoch": 3.630048465266559,
  "eval_steps": 75,
  "global_step": 1125,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0032310177705977385,
      "grad_norm": 0.0250613521784544,
      "learning_rate": 0.0,
      "loss": 0.1103,
      "step": 1
    },
    {
      "epoch": 0.006462035541195477,
      "grad_norm": 0.03354981914162636,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 0.1336,
      "step": 2
    },
    {
      "epoch": 0.009693053311793215,
      "grad_norm": 0.024361876770853996,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.1158,
      "step": 3
    },
    {
      "epoch": 0.012924071082390954,
      "grad_norm": 0.024172648787498474,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 0.0937,
      "step": 4
    },
    {
      "epoch": 0.01615508885298869,
      "grad_norm": 0.024678289890289307,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.1047,
      "step": 5
    },
    {
      "epoch": 0.01938610662358643,
      "grad_norm": 0.031230105087161064,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.1479,
      "step": 6
    },
    {
      "epoch": 0.022617124394184167,
      "grad_norm": 0.024539202451705933,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.126,
      "step": 7
    },
    {
      "epoch": 0.025848142164781908,
      "grad_norm": 0.02983587421476841,
      "learning_rate": 7.000000000000001e-07,
      "loss": 0.1218,
      "step": 8
    },
    {
      "epoch": 0.029079159935379646,
      "grad_norm": 0.02570008486509323,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.1153,
      "step": 9
    },
    {
      "epoch": 0.03231017770597738,
      "grad_norm": 0.024239273741841316,
      "learning_rate": 9.000000000000001e-07,
      "loss": 0.0952,
      "step": 10
    },
    {
      "epoch": 0.035541195476575124,
      "grad_norm": 0.020495450124144554,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0958,
      "step": 11
    },
    {
      "epoch": 0.03877221324717286,
      "grad_norm": 0.028608962893486023,
      "learning_rate": 1.1e-06,
      "loss": 0.1169,
      "step": 12
    },
    {
      "epoch": 0.0420032310177706,
      "grad_norm": 0.03878644108772278,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.1371,
      "step": 13
    },
    {
      "epoch": 0.045234248788368334,
      "grad_norm": 0.02789674885571003,
      "learning_rate": 1.3e-06,
      "loss": 0.1266,
      "step": 14
    },
    {
      "epoch": 0.048465266558966075,
      "grad_norm": 0.03313566744327545,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.1301,
      "step": 15
    },
    {
      "epoch": 0.051696284329563816,
      "grad_norm": 0.0248391292989254,
      "learning_rate": 1.5e-06,
      "loss": 0.1056,
      "step": 16
    },
    {
      "epoch": 0.05492730210016155,
      "grad_norm": 0.024395154789090157,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.112,
      "step": 17
    },
    {
      "epoch": 0.05815831987075929,
      "grad_norm": 0.03043658658862114,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 0.1037,
      "step": 18
    },
    {
      "epoch": 0.061389337641357025,
      "grad_norm": 0.02323235385119915,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.0996,
      "step": 19
    },
    {
      "epoch": 0.06462035541195477,
      "grad_norm": 0.03656580671668053,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 0.1327,
      "step": 20
    },
    {
      "epoch": 0.06785137318255251,
      "grad_norm": 0.02677535079419613,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.1205,
      "step": 21
    },
    {
      "epoch": 0.07108239095315025,
      "grad_norm": 0.029969926923513412,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.126,
      "step": 22
    },
    {
      "epoch": 0.07431340872374798,
      "grad_norm": 0.02832009270787239,
      "learning_rate": 2.2e-06,
      "loss": 0.1208,
      "step": 23
    },
    {
      "epoch": 0.07754442649434572,
      "grad_norm": 0.023000000044703484,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 0.102,
      "step": 24
    },
    {
      "epoch": 0.08077544426494346,
      "grad_norm": 0.04773552715778351,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.1403,
      "step": 25
    },
    {
      "epoch": 0.0840064620355412,
      "grad_norm": 0.03021993860602379,
      "learning_rate": 2.5e-06,
      "loss": 0.1116,
      "step": 26
    },
    {
      "epoch": 0.08723747980613894,
      "grad_norm": 0.026623567566275597,
      "learning_rate": 2.6e-06,
      "loss": 0.1052,
      "step": 27
    },
    {
      "epoch": 0.09046849757673667,
      "grad_norm": 0.02503894828259945,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 0.1016,
      "step": 28
    },
    {
      "epoch": 0.09369951534733441,
      "grad_norm": 0.026578862220048904,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.1026,
      "step": 29
    },
    {
      "epoch": 0.09693053311793215,
      "grad_norm": 0.029514916241168976,
      "learning_rate": 2.9e-06,
      "loss": 0.1047,
      "step": 30
    },
    {
      "epoch": 0.10016155088852989,
      "grad_norm": 0.025679711252450943,
      "learning_rate": 3e-06,
      "loss": 0.1228,
      "step": 31
    },
    {
      "epoch": 0.10339256865912763,
      "grad_norm": 0.023719631135463715,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 0.0997,
      "step": 32
    },
    {
      "epoch": 0.10662358642972536,
      "grad_norm": 0.03468646854162216,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.1354,
      "step": 33
    },
    {
      "epoch": 0.1098546042003231,
      "grad_norm": 0.02570706605911255,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 0.1007,
      "step": 34
    },
    {
      "epoch": 0.11308562197092084,
      "grad_norm": 0.03752969205379486,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.1512,
      "step": 35
    },
    {
      "epoch": 0.11631663974151858,
      "grad_norm": 0.034612756222486496,
      "learning_rate": 3.5e-06,
      "loss": 0.1212,
      "step": 36
    },
    {
      "epoch": 0.11954765751211632,
      "grad_norm": 0.022866638377308846,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.0962,
      "step": 37
    },
    {
      "epoch": 0.12277867528271405,
      "grad_norm": 0.03314971551299095,
      "learning_rate": 3.7e-06,
      "loss": 0.1275,
      "step": 38
    },
    {
      "epoch": 0.1260096930533118,
      "grad_norm": 0.026049382984638214,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.1095,
      "step": 39
    },
    {
      "epoch": 0.12924071082390953,
      "grad_norm": 0.027846891433000565,
      "learning_rate": 3.900000000000001e-06,
      "loss": 0.1208,
      "step": 40
    },
    {
      "epoch": 0.13247172859450726,
      "grad_norm": 0.03061266988515854,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.1071,
      "step": 41
    },
    {
      "epoch": 0.13570274636510501,
      "grad_norm": 0.028831277042627335,
      "learning_rate": 4.1e-06,
      "loss": 0.122,
      "step": 42
    },
    {
      "epoch": 0.13893376413570274,
      "grad_norm": 0.024587510153651237,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.0891,
      "step": 43
    },
    {
      "epoch": 0.1421647819063005,
      "grad_norm": 0.03614000231027603,
      "learning_rate": 4.3e-06,
      "loss": 0.1443,
      "step": 44
    },
    {
      "epoch": 0.14539579967689822,
      "grad_norm": 0.0399002730846405,
      "learning_rate": 4.4e-06,
      "loss": 0.1306,
      "step": 45
    },
    {
      "epoch": 0.14862681744749595,
      "grad_norm": 0.03731178864836693,
      "learning_rate": 4.5e-06,
      "loss": 0.1552,
      "step": 46
    },
    {
      "epoch": 0.1518578352180937,
      "grad_norm": 0.03669052943587303,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.1285,
      "step": 47
    },
    {
      "epoch": 0.15508885298869143,
      "grad_norm": 0.029436811804771423,
      "learning_rate": 4.7e-06,
      "loss": 0.1128,
      "step": 48
    },
    {
      "epoch": 0.1583198707592892,
      "grad_norm": 0.03553691506385803,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.1295,
      "step": 49
    },
    {
      "epoch": 0.16155088852988692,
      "grad_norm": 0.045196086168289185,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.1589,
      "step": 50
    },
    {
      "epoch": 0.16478190630048464,
      "grad_norm": 0.03583088517189026,
      "learning_rate": 5e-06,
      "loss": 0.1195,
      "step": 51
    },
    {
      "epoch": 0.1680129240710824,
      "grad_norm": 0.03799896687269211,
      "learning_rate": 5.1e-06,
      "loss": 0.1293,
      "step": 52
    },
    {
      "epoch": 0.17124394184168013,
      "grad_norm": 0.0341208279132843,
      "learning_rate": 5.2e-06,
      "loss": 0.1121,
      "step": 53
    },
    {
      "epoch": 0.17447495961227788,
      "grad_norm": 0.0367840901017189,
      "learning_rate": 5.300000000000001e-06,
      "loss": 0.1288,
      "step": 54
    },
    {
      "epoch": 0.1777059773828756,
      "grad_norm": 0.03497275337576866,
      "learning_rate": 5.400000000000001e-06,
      "loss": 0.1146,
      "step": 55
    },
    {
      "epoch": 0.18093699515347333,
      "grad_norm": 0.04450898617506027,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.1261,
      "step": 56
    },
    {
      "epoch": 0.1841680129240711,
      "grad_norm": 0.029873637482523918,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.0981,
      "step": 57
    },
    {
      "epoch": 0.18739903069466882,
      "grad_norm": 0.030145753175020218,
      "learning_rate": 5.7e-06,
      "loss": 0.1121,
      "step": 58
    },
    {
      "epoch": 0.19063004846526657,
      "grad_norm": 0.03658242151141167,
      "learning_rate": 5.8e-06,
      "loss": 0.1416,
      "step": 59
    },
    {
      "epoch": 0.1938610662358643,
      "grad_norm": 0.049440935254096985,
      "learning_rate": 5.9e-06,
      "loss": 0.125,
      "step": 60
    },
    {
      "epoch": 0.19709208400646203,
      "grad_norm": 0.0388176292181015,
      "learning_rate": 6e-06,
      "loss": 0.1195,
      "step": 61
    },
    {
      "epoch": 0.20032310177705978,
      "grad_norm": 0.03422081843018532,
      "learning_rate": 6.1e-06,
      "loss": 0.1191,
      "step": 62
    },
    {
      "epoch": 0.2035541195476575,
      "grad_norm": 0.047777559608221054,
      "learning_rate": 6.200000000000001e-06,
      "loss": 0.1304,
      "step": 63
    },
    {
      "epoch": 0.20678513731825526,
      "grad_norm": 0.031583212316036224,
      "learning_rate": 6.300000000000001e-06,
      "loss": 0.0996,
      "step": 64
    },
    {
      "epoch": 0.210016155088853,
      "grad_norm": 0.03744835779070854,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.1124,
      "step": 65
    },
    {
      "epoch": 0.21324717285945072,
      "grad_norm": 0.04165760055184364,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.1041,
      "step": 66
    },
    {
      "epoch": 0.21647819063004847,
      "grad_norm": 0.03987026587128639,
      "learning_rate": 6.600000000000001e-06,
      "loss": 0.133,
      "step": 67
    },
    {
      "epoch": 0.2197092084006462,
      "grad_norm": 0.040481116622686386,
      "learning_rate": 6.700000000000001e-06,
      "loss": 0.0944,
      "step": 68
    },
    {
      "epoch": 0.22294022617124395,
      "grad_norm": 0.03691282495856285,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.1018,
      "step": 69
    },
    {
      "epoch": 0.22617124394184168,
      "grad_norm": 0.04485676437616348,
      "learning_rate": 6.9e-06,
      "loss": 0.137,
      "step": 70
    },
    {
      "epoch": 0.2294022617124394,
      "grad_norm": 0.02854587510228157,
      "learning_rate": 7e-06,
      "loss": 0.0885,
      "step": 71
    },
    {
      "epoch": 0.23263327948303716,
      "grad_norm": 0.04462384432554245,
      "learning_rate": 7.100000000000001e-06,
      "loss": 0.1429,
      "step": 72
    },
    {
      "epoch": 0.2358642972536349,
      "grad_norm": 0.040676336735486984,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.121,
      "step": 73
    },
    {
      "epoch": 0.23909531502423265,
      "grad_norm": 0.0420430488884449,
      "learning_rate": 7.3e-06,
      "loss": 0.1197,
      "step": 74
    },
    {
      "epoch": 0.24232633279483037,
      "grad_norm": 0.034694962203502655,
      "learning_rate": 7.4e-06,
      "loss": 0.1177,
      "step": 75
    },
    {
      "epoch": 0.24232633279483037,
      "eval_loss": 0.12372970581054688,
      "eval_runtime": 188.2884,
      "eval_samples_per_second": 1.046,
      "eval_steps_per_second": 1.046,
      "step": 75
    },
    {
      "epoch": 0.2455573505654281,
      "grad_norm": 0.03579672798514366,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.1058,
      "step": 76
    },
    {
      "epoch": 0.24878836833602586,
      "grad_norm": 0.029704652726650238,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.0882,
      "step": 77
    },
    {
      "epoch": 0.2520193861066236,
      "grad_norm": 0.04732828587293625,
      "learning_rate": 7.7e-06,
      "loss": 0.1277,
      "step": 78
    },
    {
      "epoch": 0.2552504038772213,
      "grad_norm": 0.027987899258732796,
      "learning_rate": 7.800000000000002e-06,
      "loss": 0.0751,
      "step": 79
    },
    {
      "epoch": 0.25848142164781907,
      "grad_norm": 0.04119185730814934,
      "learning_rate": 7.9e-06,
      "loss": 0.1043,
      "step": 80
    },
    {
      "epoch": 0.2617124394184168,
      "grad_norm": 0.04631367698311806,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.118,
      "step": 81
    },
    {
      "epoch": 0.2649434571890145,
      "grad_norm": 0.03165270760655403,
      "learning_rate": 8.1e-06,
      "loss": 0.0889,
      "step": 82
    },
    {
      "epoch": 0.2681744749596123,
      "grad_norm": 0.03328806161880493,
      "learning_rate": 8.2e-06,
      "loss": 0.097,
      "step": 83
    },
    {
      "epoch": 0.27140549273021003,
      "grad_norm": 0.05337163060903549,
      "learning_rate": 8.3e-06,
      "loss": 0.1373,
      "step": 84
    },
    {
      "epoch": 0.27463651050080773,
      "grad_norm": 0.030968431383371353,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.077,
      "step": 85
    },
    {
      "epoch": 0.2778675282714055,
      "grad_norm": 0.03477643430233002,
      "learning_rate": 8.5e-06,
      "loss": 0.1016,
      "step": 86
    },
    {
      "epoch": 0.28109854604200324,
      "grad_norm": 0.04051528871059418,
      "learning_rate": 8.6e-06,
      "loss": 0.1017,
      "step": 87
    },
    {
      "epoch": 0.284329563812601,
      "grad_norm": 0.039160750806331635,
      "learning_rate": 8.700000000000001e-06,
      "loss": 0.112,
      "step": 88
    },
    {
      "epoch": 0.2875605815831987,
      "grad_norm": 0.03572917729616165,
      "learning_rate": 8.8e-06,
      "loss": 0.083,
      "step": 89
    },
    {
      "epoch": 0.29079159935379645,
      "grad_norm": 0.05116155743598938,
      "learning_rate": 8.900000000000001e-06,
      "loss": 0.1262,
      "step": 90
    },
    {
      "epoch": 0.2940226171243942,
      "grad_norm": 0.043991196900606155,
      "learning_rate": 9e-06,
      "loss": 0.1147,
      "step": 91
    },
    {
      "epoch": 0.2972536348949919,
      "grad_norm": 0.03514918312430382,
      "learning_rate": 9.100000000000001e-06,
      "loss": 0.0874,
      "step": 92
    },
    {
      "epoch": 0.30048465266558966,
      "grad_norm": 0.03676354140043259,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.0988,
      "step": 93
    },
    {
      "epoch": 0.3037156704361874,
      "grad_norm": 0.04684548079967499,
      "learning_rate": 9.3e-06,
      "loss": 0.129,
      "step": 94
    },
    {
      "epoch": 0.3069466882067851,
      "grad_norm": 0.033971093595027924,
      "learning_rate": 9.4e-06,
      "loss": 0.09,
      "step": 95
    },
    {
      "epoch": 0.31017770597738287,
      "grad_norm": 0.0339505635201931,
      "learning_rate": 9.5e-06,
      "loss": 0.083,
      "step": 96
    },
    {
      "epoch": 0.3134087237479806,
      "grad_norm": 0.04067440703511238,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.1093,
      "step": 97
    },
    {
      "epoch": 0.3166397415185784,
      "grad_norm": 0.036671143025159836,
      "learning_rate": 9.7e-06,
      "loss": 0.0929,
      "step": 98
    },
    {
      "epoch": 0.3198707592891761,
      "grad_norm": 0.0459955595433712,
      "learning_rate": 9.800000000000001e-06,
      "loss": 0.1273,
      "step": 99
    },
    {
      "epoch": 0.32310177705977383,
      "grad_norm": 0.05120276287198067,
      "learning_rate": 9.9e-06,
      "loss": 0.1349,
      "step": 100
    },
    {
      "epoch": 0.3263327948303716,
      "grad_norm": 0.03149951994419098,
      "learning_rate": 1e-05,
      "loss": 0.074,
      "step": 101
    },
    {
      "epoch": 0.3295638126009693,
      "grad_norm": 0.2025747299194336,
      "learning_rate": 1.0100000000000002e-05,
      "loss": 0.099,
      "step": 102
    },
    {
      "epoch": 0.33279483037156704,
      "grad_norm": 0.03870416432619095,
      "learning_rate": 1.02e-05,
      "loss": 0.0984,
      "step": 103
    },
    {
      "epoch": 0.3360258481421648,
      "grad_norm": 0.05344085767865181,
      "learning_rate": 1.0300000000000001e-05,
      "loss": 0.1057,
      "step": 104
    },
    {
      "epoch": 0.3392568659127625,
      "grad_norm": 0.04757027328014374,
      "learning_rate": 1.04e-05,
      "loss": 0.124,
      "step": 105
    },
    {
      "epoch": 0.34248788368336025,
      "grad_norm": 0.03201949968934059,
      "learning_rate": 1.0500000000000001e-05,
      "loss": 0.0807,
      "step": 106
    },
    {
      "epoch": 0.345718901453958,
      "grad_norm": 0.03873045742511749,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 0.1048,
      "step": 107
    },
    {
      "epoch": 0.34894991922455576,
      "grad_norm": 0.027501031756401062,
      "learning_rate": 1.0700000000000001e-05,
      "loss": 0.0613,
      "step": 108
    },
    {
      "epoch": 0.35218093699515346,
      "grad_norm": 0.03909388184547424,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 0.0929,
      "step": 109
    },
    {
      "epoch": 0.3554119547657512,
      "grad_norm": 0.08006097376346588,
      "learning_rate": 1.0900000000000002e-05,
      "loss": 0.0908,
      "step": 110
    },
    {
      "epoch": 0.35864297253634897,
      "grad_norm": 0.03813672810792923,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.0764,
      "step": 111
    },
    {
      "epoch": 0.36187399030694667,
      "grad_norm": 0.030327340587973595,
      "learning_rate": 1.1100000000000002e-05,
      "loss": 0.077,
      "step": 112
    },
    {
      "epoch": 0.3651050080775444,
      "grad_norm": 0.03776196017861366,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.0759,
      "step": 113
    },
    {
      "epoch": 0.3683360258481422,
      "grad_norm": 0.039037927985191345,
      "learning_rate": 1.13e-05,
      "loss": 0.0926,
      "step": 114
    },
    {
      "epoch": 0.3715670436187399,
      "grad_norm": 0.0416683666408062,
      "learning_rate": 1.14e-05,
      "loss": 0.0914,
      "step": 115
    },
    {
      "epoch": 0.37479806138933763,
      "grad_norm": 0.04185537248849869,
      "learning_rate": 1.15e-05,
      "loss": 0.0893,
      "step": 116
    },
    {
      "epoch": 0.3780290791599354,
      "grad_norm": 0.04651897773146629,
      "learning_rate": 1.16e-05,
      "loss": 0.1094,
      "step": 117
    },
    {
      "epoch": 0.38126009693053314,
      "grad_norm": 0.04604775831103325,
      "learning_rate": 1.17e-05,
      "loss": 0.1068,
      "step": 118
    },
    {
      "epoch": 0.38449111470113084,
      "grad_norm": 0.02846536412835121,
      "learning_rate": 1.18e-05,
      "loss": 0.064,
      "step": 119
    },
    {
      "epoch": 0.3877221324717286,
      "grad_norm": 0.033402059227228165,
      "learning_rate": 1.1900000000000001e-05,
      "loss": 0.0725,
      "step": 120
    },
    {
      "epoch": 0.39095315024232635,
      "grad_norm": 0.044676899909973145,
      "learning_rate": 1.2e-05,
      "loss": 0.1017,
      "step": 121
    },
    {
      "epoch": 0.39418416801292405,
      "grad_norm": 0.05336389318108559,
      "learning_rate": 1.2100000000000001e-05,
      "loss": 0.1063,
      "step": 122
    },
    {
      "epoch": 0.3974151857835218,
      "grad_norm": 0.0402502678334713,
      "learning_rate": 1.22e-05,
      "loss": 0.0764,
      "step": 123
    },
    {
      "epoch": 0.40064620355411956,
      "grad_norm": 0.04342082887887955,
      "learning_rate": 1.23e-05,
      "loss": 0.1008,
      "step": 124
    },
    {
      "epoch": 0.40387722132471726,
      "grad_norm": 0.047081444412469864,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 0.1029,
      "step": 125
    },
    {
      "epoch": 0.407108239095315,
      "grad_norm": 0.038031261414289474,
      "learning_rate": 1.25e-05,
      "loss": 0.0783,
      "step": 126
    },
    {
      "epoch": 0.41033925686591277,
      "grad_norm": 0.03746628388762474,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 0.0764,
      "step": 127
    },
    {
      "epoch": 0.4135702746365105,
      "grad_norm": 0.04288509115576744,
      "learning_rate": 1.27e-05,
      "loss": 0.0892,
      "step": 128
    },
    {
      "epoch": 0.4168012924071082,
      "grad_norm": 0.042407114058732986,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.0925,
      "step": 129
    },
    {
      "epoch": 0.420032310177706,
      "grad_norm": 0.03754522651433945,
      "learning_rate": 1.2900000000000002e-05,
      "loss": 0.0833,
      "step": 130
    },
    {
      "epoch": 0.42326332794830374,
      "grad_norm": 0.04337688535451889,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.0976,
      "step": 131
    },
    {
      "epoch": 0.42649434571890144,
      "grad_norm": 0.03174331411719322,
      "learning_rate": 1.3100000000000002e-05,
      "loss": 0.0684,
      "step": 132
    },
    {
      "epoch": 0.4297253634894992,
      "grad_norm": 0.04556446522474289,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 0.0979,
      "step": 133
    },
    {
      "epoch": 0.43295638126009695,
      "grad_norm": 0.04222508519887924,
      "learning_rate": 1.3300000000000001e-05,
      "loss": 0.0923,
      "step": 134
    },
    {
      "epoch": 0.43618739903069464,
      "grad_norm": 0.047948531806468964,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 0.0991,
      "step": 135
    },
    {
      "epoch": 0.4394184168012924,
      "grad_norm": 0.04752594605088234,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.0864,
      "step": 136
    },
    {
      "epoch": 0.44264943457189015,
      "grad_norm": 0.049508459866046906,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.1035,
      "step": 137
    },
    {
      "epoch": 0.4458804523424879,
      "grad_norm": 0.05521553382277489,
      "learning_rate": 1.3700000000000003e-05,
      "loss": 0.0978,
      "step": 138
    },
    {
      "epoch": 0.4491114701130856,
      "grad_norm": 0.05352664738893509,
      "learning_rate": 1.38e-05,
      "loss": 0.1097,
      "step": 139
    },
    {
      "epoch": 0.45234248788368336,
      "grad_norm": 0.04332451522350311,
      "learning_rate": 1.39e-05,
      "loss": 0.0917,
      "step": 140
    },
    {
      "epoch": 0.4555735056542811,
      "grad_norm": 0.05932965502142906,
      "learning_rate": 1.4e-05,
      "loss": 0.0909,
      "step": 141
    },
    {
      "epoch": 0.4588045234248788,
      "grad_norm": 0.04634483903646469,
      "learning_rate": 1.41e-05,
      "loss": 0.0806,
      "step": 142
    },
    {
      "epoch": 0.4620355411954766,
      "grad_norm": 0.037898797541856766,
      "learning_rate": 1.4200000000000001e-05,
      "loss": 0.0802,
      "step": 143
    },
    {
      "epoch": 0.46526655896607433,
      "grad_norm": 0.04366337135434151,
      "learning_rate": 1.43e-05,
      "loss": 0.0835,
      "step": 144
    },
    {
      "epoch": 0.46849757673667203,
      "grad_norm": 0.03588287532329559,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.0746,
      "step": 145
    },
    {
      "epoch": 0.4717285945072698,
      "grad_norm": 0.04979556053876877,
      "learning_rate": 1.45e-05,
      "loss": 0.0914,
      "step": 146
    },
    {
      "epoch": 0.47495961227786754,
      "grad_norm": 0.03938375040888786,
      "learning_rate": 1.46e-05,
      "loss": 0.0696,
      "step": 147
    },
    {
      "epoch": 0.4781906300484653,
      "grad_norm": 0.04531609266996384,
      "learning_rate": 1.4700000000000002e-05,
      "loss": 0.0847,
      "step": 148
    },
    {
      "epoch": 0.481421647819063,
      "grad_norm": 0.04314682260155678,
      "learning_rate": 1.48e-05,
      "loss": 0.0872,
      "step": 149
    },
    {
      "epoch": 0.48465266558966075,
      "grad_norm": 0.04925313591957092,
      "learning_rate": 1.4900000000000001e-05,
      "loss": 0.0905,
      "step": 150
    },
    {
      "epoch": 0.48465266558966075,
      "eval_loss": 0.09467896819114685,
      "eval_runtime": 188.3915,
      "eval_samples_per_second": 1.046,
      "eval_steps_per_second": 1.046,
      "step": 150
    },
    {
      "epoch": 0.4878836833602585,
      "grad_norm": 0.04450507089495659,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.0704,
      "step": 151
    },
    {
      "epoch": 0.4911147011308562,
      "grad_norm": 0.03887806460261345,
      "learning_rate": 1.5100000000000001e-05,
      "loss": 0.0727,
      "step": 152
    },
    {
      "epoch": 0.49434571890145396,
      "grad_norm": 0.04298221319913864,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.0763,
      "step": 153
    },
    {
      "epoch": 0.4975767366720517,
      "grad_norm": 0.046640265733003616,
      "learning_rate": 1.5300000000000003e-05,
      "loss": 0.0794,
      "step": 154
    },
    {
      "epoch": 0.5008077544426495,
      "grad_norm": 0.03670887276530266,
      "learning_rate": 1.54e-05,
      "loss": 0.0681,
      "step": 155
    },
    {
      "epoch": 0.5040387722132472,
      "grad_norm": 0.03891611844301224,
      "learning_rate": 1.55e-05,
      "loss": 0.0712,
      "step": 156
    },
    {
      "epoch": 0.5072697899838449,
      "grad_norm": 0.042465128004550934,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 0.0828,
      "step": 157
    },
    {
      "epoch": 0.5105008077544426,
      "grad_norm": 0.054676711559295654,
      "learning_rate": 1.5700000000000002e-05,
      "loss": 0.0978,
      "step": 158
    },
    {
      "epoch": 0.5137318255250404,
      "grad_norm": 0.040950365364551544,
      "learning_rate": 1.58e-05,
      "loss": 0.0822,
      "step": 159
    },
    {
      "epoch": 0.5169628432956381,
      "grad_norm": 0.04756248742341995,
      "learning_rate": 1.5900000000000004e-05,
      "loss": 0.0914,
      "step": 160
    },
    {
      "epoch": 0.5201938610662359,
      "grad_norm": 0.046163927763700485,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.089,
      "step": 161
    },
    {
      "epoch": 0.5234248788368336,
      "grad_norm": 0.04039200022816658,
      "learning_rate": 1.6100000000000002e-05,
      "loss": 0.0657,
      "step": 162
    },
    {
      "epoch": 0.5266558966074314,
      "grad_norm": 0.05602607503533363,
      "learning_rate": 1.62e-05,
      "loss": 0.1044,
      "step": 163
    },
    {
      "epoch": 0.529886914378029,
      "grad_norm": 0.04312260076403618,
      "learning_rate": 1.63e-05,
      "loss": 0.0821,
      "step": 164
    },
    {
      "epoch": 0.5331179321486268,
      "grad_norm": 0.047874368727207184,
      "learning_rate": 1.64e-05,
      "loss": 0.0824,
      "step": 165
    },
    {
      "epoch": 0.5363489499192245,
      "grad_norm": 0.06421804428100586,
      "learning_rate": 1.65e-05,
      "loss": 0.1126,
      "step": 166
    },
    {
      "epoch": 0.5395799676898223,
      "grad_norm": 0.05542091280221939,
      "learning_rate": 1.66e-05,
      "loss": 0.0959,
      "step": 167
    },
    {
      "epoch": 0.5428109854604201,
      "grad_norm": 0.05822491645812988,
      "learning_rate": 1.67e-05,
      "loss": 0.099,
      "step": 168
    },
    {
      "epoch": 0.5460420032310178,
      "grad_norm": 0.04154228791594505,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.0737,
      "step": 169
    },
    {
      "epoch": 0.5492730210016155,
      "grad_norm": 0.04687827080488205,
      "learning_rate": 1.69e-05,
      "loss": 0.0725,
      "step": 170
    },
    {
      "epoch": 0.5525040387722132,
      "grad_norm": 0.056682366877794266,
      "learning_rate": 1.7e-05,
      "loss": 0.1004,
      "step": 171
    },
    {
      "epoch": 0.555735056542811,
      "grad_norm": 0.0667276531457901,
      "learning_rate": 1.7100000000000002e-05,
      "loss": 0.0917,
      "step": 172
    },
    {
      "epoch": 0.5589660743134087,
      "grad_norm": 0.044881295412778854,
      "learning_rate": 1.72e-05,
      "loss": 0.0689,
      "step": 173
    },
    {
      "epoch": 0.5621970920840065,
      "grad_norm": 0.0385456345975399,
      "learning_rate": 1.73e-05,
      "loss": 0.0592,
      "step": 174
    },
    {
      "epoch": 0.5654281098546042,
      "grad_norm": 0.05141144245862961,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 0.0895,
      "step": 175
    },
    {
      "epoch": 0.568659127625202,
      "grad_norm": 0.06854357570409775,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.0692,
      "step": 176
    },
    {
      "epoch": 0.5718901453957996,
      "grad_norm": 0.04410829395055771,
      "learning_rate": 1.76e-05,
      "loss": 0.0688,
      "step": 177
    },
    {
      "epoch": 0.5751211631663974,
      "grad_norm": 0.03727763518691063,
      "learning_rate": 1.77e-05,
      "loss": 0.0638,
      "step": 178
    },
    {
      "epoch": 0.5783521809369951,
      "grad_norm": 0.044415879994630814,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 0.0682,
      "step": 179
    },
    {
      "epoch": 0.5815831987075929,
      "grad_norm": 0.06855777651071548,
      "learning_rate": 1.79e-05,
      "loss": 0.1018,
      "step": 180
    },
    {
      "epoch": 0.5848142164781907,
      "grad_norm": 0.053684502840042114,
      "learning_rate": 1.8e-05,
      "loss": 0.0862,
      "step": 181
    },
    {
      "epoch": 0.5880452342487884,
      "grad_norm": 0.0487506277859211,
      "learning_rate": 1.8100000000000003e-05,
      "loss": 0.073,
      "step": 182
    },
    {
      "epoch": 0.5912762520193862,
      "grad_norm": 0.04568934440612793,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 0.0726,
      "step": 183
    },
    {
      "epoch": 0.5945072697899838,
      "grad_norm": 0.04607719928026199,
      "learning_rate": 1.83e-05,
      "loss": 0.0685,
      "step": 184
    },
    {
      "epoch": 0.5977382875605816,
      "grad_norm": 0.05040200799703598,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 0.0721,
      "step": 185
    },
    {
      "epoch": 0.6009693053311793,
      "grad_norm": 0.0538799948990345,
      "learning_rate": 1.8500000000000002e-05,
      "loss": 0.0769,
      "step": 186
    },
    {
      "epoch": 0.6042003231017771,
      "grad_norm": 0.058767516165971756,
      "learning_rate": 1.86e-05,
      "loss": 0.1015,
      "step": 187
    },
    {
      "epoch": 0.6074313408723748,
      "grad_norm": 0.056379787623882294,
      "learning_rate": 1.8700000000000004e-05,
      "loss": 0.0887,
      "step": 188
    },
    {
      "epoch": 0.6106623586429726,
      "grad_norm": 0.04885280132293701,
      "learning_rate": 1.88e-05,
      "loss": 0.0767,
      "step": 189
    },
    {
      "epoch": 0.6138933764135702,
      "grad_norm": 0.04340769350528717,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 0.061,
      "step": 190
    },
    {
      "epoch": 0.617124394184168,
      "grad_norm": 0.051385678350925446,
      "learning_rate": 1.9e-05,
      "loss": 0.0794,
      "step": 191
    },
    {
      "epoch": 0.6203554119547657,
      "grad_norm": 0.03737674281001091,
      "learning_rate": 1.91e-05,
      "loss": 0.0594,
      "step": 192
    },
    {
      "epoch": 0.6235864297253635,
      "grad_norm": 0.047995615750551224,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.06,
      "step": 193
    },
    {
      "epoch": 0.6268174474959612,
      "grad_norm": 0.04322716221213341,
      "learning_rate": 1.93e-05,
      "loss": 0.0712,
      "step": 194
    },
    {
      "epoch": 0.630048465266559,
      "grad_norm": 0.044713038951158524,
      "learning_rate": 1.94e-05,
      "loss": 0.0679,
      "step": 195
    },
    {
      "epoch": 0.6332794830371568,
      "grad_norm": 0.05132253095507622,
      "learning_rate": 1.95e-05,
      "loss": 0.0723,
      "step": 196
    },
    {
      "epoch": 0.6365105008077544,
      "grad_norm": 0.039808765053749084,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 0.057,
      "step": 197
    },
    {
      "epoch": 0.6397415185783522,
      "grad_norm": 0.05255698040127754,
      "learning_rate": 1.97e-05,
      "loss": 0.0774,
      "step": 198
    },
    {
      "epoch": 0.6429725363489499,
      "grad_norm": 0.05560529604554176,
      "learning_rate": 1.98e-05,
      "loss": 0.0839,
      "step": 199
    },
    {
      "epoch": 0.6462035541195477,
      "grad_norm": 0.0640430748462677,
      "learning_rate": 1.9900000000000003e-05,
      "loss": 0.0993,
      "step": 200
    },
    {
      "epoch": 0.6494345718901454,
      "grad_norm": 0.05945132300257683,
      "learning_rate": 2e-05,
      "loss": 0.0917,
      "step": 201
    },
    {
      "epoch": 0.6526655896607432,
      "grad_norm": 0.0569840632379055,
      "learning_rate": 1.9995490417136416e-05,
      "loss": 0.096,
      "step": 202
    },
    {
      "epoch": 0.6558966074313409,
      "grad_norm": 0.052612412720918655,
      "learning_rate": 1.999098083427283e-05,
      "loss": 0.079,
      "step": 203
    },
    {
      "epoch": 0.6591276252019386,
      "grad_norm": 0.05563991889357567,
      "learning_rate": 1.9986471251409248e-05,
      "loss": 0.0748,
      "step": 204
    },
    {
      "epoch": 0.6623586429725363,
      "grad_norm": 0.0527903214097023,
      "learning_rate": 1.9981961668545663e-05,
      "loss": 0.0789,
      "step": 205
    },
    {
      "epoch": 0.6655896607431341,
      "grad_norm": 0.06214462220668793,
      "learning_rate": 1.9977452085682077e-05,
      "loss": 0.0806,
      "step": 206
    },
    {
      "epoch": 0.6688206785137318,
      "grad_norm": 0.05324917659163475,
      "learning_rate": 1.997294250281849e-05,
      "loss": 0.0778,
      "step": 207
    },
    {
      "epoch": 0.6720516962843296,
      "grad_norm": 0.03270899876952171,
      "learning_rate": 1.9968432919954906e-05,
      "loss": 0.0469,
      "step": 208
    },
    {
      "epoch": 0.6752827140549273,
      "grad_norm": 0.060554295778274536,
      "learning_rate": 1.996392333709132e-05,
      "loss": 0.0926,
      "step": 209
    },
    {
      "epoch": 0.678513731825525,
      "grad_norm": 0.05076554790139198,
      "learning_rate": 1.9959413754227738e-05,
      "loss": 0.0752,
      "step": 210
    },
    {
      "epoch": 0.6817447495961227,
      "grad_norm": 0.05301008000969887,
      "learning_rate": 1.9954904171364152e-05,
      "loss": 0.0783,
      "step": 211
    },
    {
      "epoch": 0.6849757673667205,
      "grad_norm": 0.04376392439007759,
      "learning_rate": 1.9950394588500567e-05,
      "loss": 0.0603,
      "step": 212
    },
    {
      "epoch": 0.6882067851373183,
      "grad_norm": 0.049395546317100525,
      "learning_rate": 1.994588500563698e-05,
      "loss": 0.071,
      "step": 213
    },
    {
      "epoch": 0.691437802907916,
      "grad_norm": 0.061137910932302475,
      "learning_rate": 1.9941375422773395e-05,
      "loss": 0.0995,
      "step": 214
    },
    {
      "epoch": 0.6946688206785138,
      "grad_norm": 0.05603012442588806,
      "learning_rate": 1.993686583990981e-05,
      "loss": 0.0787,
      "step": 215
    },
    {
      "epoch": 0.6978998384491115,
      "grad_norm": 0.06130882352590561,
      "learning_rate": 1.9932356257046224e-05,
      "loss": 0.0745,
      "step": 216
    },
    {
      "epoch": 0.7011308562197092,
      "grad_norm": 0.06154336780309677,
      "learning_rate": 1.9927846674182642e-05,
      "loss": 0.0707,
      "step": 217
    },
    {
      "epoch": 0.7043618739903069,
      "grad_norm": 0.06344747543334961,
      "learning_rate": 1.9923337091319056e-05,
      "loss": 0.0978,
      "step": 218
    },
    {
      "epoch": 0.7075928917609047,
      "grad_norm": 0.06630375236272812,
      "learning_rate": 1.991882750845547e-05,
      "loss": 0.0721,
      "step": 219
    },
    {
      "epoch": 0.7108239095315024,
      "grad_norm": 0.04976838827133179,
      "learning_rate": 1.9914317925591885e-05,
      "loss": 0.0595,
      "step": 220
    },
    {
      "epoch": 0.7140549273021002,
      "grad_norm": 0.05968477949500084,
      "learning_rate": 1.99098083427283e-05,
      "loss": 0.0792,
      "step": 221
    },
    {
      "epoch": 0.7172859450726979,
      "grad_norm": 0.0983344092965126,
      "learning_rate": 1.9905298759864714e-05,
      "loss": 0.0872,
      "step": 222
    },
    {
      "epoch": 0.7205169628432956,
      "grad_norm": 0.08061773329973221,
      "learning_rate": 1.990078917700113e-05,
      "loss": 0.0824,
      "step": 223
    },
    {
      "epoch": 0.7237479806138933,
      "grad_norm": 0.055074963718652725,
      "learning_rate": 1.9896279594137543e-05,
      "loss": 0.0677,
      "step": 224
    },
    {
      "epoch": 0.7269789983844911,
      "grad_norm": 0.06062469258904457,
      "learning_rate": 1.9891770011273957e-05,
      "loss": 0.0658,
      "step": 225
    },
    {
      "epoch": 0.7269789983844911,
      "eval_loss": 0.08582010865211487,
      "eval_runtime": 188.2111,
      "eval_samples_per_second": 1.047,
      "eval_steps_per_second": 1.047,
      "step": 225
    },
    {
      "epoch": 0.7302100161550888,
      "grad_norm": 0.07606152445077896,
      "learning_rate": 1.988726042841037e-05,
      "loss": 0.1078,
      "step": 226
    },
    {
      "epoch": 0.7334410339256866,
      "grad_norm": 0.06171920895576477,
      "learning_rate": 1.988275084554679e-05,
      "loss": 0.086,
      "step": 227
    },
    {
      "epoch": 0.7366720516962844,
      "grad_norm": 0.03934045881032944,
      "learning_rate": 1.9878241262683204e-05,
      "loss": 0.0516,
      "step": 228
    },
    {
      "epoch": 0.7399030694668821,
      "grad_norm": 0.0552021786570549,
      "learning_rate": 1.9873731679819618e-05,
      "loss": 0.0761,
      "step": 229
    },
    {
      "epoch": 0.7431340872374798,
      "grad_norm": 0.05151893198490143,
      "learning_rate": 1.9869222096956032e-05,
      "loss": 0.0726,
      "step": 230
    },
    {
      "epoch": 0.7463651050080775,
      "grad_norm": 0.0533306822180748,
      "learning_rate": 1.9864712514092447e-05,
      "loss": 0.0717,
      "step": 231
    },
    {
      "epoch": 0.7495961227786753,
      "grad_norm": 0.052841685712337494,
      "learning_rate": 1.986020293122886e-05,
      "loss": 0.0755,
      "step": 232
    },
    {
      "epoch": 0.752827140549273,
      "grad_norm": 0.040998924523591995,
      "learning_rate": 1.9855693348365276e-05,
      "loss": 0.0554,
      "step": 233
    },
    {
      "epoch": 0.7560581583198708,
      "grad_norm": 0.057859815657138824,
      "learning_rate": 1.9851183765501693e-05,
      "loss": 0.0719,
      "step": 234
    },
    {
      "epoch": 0.7592891760904685,
      "grad_norm": 0.04167502373456955,
      "learning_rate": 1.9846674182638108e-05,
      "loss": 0.0625,
      "step": 235
    },
    {
      "epoch": 0.7625201938610663,
      "grad_norm": 0.058570049703121185,
      "learning_rate": 1.9842164599774522e-05,
      "loss": 0.0773,
      "step": 236
    },
    {
      "epoch": 0.7657512116316639,
      "grad_norm": 0.06181475892663002,
      "learning_rate": 1.9837655016910937e-05,
      "loss": 0.083,
      "step": 237
    },
    {
      "epoch": 0.7689822294022617,
      "grad_norm": 0.06188640370965004,
      "learning_rate": 1.983314543404735e-05,
      "loss": 0.0838,
      "step": 238
    },
    {
      "epoch": 0.7722132471728594,
      "grad_norm": 0.0784875750541687,
      "learning_rate": 1.9828635851183765e-05,
      "loss": 0.1032,
      "step": 239
    },
    {
      "epoch": 0.7754442649434572,
      "grad_norm": 0.06791771203279495,
      "learning_rate": 1.9824126268320183e-05,
      "loss": 0.0888,
      "step": 240
    },
    {
      "epoch": 0.778675282714055,
      "grad_norm": 0.06790699809789658,
      "learning_rate": 1.9819616685456598e-05,
      "loss": 0.0919,
      "step": 241
    },
    {
      "epoch": 0.7819063004846527,
      "grad_norm": 0.04812704026699066,
      "learning_rate": 1.9815107102593012e-05,
      "loss": 0.058,
      "step": 242
    },
    {
      "epoch": 0.7851373182552503,
      "grad_norm": 0.0617465041577816,
      "learning_rate": 1.9810597519729426e-05,
      "loss": 0.0852,
      "step": 243
    },
    {
      "epoch": 0.7883683360258481,
      "grad_norm": 0.049844078719615936,
      "learning_rate": 1.980608793686584e-05,
      "loss": 0.0668,
      "step": 244
    },
    {
      "epoch": 0.7915993537964459,
      "grad_norm": 0.0725836232304573,
      "learning_rate": 1.9801578354002255e-05,
      "loss": 0.0929,
      "step": 245
    },
    {
      "epoch": 0.7948303715670436,
      "grad_norm": 0.0587320439517498,
      "learning_rate": 1.9797068771138673e-05,
      "loss": 0.0764,
      "step": 246
    },
    {
      "epoch": 0.7980613893376414,
      "grad_norm": 0.04824487864971161,
      "learning_rate": 1.9792559188275087e-05,
      "loss": 0.0562,
      "step": 247
    },
    {
      "epoch": 0.8012924071082391,
      "grad_norm": 0.06668104231357574,
      "learning_rate": 1.97880496054115e-05,
      "loss": 0.0842,
      "step": 248
    },
    {
      "epoch": 0.8045234248788369,
      "grad_norm": 0.057721976190805435,
      "learning_rate": 1.9783540022547916e-05,
      "loss": 0.0838,
      "step": 249
    },
    {
      "epoch": 0.8077544426494345,
      "grad_norm": 0.07014774531126022,
      "learning_rate": 1.977903043968433e-05,
      "loss": 0.0831,
      "step": 250
    },
    {
      "epoch": 0.8109854604200323,
      "grad_norm": 0.0693356841802597,
      "learning_rate": 1.9774520856820745e-05,
      "loss": 0.0903,
      "step": 251
    },
    {
      "epoch": 0.81421647819063,
      "grad_norm": 0.05464401841163635,
      "learning_rate": 1.9770011273957163e-05,
      "loss": 0.0682,
      "step": 252
    },
    {
      "epoch": 0.8174474959612278,
      "grad_norm": 0.053677983582019806,
      "learning_rate": 1.9765501691093577e-05,
      "loss": 0.0588,
      "step": 253
    },
    {
      "epoch": 0.8206785137318255,
      "grad_norm": 0.04152385890483856,
      "learning_rate": 1.976099210822999e-05,
      "loss": 0.0564,
      "step": 254
    },
    {
      "epoch": 0.8239095315024233,
      "grad_norm": 0.06150719150900841,
      "learning_rate": 1.9756482525366406e-05,
      "loss": 0.0694,
      "step": 255
    },
    {
      "epoch": 0.827140549273021,
      "grad_norm": 0.05433864891529083,
      "learning_rate": 1.975197294250282e-05,
      "loss": 0.067,
      "step": 256
    },
    {
      "epoch": 0.8303715670436187,
      "grad_norm": 0.04325372353196144,
      "learning_rate": 1.9747463359639235e-05,
      "loss": 0.0575,
      "step": 257
    },
    {
      "epoch": 0.8336025848142165,
      "grad_norm": 0.049097690731287,
      "learning_rate": 1.9742953776775652e-05,
      "loss": 0.0571,
      "step": 258
    },
    {
      "epoch": 0.8368336025848142,
      "grad_norm": 0.06370379030704498,
      "learning_rate": 1.9738444193912067e-05,
      "loss": 0.0877,
      "step": 259
    },
    {
      "epoch": 0.840064620355412,
      "grad_norm": 0.05573710799217224,
      "learning_rate": 1.973393461104848e-05,
      "loss": 0.0759,
      "step": 260
    },
    {
      "epoch": 0.8432956381260097,
      "grad_norm": 0.06537079066038132,
      "learning_rate": 1.9729425028184896e-05,
      "loss": 0.0759,
      "step": 261
    },
    {
      "epoch": 0.8465266558966075,
      "grad_norm": 0.04301934316754341,
      "learning_rate": 1.972491544532131e-05,
      "loss": 0.0564,
      "step": 262
    },
    {
      "epoch": 0.8497576736672051,
      "grad_norm": 0.07281677424907684,
      "learning_rate": 1.9720405862457724e-05,
      "loss": 0.0685,
      "step": 263
    },
    {
      "epoch": 0.8529886914378029,
      "grad_norm": 0.06059825047850609,
      "learning_rate": 1.971589627959414e-05,
      "loss": 0.071,
      "step": 264
    },
    {
      "epoch": 0.8562197092084006,
      "grad_norm": 0.05605108663439751,
      "learning_rate": 1.9711386696730553e-05,
      "loss": 0.0634,
      "step": 265
    },
    {
      "epoch": 0.8594507269789984,
      "grad_norm": 0.07372546941041946,
      "learning_rate": 1.9706877113866967e-05,
      "loss": 0.0916,
      "step": 266
    },
    {
      "epoch": 0.8626817447495961,
      "grad_norm": 0.051352906972169876,
      "learning_rate": 1.9702367531003382e-05,
      "loss": 0.0653,
      "step": 267
    },
    {
      "epoch": 0.8659127625201939,
      "grad_norm": 0.059334397315979004,
      "learning_rate": 1.96978579481398e-05,
      "loss": 0.0717,
      "step": 268
    },
    {
      "epoch": 0.8691437802907916,
      "grad_norm": 0.06220857426524162,
      "learning_rate": 1.9693348365276214e-05,
      "loss": 0.0612,
      "step": 269
    },
    {
      "epoch": 0.8723747980613893,
      "grad_norm": 0.053203944116830826,
      "learning_rate": 1.968883878241263e-05,
      "loss": 0.0699,
      "step": 270
    },
    {
      "epoch": 0.875605815831987,
      "grad_norm": 0.06943807750940323,
      "learning_rate": 1.9684329199549043e-05,
      "loss": 0.0793,
      "step": 271
    },
    {
      "epoch": 0.8788368336025848,
      "grad_norm": 0.07023902982473373,
      "learning_rate": 1.9679819616685457e-05,
      "loss": 0.0859,
      "step": 272
    },
    {
      "epoch": 0.8820678513731826,
      "grad_norm": 0.06727661192417145,
      "learning_rate": 1.967531003382187e-05,
      "loss": 0.0796,
      "step": 273
    },
    {
      "epoch": 0.8852988691437803,
      "grad_norm": 0.08004336804151535,
      "learning_rate": 1.9670800450958286e-05,
      "loss": 0.0989,
      "step": 274
    },
    {
      "epoch": 0.8885298869143781,
      "grad_norm": 0.06687738746404648,
      "learning_rate": 1.96662908680947e-05,
      "loss": 0.0674,
      "step": 275
    },
    {
      "epoch": 0.8917609046849758,
      "grad_norm": 0.06280867755413055,
      "learning_rate": 1.9661781285231118e-05,
      "loss": 0.0735,
      "step": 276
    },
    {
      "epoch": 0.8949919224555735,
      "grad_norm": 0.06883740425109863,
      "learning_rate": 1.9657271702367533e-05,
      "loss": 0.0893,
      "step": 277
    },
    {
      "epoch": 0.8982229402261712,
      "grad_norm": 0.059292376041412354,
      "learning_rate": 1.9652762119503947e-05,
      "loss": 0.0819,
      "step": 278
    },
    {
      "epoch": 0.901453957996769,
      "grad_norm": 0.0578530877828598,
      "learning_rate": 1.964825253664036e-05,
      "loss": 0.0724,
      "step": 279
    },
    {
      "epoch": 0.9046849757673667,
      "grad_norm": 0.08932427316904068,
      "learning_rate": 1.9643742953776776e-05,
      "loss": 0.0939,
      "step": 280
    },
    {
      "epoch": 0.9079159935379645,
      "grad_norm": 0.07406419515609741,
      "learning_rate": 1.963923337091319e-05,
      "loss": 0.0868,
      "step": 281
    },
    {
      "epoch": 0.9111470113085622,
      "grad_norm": 0.05354011803865433,
      "learning_rate": 1.9634723788049608e-05,
      "loss": 0.0685,
      "step": 282
    },
    {
      "epoch": 0.9143780290791599,
      "grad_norm": 0.06414072960615158,
      "learning_rate": 1.9630214205186022e-05,
      "loss": 0.0686,
      "step": 283
    },
    {
      "epoch": 0.9176090468497576,
      "grad_norm": 0.058192793279886246,
      "learning_rate": 1.9625704622322437e-05,
      "loss": 0.0674,
      "step": 284
    },
    {
      "epoch": 0.9208400646203554,
      "grad_norm": 0.10264746099710464,
      "learning_rate": 1.962119503945885e-05,
      "loss": 0.1152,
      "step": 285
    },
    {
      "epoch": 0.9240710823909531,
      "grad_norm": 0.066757932305336,
      "learning_rate": 1.9616685456595265e-05,
      "loss": 0.0757,
      "step": 286
    },
    {
      "epoch": 0.9273021001615509,
      "grad_norm": 0.06598404794931412,
      "learning_rate": 1.9612175873731683e-05,
      "loss": 0.0768,
      "step": 287
    },
    {
      "epoch": 0.9305331179321487,
      "grad_norm": 0.07162454724311829,
      "learning_rate": 1.9607666290868098e-05,
      "loss": 0.081,
      "step": 288
    },
    {
      "epoch": 0.9337641357027464,
      "grad_norm": 0.05917588993906975,
      "learning_rate": 1.9603156708004512e-05,
      "loss": 0.0645,
      "step": 289
    },
    {
      "epoch": 0.9369951534733441,
      "grad_norm": 0.06051475182175636,
      "learning_rate": 1.9598647125140926e-05,
      "loss": 0.0656,
      "step": 290
    },
    {
      "epoch": 0.9402261712439418,
      "grad_norm": 0.06452775001525879,
      "learning_rate": 1.959413754227734e-05,
      "loss": 0.0704,
      "step": 291
    },
    {
      "epoch": 0.9434571890145396,
      "grad_norm": 0.06445769965648651,
      "learning_rate": 1.9589627959413755e-05,
      "loss": 0.0759,
      "step": 292
    },
    {
      "epoch": 0.9466882067851373,
      "grad_norm": 0.06948834657669067,
      "learning_rate": 1.9585118376550173e-05,
      "loss": 0.0776,
      "step": 293
    },
    {
      "epoch": 0.9499192245557351,
      "grad_norm": 0.05026319995522499,
      "learning_rate": 1.9580608793686587e-05,
      "loss": 0.0574,
      "step": 294
    },
    {
      "epoch": 0.9531502423263328,
      "grad_norm": 0.08733383566141129,
      "learning_rate": 1.9576099210823002e-05,
      "loss": 0.0898,
      "step": 295
    },
    {
      "epoch": 0.9563812600969306,
      "grad_norm": 0.05138668045401573,
      "learning_rate": 1.9571589627959416e-05,
      "loss": 0.0525,
      "step": 296
    },
    {
      "epoch": 0.9596122778675282,
      "grad_norm": 0.0710444375872612,
      "learning_rate": 1.956708004509583e-05,
      "loss": 0.0832,
      "step": 297
    },
    {
      "epoch": 0.962843295638126,
      "grad_norm": 0.06288463622331619,
      "learning_rate": 1.9562570462232245e-05,
      "loss": 0.0673,
      "step": 298
    },
    {
      "epoch": 0.9660743134087237,
      "grad_norm": 0.05722356587648392,
      "learning_rate": 1.9558060879368663e-05,
      "loss": 0.0702,
      "step": 299
    },
    {
      "epoch": 0.9693053311793215,
      "grad_norm": 0.07167758047580719,
      "learning_rate": 1.9553551296505077e-05,
      "loss": 0.0809,
      "step": 300
    },
    {
      "epoch": 0.9693053311793215,
      "eval_loss": 0.08212888240814209,
      "eval_runtime": 188.3411,
      "eval_samples_per_second": 1.046,
      "eval_steps_per_second": 1.046,
      "step": 300
    },
    {
      "epoch": 0.9725363489499192,
      "grad_norm": 0.04536513611674309,
      "learning_rate": 1.954904171364149e-05,
      "loss": 0.0485,
      "step": 301
    },
    {
      "epoch": 0.975767366720517,
      "grad_norm": 0.07035136222839355,
      "learning_rate": 1.9544532130777906e-05,
      "loss": 0.0763,
      "step": 302
    },
    {
      "epoch": 0.9789983844911146,
      "grad_norm": 0.06417107582092285,
      "learning_rate": 1.954002254791432e-05,
      "loss": 0.0735,
      "step": 303
    },
    {
      "epoch": 0.9822294022617124,
      "grad_norm": 0.06369137018918991,
      "learning_rate": 1.9535512965050735e-05,
      "loss": 0.07,
      "step": 304
    },
    {
      "epoch": 0.9854604200323102,
      "grad_norm": 0.053664304316043854,
      "learning_rate": 1.953100338218715e-05,
      "loss": 0.058,
      "step": 305
    },
    {
      "epoch": 0.9886914378029079,
      "grad_norm": 0.07393426448106766,
      "learning_rate": 1.9526493799323563e-05,
      "loss": 0.0783,
      "step": 306
    },
    {
      "epoch": 0.9919224555735057,
      "grad_norm": 0.05141662806272507,
      "learning_rate": 1.9521984216459978e-05,
      "loss": 0.0556,
      "step": 307
    },
    {
      "epoch": 0.9951534733441034,
      "grad_norm": 0.06411275267601013,
      "learning_rate": 1.9517474633596392e-05,
      "loss": 0.0687,
      "step": 308
    },
    {
      "epoch": 0.9983844911147012,
      "grad_norm": 0.08877477794885635,
      "learning_rate": 1.951296505073281e-05,
      "loss": 0.0846,
      "step": 309
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.08283062279224396,
      "learning_rate": 1.9508455467869224e-05,
      "loss": 0.0541,
      "step": 310
    },
    {
      "epoch": 1.0032310177705976,
      "grad_norm": 0.06769707798957825,
      "learning_rate": 1.950394588500564e-05,
      "loss": 0.0752,
      "step": 311
    },
    {
      "epoch": 1.0064620355411955,
      "grad_norm": 0.06125921383500099,
      "learning_rate": 1.9499436302142053e-05,
      "loss": 0.0719,
      "step": 312
    },
    {
      "epoch": 1.0096930533117932,
      "grad_norm": 0.03994071111083031,
      "learning_rate": 1.9494926719278468e-05,
      "loss": 0.0421,
      "step": 313
    },
    {
      "epoch": 1.012924071082391,
      "grad_norm": 0.05433064326643944,
      "learning_rate": 1.9490417136414882e-05,
      "loss": 0.066,
      "step": 314
    },
    {
      "epoch": 1.0161550888529887,
      "grad_norm": 0.06107380986213684,
      "learning_rate": 1.9485907553551296e-05,
      "loss": 0.0724,
      "step": 315
    },
    {
      "epoch": 1.0193861066235865,
      "grad_norm": 0.0669042244553566,
      "learning_rate": 1.948139797068771e-05,
      "loss": 0.0772,
      "step": 316
    },
    {
      "epoch": 1.0226171243941842,
      "grad_norm": 0.0474565327167511,
      "learning_rate": 1.947688838782413e-05,
      "loss": 0.0491,
      "step": 317
    },
    {
      "epoch": 1.0258481421647818,
      "grad_norm": 0.054098691791296005,
      "learning_rate": 1.9472378804960543e-05,
      "loss": 0.0618,
      "step": 318
    },
    {
      "epoch": 1.0290791599353797,
      "grad_norm": 0.06151336431503296,
      "learning_rate": 1.9467869222096957e-05,
      "loss": 0.0604,
      "step": 319
    },
    {
      "epoch": 1.0323101777059773,
      "grad_norm": 0.051618464291095734,
      "learning_rate": 1.946335963923337e-05,
      "loss": 0.0551,
      "step": 320
    },
    {
      "epoch": 1.0355411954765752,
      "grad_norm": 0.08121399581432343,
      "learning_rate": 1.9458850056369786e-05,
      "loss": 0.0939,
      "step": 321
    },
    {
      "epoch": 1.0387722132471728,
      "grad_norm": 0.05889379233121872,
      "learning_rate": 1.94543404735062e-05,
      "loss": 0.0687,
      "step": 322
    },
    {
      "epoch": 1.0420032310177707,
      "grad_norm": 0.06208242103457451,
      "learning_rate": 1.9449830890642618e-05,
      "loss": 0.0744,
      "step": 323
    },
    {
      "epoch": 1.0452342487883683,
      "grad_norm": 0.06454786658287048,
      "learning_rate": 1.9445321307779033e-05,
      "loss": 0.0684,
      "step": 324
    },
    {
      "epoch": 1.048465266558966,
      "grad_norm": 0.07085470855236053,
      "learning_rate": 1.9440811724915447e-05,
      "loss": 0.0727,
      "step": 325
    },
    {
      "epoch": 1.0516962843295639,
      "grad_norm": 0.07236117869615555,
      "learning_rate": 1.943630214205186e-05,
      "loss": 0.0892,
      "step": 326
    },
    {
      "epoch": 1.0549273021001615,
      "grad_norm": 0.054056137800216675,
      "learning_rate": 1.9431792559188276e-05,
      "loss": 0.0634,
      "step": 327
    },
    {
      "epoch": 1.0581583198707594,
      "grad_norm": 0.05462612211704254,
      "learning_rate": 1.942728297632469e-05,
      "loss": 0.0631,
      "step": 328
    },
    {
      "epoch": 1.061389337641357,
      "grad_norm": 0.0674949586391449,
      "learning_rate": 1.9422773393461108e-05,
      "loss": 0.0641,
      "step": 329
    },
    {
      "epoch": 1.0646203554119547,
      "grad_norm": 0.07532529532909393,
      "learning_rate": 1.9418263810597522e-05,
      "loss": 0.0765,
      "step": 330
    },
    {
      "epoch": 1.0678513731825525,
      "grad_norm": 0.06264142692089081,
      "learning_rate": 1.9413754227733937e-05,
      "loss": 0.0704,
      "step": 331
    },
    {
      "epoch": 1.0710823909531502,
      "grad_norm": 0.06789285689592361,
      "learning_rate": 1.940924464487035e-05,
      "loss": 0.0667,
      "step": 332
    },
    {
      "epoch": 1.074313408723748,
      "grad_norm": 0.06181450933218002,
      "learning_rate": 1.9404735062006765e-05,
      "loss": 0.0648,
      "step": 333
    },
    {
      "epoch": 1.0775444264943457,
      "grad_norm": 0.07014179229736328,
      "learning_rate": 1.940022547914318e-05,
      "loss": 0.076,
      "step": 334
    },
    {
      "epoch": 1.0807754442649435,
      "grad_norm": 0.07433414459228516,
      "learning_rate": 1.9395715896279598e-05,
      "loss": 0.0757,
      "step": 335
    },
    {
      "epoch": 1.0840064620355412,
      "grad_norm": 0.04758503660559654,
      "learning_rate": 1.9391206313416012e-05,
      "loss": 0.0492,
      "step": 336
    },
    {
      "epoch": 1.0872374798061388,
      "grad_norm": 0.06751306354999542,
      "learning_rate": 1.9386696730552426e-05,
      "loss": 0.0682,
      "step": 337
    },
    {
      "epoch": 1.0904684975767367,
      "grad_norm": 0.06028216332197189,
      "learning_rate": 1.938218714768884e-05,
      "loss": 0.059,
      "step": 338
    },
    {
      "epoch": 1.0936995153473343,
      "grad_norm": 0.060358040034770966,
      "learning_rate": 1.9377677564825255e-05,
      "loss": 0.0659,
      "step": 339
    },
    {
      "epoch": 1.0969305331179322,
      "grad_norm": 0.06687436252832413,
      "learning_rate": 1.937316798196167e-05,
      "loss": 0.0517,
      "step": 340
    },
    {
      "epoch": 1.1001615508885298,
      "grad_norm": 0.07463373243808746,
      "learning_rate": 1.9368658399098087e-05,
      "loss": 0.0674,
      "step": 341
    },
    {
      "epoch": 1.1033925686591277,
      "grad_norm": 0.06248531863093376,
      "learning_rate": 1.9364148816234502e-05,
      "loss": 0.0681,
      "step": 342
    },
    {
      "epoch": 1.1066235864297254,
      "grad_norm": 0.06864578276872635,
      "learning_rate": 1.9359639233370916e-05,
      "loss": 0.0703,
      "step": 343
    },
    {
      "epoch": 1.109854604200323,
      "grad_norm": 0.0693066269159317,
      "learning_rate": 1.935512965050733e-05,
      "loss": 0.0599,
      "step": 344
    },
    {
      "epoch": 1.1130856219709209,
      "grad_norm": 0.13610310852527618,
      "learning_rate": 1.9350620067643745e-05,
      "loss": 0.0789,
      "step": 345
    },
    {
      "epoch": 1.1163166397415185,
      "grad_norm": 0.09487364441156387,
      "learning_rate": 1.934611048478016e-05,
      "loss": 0.094,
      "step": 346
    },
    {
      "epoch": 1.1195476575121164,
      "grad_norm": 0.0767926499247551,
      "learning_rate": 1.9341600901916574e-05,
      "loss": 0.0751,
      "step": 347
    },
    {
      "epoch": 1.122778675282714,
      "grad_norm": 0.1105605959892273,
      "learning_rate": 1.9337091319052988e-05,
      "loss": 0.0908,
      "step": 348
    },
    {
      "epoch": 1.1260096930533119,
      "grad_norm": 0.06821838766336441,
      "learning_rate": 1.9332581736189403e-05,
      "loss": 0.0702,
      "step": 349
    },
    {
      "epoch": 1.1292407108239095,
      "grad_norm": 0.07123742997646332,
      "learning_rate": 1.932807215332582e-05,
      "loss": 0.0637,
      "step": 350
    },
    {
      "epoch": 1.1324717285945072,
      "grad_norm": 0.08340942859649658,
      "learning_rate": 1.9323562570462235e-05,
      "loss": 0.0803,
      "step": 351
    },
    {
      "epoch": 1.135702746365105,
      "grad_norm": 0.06730187684297562,
      "learning_rate": 1.931905298759865e-05,
      "loss": 0.0644,
      "step": 352
    },
    {
      "epoch": 1.1389337641357027,
      "grad_norm": 0.06728731095790863,
      "learning_rate": 1.9314543404735063e-05,
      "loss": 0.0633,
      "step": 353
    },
    {
      "epoch": 1.1421647819063006,
      "grad_norm": 0.07192697376012802,
      "learning_rate": 1.9310033821871478e-05,
      "loss": 0.0664,
      "step": 354
    },
    {
      "epoch": 1.1453957996768982,
      "grad_norm": 0.07150010764598846,
      "learning_rate": 1.9305524239007892e-05,
      "loss": 0.0745,
      "step": 355
    },
    {
      "epoch": 1.148626817447496,
      "grad_norm": 0.05815986543893814,
      "learning_rate": 1.9301014656144307e-05,
      "loss": 0.0501,
      "step": 356
    },
    {
      "epoch": 1.1518578352180937,
      "grad_norm": 0.063558429479599,
      "learning_rate": 1.929650507328072e-05,
      "loss": 0.064,
      "step": 357
    },
    {
      "epoch": 1.1550888529886914,
      "grad_norm": 0.08062389492988586,
      "learning_rate": 1.9291995490417135e-05,
      "loss": 0.0792,
      "step": 358
    },
    {
      "epoch": 1.1583198707592892,
      "grad_norm": 0.06872212886810303,
      "learning_rate": 1.9287485907553553e-05,
      "loss": 0.0756,
      "step": 359
    },
    {
      "epoch": 1.1615508885298869,
      "grad_norm": 0.06003013253211975,
      "learning_rate": 1.9282976324689968e-05,
      "loss": 0.0577,
      "step": 360
    },
    {
      "epoch": 1.1647819063004847,
      "grad_norm": 0.07533125579357147,
      "learning_rate": 1.9278466741826382e-05,
      "loss": 0.0745,
      "step": 361
    },
    {
      "epoch": 1.1680129240710824,
      "grad_norm": 0.0708516389131546,
      "learning_rate": 1.9273957158962796e-05,
      "loss": 0.0671,
      "step": 362
    },
    {
      "epoch": 1.1712439418416802,
      "grad_norm": 0.10226985812187195,
      "learning_rate": 1.926944757609921e-05,
      "loss": 0.1126,
      "step": 363
    },
    {
      "epoch": 1.1744749596122779,
      "grad_norm": 0.067733995616436,
      "learning_rate": 1.9264937993235625e-05,
      "loss": 0.0554,
      "step": 364
    },
    {
      "epoch": 1.1777059773828755,
      "grad_norm": 0.08708222955465317,
      "learning_rate": 1.9260428410372043e-05,
      "loss": 0.0806,
      "step": 365
    },
    {
      "epoch": 1.1809369951534734,
      "grad_norm": 0.06153462454676628,
      "learning_rate": 1.9255918827508457e-05,
      "loss": 0.0532,
      "step": 366
    },
    {
      "epoch": 1.184168012924071,
      "grad_norm": 0.051941219717264175,
      "learning_rate": 1.9251409244644872e-05,
      "loss": 0.0503,
      "step": 367
    },
    {
      "epoch": 1.187399030694669,
      "grad_norm": 0.09817774593830109,
      "learning_rate": 1.9246899661781286e-05,
      "loss": 0.0801,
      "step": 368
    },
    {
      "epoch": 1.1906300484652665,
      "grad_norm": 0.08504205197095871,
      "learning_rate": 1.92423900789177e-05,
      "loss": 0.08,
      "step": 369
    },
    {
      "epoch": 1.1938610662358644,
      "grad_norm": 0.0611301064491272,
      "learning_rate": 1.9237880496054118e-05,
      "loss": 0.061,
      "step": 370
    },
    {
      "epoch": 1.197092084006462,
      "grad_norm": 0.06038827449083328,
      "learning_rate": 1.9233370913190533e-05,
      "loss": 0.0512,
      "step": 371
    },
    {
      "epoch": 1.2003231017770597,
      "grad_norm": 0.08283468335866928,
      "learning_rate": 1.9228861330326947e-05,
      "loss": 0.0753,
      "step": 372
    },
    {
      "epoch": 1.2035541195476576,
      "grad_norm": 0.09547346830368042,
      "learning_rate": 1.922435174746336e-05,
      "loss": 0.0809,
      "step": 373
    },
    {
      "epoch": 1.2067851373182552,
      "grad_norm": 0.058611780405044556,
      "learning_rate": 1.9219842164599776e-05,
      "loss": 0.0563,
      "step": 374
    },
    {
      "epoch": 1.210016155088853,
      "grad_norm": 0.08549389988183975,
      "learning_rate": 1.921533258173619e-05,
      "loss": 0.0733,
      "step": 375
    },
    {
      "epoch": 1.210016155088853,
      "eval_loss": 0.0792868584394455,
      "eval_runtime": 188.2838,
      "eval_samples_per_second": 1.046,
      "eval_steps_per_second": 1.046,
      "step": 375
    },
    {
      "epoch": 1.2132471728594507,
      "grad_norm": 0.08417635411024094,
      "learning_rate": 1.9210822998872608e-05,
      "loss": 0.0841,
      "step": 376
    },
    {
      "epoch": 1.2164781906300486,
      "grad_norm": 0.08157463371753693,
      "learning_rate": 1.9206313416009022e-05,
      "loss": 0.0714,
      "step": 377
    },
    {
      "epoch": 1.2197092084006462,
      "grad_norm": 0.05649822950363159,
      "learning_rate": 1.9201803833145437e-05,
      "loss": 0.0503,
      "step": 378
    },
    {
      "epoch": 1.2229402261712439,
      "grad_norm": 0.07617928087711334,
      "learning_rate": 1.919729425028185e-05,
      "loss": 0.0727,
      "step": 379
    },
    {
      "epoch": 1.2261712439418417,
      "grad_norm": 0.0574098639190197,
      "learning_rate": 1.9192784667418266e-05,
      "loss": 0.0506,
      "step": 380
    },
    {
      "epoch": 1.2294022617124394,
      "grad_norm": 0.07354257255792618,
      "learning_rate": 1.918827508455468e-05,
      "loss": 0.0728,
      "step": 381
    },
    {
      "epoch": 1.2326332794830372,
      "grad_norm": 0.07268121838569641,
      "learning_rate": 1.9183765501691098e-05,
      "loss": 0.0679,
      "step": 382
    },
    {
      "epoch": 1.235864297253635,
      "grad_norm": 0.07641527056694031,
      "learning_rate": 1.9179255918827512e-05,
      "loss": 0.0663,
      "step": 383
    },
    {
      "epoch": 1.2390953150242328,
      "grad_norm": 0.059996772557497025,
      "learning_rate": 1.9174746335963926e-05,
      "loss": 0.0523,
      "step": 384
    },
    {
      "epoch": 1.2423263327948304,
      "grad_norm": 0.07397306710481644,
      "learning_rate": 1.917023675310034e-05,
      "loss": 0.0662,
      "step": 385
    },
    {
      "epoch": 1.245557350565428,
      "grad_norm": 0.09324625134468079,
      "learning_rate": 1.9165727170236755e-05,
      "loss": 0.083,
      "step": 386
    },
    {
      "epoch": 1.248788368336026,
      "grad_norm": 0.08019818365573883,
      "learning_rate": 1.916121758737317e-05,
      "loss": 0.0682,
      "step": 387
    },
    {
      "epoch": 1.2520193861066236,
      "grad_norm": 0.08203406631946564,
      "learning_rate": 1.9156708004509584e-05,
      "loss": 0.0788,
      "step": 388
    },
    {
      "epoch": 1.2552504038772212,
      "grad_norm": 0.07293461263179779,
      "learning_rate": 1.9152198421646e-05,
      "loss": 0.0583,
      "step": 389
    },
    {
      "epoch": 1.258481421647819,
      "grad_norm": 0.07020010054111481,
      "learning_rate": 1.9147688838782413e-05,
      "loss": 0.0546,
      "step": 390
    },
    {
      "epoch": 1.261712439418417,
      "grad_norm": 0.0655217245221138,
      "learning_rate": 1.914317925591883e-05,
      "loss": 0.0561,
      "step": 391
    },
    {
      "epoch": 1.2649434571890146,
      "grad_norm": 0.0773930773139,
      "learning_rate": 1.9138669673055245e-05,
      "loss": 0.0805,
      "step": 392
    },
    {
      "epoch": 1.2681744749596122,
      "grad_norm": 0.06243716925382614,
      "learning_rate": 1.913416009019166e-05,
      "loss": 0.0573,
      "step": 393
    },
    {
      "epoch": 1.27140549273021,
      "grad_norm": 0.07922864705324173,
      "learning_rate": 1.9129650507328074e-05,
      "loss": 0.0679,
      "step": 394
    },
    {
      "epoch": 1.2746365105008077,
      "grad_norm": 0.10133316367864609,
      "learning_rate": 1.9125140924464488e-05,
      "loss": 0.1135,
      "step": 395
    },
    {
      "epoch": 1.2778675282714054,
      "grad_norm": 0.0727897360920906,
      "learning_rate": 1.9120631341600903e-05,
      "loss": 0.0658,
      "step": 396
    },
    {
      "epoch": 1.2810985460420032,
      "grad_norm": 0.0690392330288887,
      "learning_rate": 1.9116121758737317e-05,
      "loss": 0.0656,
      "step": 397
    },
    {
      "epoch": 1.284329563812601,
      "grad_norm": 0.062050607055425644,
      "learning_rate": 1.911161217587373e-05,
      "loss": 0.0542,
      "step": 398
    },
    {
      "epoch": 1.2875605815831987,
      "grad_norm": 0.0690266340970993,
      "learning_rate": 1.9107102593010146e-05,
      "loss": 0.065,
      "step": 399
    },
    {
      "epoch": 1.2907915993537964,
      "grad_norm": 0.07588627934455872,
      "learning_rate": 1.9102593010146564e-05,
      "loss": 0.0761,
      "step": 400
    },
    {
      "epoch": 1.2940226171243943,
      "grad_norm": 0.07368933409452438,
      "learning_rate": 1.9098083427282978e-05,
      "loss": 0.0637,
      "step": 401
    },
    {
      "epoch": 1.297253634894992,
      "grad_norm": 0.0670572817325592,
      "learning_rate": 1.9093573844419392e-05,
      "loss": 0.0656,
      "step": 402
    },
    {
      "epoch": 1.3004846526655895,
      "grad_norm": 0.06778164952993393,
      "learning_rate": 1.9089064261555807e-05,
      "loss": 0.0577,
      "step": 403
    },
    {
      "epoch": 1.3037156704361874,
      "grad_norm": 0.10589181631803513,
      "learning_rate": 1.908455467869222e-05,
      "loss": 0.0849,
      "step": 404
    },
    {
      "epoch": 1.306946688206785,
      "grad_norm": 0.060739271342754364,
      "learning_rate": 1.9080045095828635e-05,
      "loss": 0.0563,
      "step": 405
    },
    {
      "epoch": 1.310177705977383,
      "grad_norm": 0.062488917261362076,
      "learning_rate": 1.9075535512965053e-05,
      "loss": 0.0555,
      "step": 406
    },
    {
      "epoch": 1.3134087237479806,
      "grad_norm": 0.08088962733745575,
      "learning_rate": 1.9071025930101468e-05,
      "loss": 0.0683,
      "step": 407
    },
    {
      "epoch": 1.3166397415185784,
      "grad_norm": 0.07679299265146255,
      "learning_rate": 1.9066516347237882e-05,
      "loss": 0.0713,
      "step": 408
    },
    {
      "epoch": 1.319870759289176,
      "grad_norm": 0.09731165319681168,
      "learning_rate": 1.9062006764374296e-05,
      "loss": 0.094,
      "step": 409
    },
    {
      "epoch": 1.3231017770597737,
      "grad_norm": 0.09488274902105331,
      "learning_rate": 1.905749718151071e-05,
      "loss": 0.0855,
      "step": 410
    },
    {
      "epoch": 1.3263327948303716,
      "grad_norm": 0.08556380867958069,
      "learning_rate": 1.9052987598647125e-05,
      "loss": 0.0872,
      "step": 411
    },
    {
      "epoch": 1.3295638126009692,
      "grad_norm": 0.0821579322218895,
      "learning_rate": 1.9048478015783543e-05,
      "loss": 0.0802,
      "step": 412
    },
    {
      "epoch": 1.332794830371567,
      "grad_norm": 0.08073097467422485,
      "learning_rate": 1.9043968432919957e-05,
      "loss": 0.0764,
      "step": 413
    },
    {
      "epoch": 1.3360258481421647,
      "grad_norm": 0.07449216395616531,
      "learning_rate": 1.9039458850056372e-05,
      "loss": 0.0681,
      "step": 414
    },
    {
      "epoch": 1.3392568659127626,
      "grad_norm": 0.05690048635005951,
      "learning_rate": 1.9034949267192786e-05,
      "loss": 0.0479,
      "step": 415
    },
    {
      "epoch": 1.3424878836833603,
      "grad_norm": 0.08106525242328644,
      "learning_rate": 1.90304396843292e-05,
      "loss": 0.0751,
      "step": 416
    },
    {
      "epoch": 1.345718901453958,
      "grad_norm": 0.06518511474132538,
      "learning_rate": 1.9025930101465615e-05,
      "loss": 0.0594,
      "step": 417
    },
    {
      "epoch": 1.3489499192245558,
      "grad_norm": 0.08395849913358688,
      "learning_rate": 1.9021420518602033e-05,
      "loss": 0.0756,
      "step": 418
    },
    {
      "epoch": 1.3521809369951534,
      "grad_norm": 0.05958770960569382,
      "learning_rate": 1.9016910935738447e-05,
      "loss": 0.0491,
      "step": 419
    },
    {
      "epoch": 1.3554119547657513,
      "grad_norm": 0.07311136275529861,
      "learning_rate": 1.901240135287486e-05,
      "loss": 0.0719,
      "step": 420
    },
    {
      "epoch": 1.358642972536349,
      "grad_norm": 0.09682740271091461,
      "learning_rate": 1.9007891770011276e-05,
      "loss": 0.0895,
      "step": 421
    },
    {
      "epoch": 1.3618739903069468,
      "grad_norm": 0.06295045465230942,
      "learning_rate": 1.900338218714769e-05,
      "loss": 0.0589,
      "step": 422
    },
    {
      "epoch": 1.3651050080775444,
      "grad_norm": 0.0831819698214531,
      "learning_rate": 1.8998872604284105e-05,
      "loss": 0.0727,
      "step": 423
    },
    {
      "epoch": 1.368336025848142,
      "grad_norm": 0.06702585518360138,
      "learning_rate": 1.8994363021420522e-05,
      "loss": 0.0617,
      "step": 424
    },
    {
      "epoch": 1.37156704361874,
      "grad_norm": 0.06618952006101608,
      "learning_rate": 1.8989853438556937e-05,
      "loss": 0.0517,
      "step": 425
    },
    {
      "epoch": 1.3747980613893376,
      "grad_norm": 0.07830128818750381,
      "learning_rate": 1.898534385569335e-05,
      "loss": 0.0747,
      "step": 426
    },
    {
      "epoch": 1.3780290791599354,
      "grad_norm": 0.07554402947425842,
      "learning_rate": 1.8980834272829766e-05,
      "loss": 0.078,
      "step": 427
    },
    {
      "epoch": 1.381260096930533,
      "grad_norm": 0.07517927139997482,
      "learning_rate": 1.897632468996618e-05,
      "loss": 0.0715,
      "step": 428
    },
    {
      "epoch": 1.384491114701131,
      "grad_norm": 0.05810945853590965,
      "learning_rate": 1.8971815107102594e-05,
      "loss": 0.0524,
      "step": 429
    },
    {
      "epoch": 1.3877221324717286,
      "grad_norm": 0.1092490404844284,
      "learning_rate": 1.896730552423901e-05,
      "loss": 0.0987,
      "step": 430
    },
    {
      "epoch": 1.3909531502423262,
      "grad_norm": 0.08325308561325073,
      "learning_rate": 1.8962795941375423e-05,
      "loss": 0.0746,
      "step": 431
    },
    {
      "epoch": 1.394184168012924,
      "grad_norm": 0.08017408847808838,
      "learning_rate": 1.895828635851184e-05,
      "loss": 0.0676,
      "step": 432
    },
    {
      "epoch": 1.3974151857835218,
      "grad_norm": 0.09756331145763397,
      "learning_rate": 1.8953776775648255e-05,
      "loss": 0.0788,
      "step": 433
    },
    {
      "epoch": 1.4006462035541196,
      "grad_norm": 0.0654483512043953,
      "learning_rate": 1.894926719278467e-05,
      "loss": 0.0552,
      "step": 434
    },
    {
      "epoch": 1.4038772213247173,
      "grad_norm": 0.07338982075452805,
      "learning_rate": 1.8944757609921084e-05,
      "loss": 0.0597,
      "step": 435
    },
    {
      "epoch": 1.4071082390953151,
      "grad_norm": 0.06292750686407089,
      "learning_rate": 1.89402480270575e-05,
      "loss": 0.0482,
      "step": 436
    },
    {
      "epoch": 1.4103392568659128,
      "grad_norm": 0.09405938535928726,
      "learning_rate": 1.8935738444193913e-05,
      "loss": 0.0795,
      "step": 437
    },
    {
      "epoch": 1.4135702746365104,
      "grad_norm": 0.09486392885446548,
      "learning_rate": 1.8931228861330327e-05,
      "loss": 0.0811,
      "step": 438
    },
    {
      "epoch": 1.4168012924071083,
      "grad_norm": 0.0729052945971489,
      "learning_rate": 1.892671927846674e-05,
      "loss": 0.0682,
      "step": 439
    },
    {
      "epoch": 1.420032310177706,
      "grad_norm": 0.06790515035390854,
      "learning_rate": 1.8922209695603156e-05,
      "loss": 0.0541,
      "step": 440
    },
    {
      "epoch": 1.4232633279483038,
      "grad_norm": 0.08173596858978271,
      "learning_rate": 1.891770011273957e-05,
      "loss": 0.0697,
      "step": 441
    },
    {
      "epoch": 1.4264943457189014,
      "grad_norm": 0.0874050036072731,
      "learning_rate": 1.8913190529875988e-05,
      "loss": 0.0689,
      "step": 442
    },
    {
      "epoch": 1.4297253634894993,
      "grad_norm": 0.07508452981710434,
      "learning_rate": 1.8908680947012403e-05,
      "loss": 0.069,
      "step": 443
    },
    {
      "epoch": 1.432956381260097,
      "grad_norm": 0.09134234488010406,
      "learning_rate": 1.8904171364148817e-05,
      "loss": 0.072,
      "step": 444
    },
    {
      "epoch": 1.4361873990306946,
      "grad_norm": 0.0830577090382576,
      "learning_rate": 1.889966178128523e-05,
      "loss": 0.0681,
      "step": 445
    },
    {
      "epoch": 1.4394184168012925,
      "grad_norm": 0.0741642490029335,
      "learning_rate": 1.8895152198421646e-05,
      "loss": 0.0642,
      "step": 446
    },
    {
      "epoch": 1.44264943457189,
      "grad_norm": 0.07305614650249481,
      "learning_rate": 1.889064261555806e-05,
      "loss": 0.0569,
      "step": 447
    },
    {
      "epoch": 1.445880452342488,
      "grad_norm": 0.05348379164934158,
      "learning_rate": 1.8886133032694478e-05,
      "loss": 0.0434,
      "step": 448
    },
    {
      "epoch": 1.4491114701130856,
      "grad_norm": 0.09780937433242798,
      "learning_rate": 1.8881623449830892e-05,
      "loss": 0.0856,
      "step": 449
    },
    {
      "epoch": 1.4523424878836835,
      "grad_norm": 0.081721231341362,
      "learning_rate": 1.8877113866967307e-05,
      "loss": 0.0655,
      "step": 450
    },
    {
      "epoch": 1.4523424878836835,
      "eval_loss": 0.0769171267747879,
      "eval_runtime": 188.2463,
      "eval_samples_per_second": 1.047,
      "eval_steps_per_second": 1.047,
      "step": 450
    },
    {
      "epoch": 1.4555735056542811,
      "grad_norm": 0.06182597577571869,
      "learning_rate": 1.887260428410372e-05,
      "loss": 0.0549,
      "step": 451
    },
    {
      "epoch": 1.4588045234248788,
      "grad_norm": 0.0831274464726448,
      "learning_rate": 1.8868094701240136e-05,
      "loss": 0.0664,
      "step": 452
    },
    {
      "epoch": 1.4620355411954766,
      "grad_norm": 0.07277555763721466,
      "learning_rate": 1.8863585118376553e-05,
      "loss": 0.0619,
      "step": 453
    },
    {
      "epoch": 1.4652665589660743,
      "grad_norm": 0.09069440513849258,
      "learning_rate": 1.8859075535512968e-05,
      "loss": 0.0717,
      "step": 454
    },
    {
      "epoch": 1.468497576736672,
      "grad_norm": 0.08567981421947479,
      "learning_rate": 1.8854565952649382e-05,
      "loss": 0.0761,
      "step": 455
    },
    {
      "epoch": 1.4717285945072698,
      "grad_norm": 0.08353572338819504,
      "learning_rate": 1.8850056369785796e-05,
      "loss": 0.0621,
      "step": 456
    },
    {
      "epoch": 1.4749596122778676,
      "grad_norm": 0.06493799388408661,
      "learning_rate": 1.884554678692221e-05,
      "loss": 0.0549,
      "step": 457
    },
    {
      "epoch": 1.4781906300484653,
      "grad_norm": 0.07239842414855957,
      "learning_rate": 1.8841037204058625e-05,
      "loss": 0.0574,
      "step": 458
    },
    {
      "epoch": 1.481421647819063,
      "grad_norm": 0.1062210276722908,
      "learning_rate": 1.8836527621195043e-05,
      "loss": 0.0831,
      "step": 459
    },
    {
      "epoch": 1.4846526655896608,
      "grad_norm": 0.06695660948753357,
      "learning_rate": 1.8832018038331457e-05,
      "loss": 0.0488,
      "step": 460
    },
    {
      "epoch": 1.4878836833602584,
      "grad_norm": 0.08332875370979309,
      "learning_rate": 1.8827508455467872e-05,
      "loss": 0.0752,
      "step": 461
    },
    {
      "epoch": 1.491114701130856,
      "grad_norm": 0.09285688400268555,
      "learning_rate": 1.8822998872604286e-05,
      "loss": 0.0811,
      "step": 462
    },
    {
      "epoch": 1.494345718901454,
      "grad_norm": 0.07672538608312607,
      "learning_rate": 1.88184892897407e-05,
      "loss": 0.0565,
      "step": 463
    },
    {
      "epoch": 1.4975767366720518,
      "grad_norm": 0.07295355945825577,
      "learning_rate": 1.8813979706877115e-05,
      "loss": 0.0615,
      "step": 464
    },
    {
      "epoch": 1.5008077544426495,
      "grad_norm": 0.05997586250305176,
      "learning_rate": 1.8809470124013533e-05,
      "loss": 0.0494,
      "step": 465
    },
    {
      "epoch": 1.504038772213247,
      "grad_norm": 0.08460883051156998,
      "learning_rate": 1.8804960541149947e-05,
      "loss": 0.0715,
      "step": 466
    },
    {
      "epoch": 1.507269789983845,
      "grad_norm": 0.08083106577396393,
      "learning_rate": 1.880045095828636e-05,
      "loss": 0.0615,
      "step": 467
    },
    {
      "epoch": 1.5105008077544426,
      "grad_norm": 0.09291260689496994,
      "learning_rate": 1.8795941375422776e-05,
      "loss": 0.0767,
      "step": 468
    },
    {
      "epoch": 1.5137318255250403,
      "grad_norm": 0.0817233994603157,
      "learning_rate": 1.879143179255919e-05,
      "loss": 0.0728,
      "step": 469
    },
    {
      "epoch": 1.5169628432956381,
      "grad_norm": 0.07894831895828247,
      "learning_rate": 1.8786922209695605e-05,
      "loss": 0.0697,
      "step": 470
    },
    {
      "epoch": 1.520193861066236,
      "grad_norm": 0.05180181935429573,
      "learning_rate": 1.878241262683202e-05,
      "loss": 0.0407,
      "step": 471
    },
    {
      "epoch": 1.5234248788368336,
      "grad_norm": 0.08214667439460754,
      "learning_rate": 1.8777903043968433e-05,
      "loss": 0.0673,
      "step": 472
    },
    {
      "epoch": 1.5266558966074313,
      "grad_norm": 0.06972946226596832,
      "learning_rate": 1.877339346110485e-05,
      "loss": 0.0501,
      "step": 473
    },
    {
      "epoch": 1.5298869143780292,
      "grad_norm": 0.08416459709405899,
      "learning_rate": 1.8768883878241266e-05,
      "loss": 0.0666,
      "step": 474
    },
    {
      "epoch": 1.5331179321486268,
      "grad_norm": 0.07642164081335068,
      "learning_rate": 1.876437429537768e-05,
      "loss": 0.0592,
      "step": 475
    },
    {
      "epoch": 1.5363489499192244,
      "grad_norm": 0.0762806385755539,
      "learning_rate": 1.8759864712514094e-05,
      "loss": 0.0573,
      "step": 476
    },
    {
      "epoch": 1.5395799676898223,
      "grad_norm": 0.06152572110295296,
      "learning_rate": 1.875535512965051e-05,
      "loss": 0.0509,
      "step": 477
    },
    {
      "epoch": 1.5428109854604202,
      "grad_norm": 0.08461987972259521,
      "learning_rate": 1.8750845546786923e-05,
      "loss": 0.0693,
      "step": 478
    },
    {
      "epoch": 1.5460420032310178,
      "grad_norm": 0.06401054561138153,
      "learning_rate": 1.8746335963923338e-05,
      "loss": 0.0523,
      "step": 479
    },
    {
      "epoch": 1.5492730210016155,
      "grad_norm": 0.07567861676216125,
      "learning_rate": 1.8741826381059752e-05,
      "loss": 0.0632,
      "step": 480
    },
    {
      "epoch": 1.5525040387722133,
      "grad_norm": 0.07169700413942337,
      "learning_rate": 1.8737316798196166e-05,
      "loss": 0.0541,
      "step": 481
    },
    {
      "epoch": 1.555735056542811,
      "grad_norm": 0.067410409450531,
      "learning_rate": 1.873280721533258e-05,
      "loss": 0.0576,
      "step": 482
    },
    {
      "epoch": 1.5589660743134086,
      "grad_norm": 0.0794718787074089,
      "learning_rate": 1.8728297632469e-05,
      "loss": 0.0602,
      "step": 483
    },
    {
      "epoch": 1.5621970920840065,
      "grad_norm": 0.09098870307207108,
      "learning_rate": 1.8723788049605413e-05,
      "loss": 0.0741,
      "step": 484
    },
    {
      "epoch": 1.5654281098546043,
      "grad_norm": 0.07266968488693237,
      "learning_rate": 1.8719278466741827e-05,
      "loss": 0.0535,
      "step": 485
    },
    {
      "epoch": 1.568659127625202,
      "grad_norm": 0.07994985580444336,
      "learning_rate": 1.8714768883878242e-05,
      "loss": 0.0642,
      "step": 486
    },
    {
      "epoch": 1.5718901453957996,
      "grad_norm": 0.09563203901052475,
      "learning_rate": 1.8710259301014656e-05,
      "loss": 0.0738,
      "step": 487
    },
    {
      "epoch": 1.5751211631663975,
      "grad_norm": 0.07337169349193573,
      "learning_rate": 1.870574971815107e-05,
      "loss": 0.0615,
      "step": 488
    },
    {
      "epoch": 1.5783521809369951,
      "grad_norm": 0.08605758100748062,
      "learning_rate": 1.8701240135287488e-05,
      "loss": 0.0737,
      "step": 489
    },
    {
      "epoch": 1.5815831987075928,
      "grad_norm": 0.08178628236055374,
      "learning_rate": 1.8696730552423903e-05,
      "loss": 0.0562,
      "step": 490
    },
    {
      "epoch": 1.5848142164781907,
      "grad_norm": 0.08635883033275604,
      "learning_rate": 1.8692220969560317e-05,
      "loss": 0.0655,
      "step": 491
    },
    {
      "epoch": 1.5880452342487885,
      "grad_norm": 0.10575321316719055,
      "learning_rate": 1.868771138669673e-05,
      "loss": 0.0857,
      "step": 492
    },
    {
      "epoch": 1.5912762520193862,
      "grad_norm": 0.10067257285118103,
      "learning_rate": 1.8683201803833146e-05,
      "loss": 0.0817,
      "step": 493
    },
    {
      "epoch": 1.5945072697899838,
      "grad_norm": 0.07644681632518768,
      "learning_rate": 1.867869222096956e-05,
      "loss": 0.0625,
      "step": 494
    },
    {
      "epoch": 1.5977382875605817,
      "grad_norm": 0.07164619863033295,
      "learning_rate": 1.8674182638105978e-05,
      "loss": 0.0597,
      "step": 495
    },
    {
      "epoch": 1.6009693053311793,
      "grad_norm": 0.07293085008859634,
      "learning_rate": 1.8669673055242392e-05,
      "loss": 0.0604,
      "step": 496
    },
    {
      "epoch": 1.604200323101777,
      "grad_norm": 0.09480689465999603,
      "learning_rate": 1.8665163472378807e-05,
      "loss": 0.0737,
      "step": 497
    },
    {
      "epoch": 1.6074313408723748,
      "grad_norm": 0.09798948466777802,
      "learning_rate": 1.866065388951522e-05,
      "loss": 0.0706,
      "step": 498
    },
    {
      "epoch": 1.6106623586429727,
      "grad_norm": 0.08216292411088943,
      "learning_rate": 1.8656144306651636e-05,
      "loss": 0.0585,
      "step": 499
    },
    {
      "epoch": 1.6138933764135701,
      "grad_norm": 0.10146701335906982,
      "learning_rate": 1.865163472378805e-05,
      "loss": 0.0631,
      "step": 500
    },
    {
      "epoch": 1.617124394184168,
      "grad_norm": 0.07699297368526459,
      "learning_rate": 1.8647125140924468e-05,
      "loss": 0.0592,
      "step": 501
    },
    {
      "epoch": 1.6203554119547658,
      "grad_norm": 0.07803017646074295,
      "learning_rate": 1.8642615558060882e-05,
      "loss": 0.063,
      "step": 502
    },
    {
      "epoch": 1.6235864297253635,
      "grad_norm": 0.08820293843746185,
      "learning_rate": 1.8638105975197297e-05,
      "loss": 0.0733,
      "step": 503
    },
    {
      "epoch": 1.6268174474959611,
      "grad_norm": 0.10102511942386627,
      "learning_rate": 1.863359639233371e-05,
      "loss": 0.0735,
      "step": 504
    },
    {
      "epoch": 1.630048465266559,
      "grad_norm": 0.08669153600931168,
      "learning_rate": 1.8629086809470125e-05,
      "loss": 0.0757,
      "step": 505
    },
    {
      "epoch": 1.6332794830371569,
      "grad_norm": 0.08120600879192352,
      "learning_rate": 1.862457722660654e-05,
      "loss": 0.0586,
      "step": 506
    },
    {
      "epoch": 1.6365105008077543,
      "grad_norm": 0.06960420310497284,
      "learning_rate": 1.8620067643742957e-05,
      "loss": 0.0519,
      "step": 507
    },
    {
      "epoch": 1.6397415185783522,
      "grad_norm": 0.08567452430725098,
      "learning_rate": 1.8615558060879372e-05,
      "loss": 0.0703,
      "step": 508
    },
    {
      "epoch": 1.64297253634895,
      "grad_norm": 0.08288481831550598,
      "learning_rate": 1.8611048478015786e-05,
      "loss": 0.0624,
      "step": 509
    },
    {
      "epoch": 1.6462035541195477,
      "grad_norm": 0.10185632109642029,
      "learning_rate": 1.86065388951522e-05,
      "loss": 0.072,
      "step": 510
    },
    {
      "epoch": 1.6494345718901453,
      "grad_norm": 0.0914456769824028,
      "learning_rate": 1.8602029312288615e-05,
      "loss": 0.0683,
      "step": 511
    },
    {
      "epoch": 1.6526655896607432,
      "grad_norm": 0.08204774558544159,
      "learning_rate": 1.859751972942503e-05,
      "loss": 0.0657,
      "step": 512
    },
    {
      "epoch": 1.655896607431341,
      "grad_norm": 0.11823786050081253,
      "learning_rate": 1.8593010146561444e-05,
      "loss": 0.0872,
      "step": 513
    },
    {
      "epoch": 1.6591276252019385,
      "grad_norm": 0.13115671277046204,
      "learning_rate": 1.858850056369786e-05,
      "loss": 0.1004,
      "step": 514
    },
    {
      "epoch": 1.6623586429725363,
      "grad_norm": 0.09443841129541397,
      "learning_rate": 1.8583990980834276e-05,
      "loss": 0.0722,
      "step": 515
    },
    {
      "epoch": 1.6655896607431342,
      "grad_norm": 0.06847009062767029,
      "learning_rate": 1.857948139797069e-05,
      "loss": 0.054,
      "step": 516
    },
    {
      "epoch": 1.6688206785137318,
      "grad_norm": 0.07960178703069687,
      "learning_rate": 1.8574971815107105e-05,
      "loss": 0.065,
      "step": 517
    },
    {
      "epoch": 1.6720516962843295,
      "grad_norm": 0.07255195826292038,
      "learning_rate": 1.857046223224352e-05,
      "loss": 0.0523,
      "step": 518
    },
    {
      "epoch": 1.6752827140549273,
      "grad_norm": 0.08610787242650986,
      "learning_rate": 1.8565952649379934e-05,
      "loss": 0.0667,
      "step": 519
    },
    {
      "epoch": 1.678513731825525,
      "grad_norm": 0.09422770887613297,
      "learning_rate": 1.8561443066516348e-05,
      "loss": 0.0772,
      "step": 520
    },
    {
      "epoch": 1.6817447495961226,
      "grad_norm": 0.10716807097196579,
      "learning_rate": 1.8556933483652762e-05,
      "loss": 0.0825,
      "step": 521
    },
    {
      "epoch": 1.6849757673667205,
      "grad_norm": 0.09894333779811859,
      "learning_rate": 1.8552423900789177e-05,
      "loss": 0.084,
      "step": 522
    },
    {
      "epoch": 1.6882067851373184,
      "grad_norm": 0.08127731829881668,
      "learning_rate": 1.854791431792559e-05,
      "loss": 0.063,
      "step": 523
    },
    {
      "epoch": 1.691437802907916,
      "grad_norm": 0.08127739280462265,
      "learning_rate": 1.854340473506201e-05,
      "loss": 0.0602,
      "step": 524
    },
    {
      "epoch": 1.6946688206785137,
      "grad_norm": 0.08109954744577408,
      "learning_rate": 1.8538895152198423e-05,
      "loss": 0.0618,
      "step": 525
    },
    {
      "epoch": 1.6946688206785137,
      "eval_loss": 0.07574764639139175,
      "eval_runtime": 188.1431,
      "eval_samples_per_second": 1.047,
      "eval_steps_per_second": 1.047,
      "step": 525
    },
    {
      "epoch": 1.6978998384491115,
      "grad_norm": 0.07915576547384262,
      "learning_rate": 1.8534385569334838e-05,
      "loss": 0.0566,
      "step": 526
    },
    {
      "epoch": 1.7011308562197092,
      "grad_norm": 0.09259936213493347,
      "learning_rate": 1.8529875986471252e-05,
      "loss": 0.0821,
      "step": 527
    },
    {
      "epoch": 1.7043618739903068,
      "grad_norm": 0.06958405673503876,
      "learning_rate": 1.8525366403607666e-05,
      "loss": 0.0513,
      "step": 528
    },
    {
      "epoch": 1.7075928917609047,
      "grad_norm": 0.11260278522968292,
      "learning_rate": 1.852085682074408e-05,
      "loss": 0.0791,
      "step": 529
    },
    {
      "epoch": 1.7108239095315025,
      "grad_norm": 0.08565714955329895,
      "learning_rate": 1.8516347237880495e-05,
      "loss": 0.0704,
      "step": 530
    },
    {
      "epoch": 1.7140549273021002,
      "grad_norm": 0.0977453961968422,
      "learning_rate": 1.8511837655016913e-05,
      "loss": 0.0666,
      "step": 531
    },
    {
      "epoch": 1.7172859450726978,
      "grad_norm": 0.09589142352342606,
      "learning_rate": 1.8507328072153327e-05,
      "loss": 0.0678,
      "step": 532
    },
    {
      "epoch": 1.7205169628432957,
      "grad_norm": 0.10372763872146606,
      "learning_rate": 1.8502818489289742e-05,
      "loss": 0.0755,
      "step": 533
    },
    {
      "epoch": 1.7237479806138933,
      "grad_norm": 0.09707041829824448,
      "learning_rate": 1.8498308906426156e-05,
      "loss": 0.0759,
      "step": 534
    },
    {
      "epoch": 1.726978998384491,
      "grad_norm": 0.07280156016349792,
      "learning_rate": 1.849379932356257e-05,
      "loss": 0.058,
      "step": 535
    },
    {
      "epoch": 1.7302100161550888,
      "grad_norm": 0.08810850977897644,
      "learning_rate": 1.848928974069899e-05,
      "loss": 0.0691,
      "step": 536
    },
    {
      "epoch": 1.7334410339256867,
      "grad_norm": 0.09844056516885757,
      "learning_rate": 1.8484780157835403e-05,
      "loss": 0.0682,
      "step": 537
    },
    {
      "epoch": 1.7366720516962844,
      "grad_norm": 0.06963982433080673,
      "learning_rate": 1.8480270574971817e-05,
      "loss": 0.0513,
      "step": 538
    },
    {
      "epoch": 1.739903069466882,
      "grad_norm": 0.08248520642518997,
      "learning_rate": 1.847576099210823e-05,
      "loss": 0.0611,
      "step": 539
    },
    {
      "epoch": 1.7431340872374799,
      "grad_norm": 0.09553173929452896,
      "learning_rate": 1.8471251409244646e-05,
      "loss": 0.0754,
      "step": 540
    },
    {
      "epoch": 1.7463651050080775,
      "grad_norm": 0.06919584423303604,
      "learning_rate": 1.846674182638106e-05,
      "loss": 0.0508,
      "step": 541
    },
    {
      "epoch": 1.7495961227786752,
      "grad_norm": 0.07004183530807495,
      "learning_rate": 1.8462232243517478e-05,
      "loss": 0.0498,
      "step": 542
    },
    {
      "epoch": 1.752827140549273,
      "grad_norm": 0.08570928126573563,
      "learning_rate": 1.8457722660653892e-05,
      "loss": 0.0631,
      "step": 543
    },
    {
      "epoch": 1.7560581583198709,
      "grad_norm": 0.0732467994093895,
      "learning_rate": 1.8453213077790307e-05,
      "loss": 0.0557,
      "step": 544
    },
    {
      "epoch": 1.7592891760904685,
      "grad_norm": 0.07687011361122131,
      "learning_rate": 1.844870349492672e-05,
      "loss": 0.0573,
      "step": 545
    },
    {
      "epoch": 1.7625201938610662,
      "grad_norm": 0.07186026871204376,
      "learning_rate": 1.8444193912063136e-05,
      "loss": 0.0502,
      "step": 546
    },
    {
      "epoch": 1.765751211631664,
      "grad_norm": 0.07176259905099869,
      "learning_rate": 1.843968432919955e-05,
      "loss": 0.0529,
      "step": 547
    },
    {
      "epoch": 1.7689822294022617,
      "grad_norm": 0.0842595249414444,
      "learning_rate": 1.8435174746335968e-05,
      "loss": 0.0674,
      "step": 548
    },
    {
      "epoch": 1.7722132471728593,
      "grad_norm": 0.07965710759162903,
      "learning_rate": 1.8430665163472382e-05,
      "loss": 0.0619,
      "step": 549
    },
    {
      "epoch": 1.7754442649434572,
      "grad_norm": 0.08953316509723663,
      "learning_rate": 1.8426155580608797e-05,
      "loss": 0.0643,
      "step": 550
    },
    {
      "epoch": 1.778675282714055,
      "grad_norm": 0.0928904190659523,
      "learning_rate": 1.842164599774521e-05,
      "loss": 0.0712,
      "step": 551
    },
    {
      "epoch": 1.7819063004846527,
      "grad_norm": 0.08743231743574142,
      "learning_rate": 1.8417136414881625e-05,
      "loss": 0.0657,
      "step": 552
    },
    {
      "epoch": 1.7851373182552503,
      "grad_norm": 0.07706678658723831,
      "learning_rate": 1.841262683201804e-05,
      "loss": 0.0568,
      "step": 553
    },
    {
      "epoch": 1.7883683360258482,
      "grad_norm": 0.0831725150346756,
      "learning_rate": 1.8408117249154454e-05,
      "loss": 0.0578,
      "step": 554
    },
    {
      "epoch": 1.7915993537964459,
      "grad_norm": 0.09395398199558258,
      "learning_rate": 1.8403607666290872e-05,
      "loss": 0.0715,
      "step": 555
    },
    {
      "epoch": 1.7948303715670435,
      "grad_norm": 0.0942830964922905,
      "learning_rate": 1.8399098083427286e-05,
      "loss": 0.0634,
      "step": 556
    },
    {
      "epoch": 1.7980613893376414,
      "grad_norm": 0.0980205312371254,
      "learning_rate": 1.83945885005637e-05,
      "loss": 0.0694,
      "step": 557
    },
    {
      "epoch": 1.8012924071082392,
      "grad_norm": 0.10699216276407242,
      "learning_rate": 1.8390078917700115e-05,
      "loss": 0.0871,
      "step": 558
    },
    {
      "epoch": 1.8045234248788369,
      "grad_norm": 0.09851755946874619,
      "learning_rate": 1.838556933483653e-05,
      "loss": 0.0644,
      "step": 559
    },
    {
      "epoch": 1.8077544426494345,
      "grad_norm": 0.09926044940948486,
      "learning_rate": 1.8381059751972944e-05,
      "loss": 0.0629,
      "step": 560
    },
    {
      "epoch": 1.8109854604200324,
      "grad_norm": 0.09520839154720306,
      "learning_rate": 1.8376550169109358e-05,
      "loss": 0.0607,
      "step": 561
    },
    {
      "epoch": 1.81421647819063,
      "grad_norm": 0.06896607577800751,
      "learning_rate": 1.8372040586245773e-05,
      "loss": 0.0559,
      "step": 562
    },
    {
      "epoch": 1.8174474959612277,
      "grad_norm": 0.09539300203323364,
      "learning_rate": 1.8367531003382187e-05,
      "loss": 0.0683,
      "step": 563
    },
    {
      "epoch": 1.8206785137318255,
      "grad_norm": 0.10596197098493576,
      "learning_rate": 1.83630214205186e-05,
      "loss": 0.075,
      "step": 564
    },
    {
      "epoch": 1.8239095315024234,
      "grad_norm": 0.13212427496910095,
      "learning_rate": 1.835851183765502e-05,
      "loss": 0.0755,
      "step": 565
    },
    {
      "epoch": 1.827140549273021,
      "grad_norm": 0.12438125163316727,
      "learning_rate": 1.8354002254791434e-05,
      "loss": 0.0853,
      "step": 566
    },
    {
      "epoch": 1.8303715670436187,
      "grad_norm": 0.06944366544485092,
      "learning_rate": 1.8349492671927848e-05,
      "loss": 0.0434,
      "step": 567
    },
    {
      "epoch": 1.8336025848142166,
      "grad_norm": 0.10360438376665115,
      "learning_rate": 1.8344983089064262e-05,
      "loss": 0.0778,
      "step": 568
    },
    {
      "epoch": 1.8368336025848142,
      "grad_norm": 0.1002860888838768,
      "learning_rate": 1.8340473506200677e-05,
      "loss": 0.0733,
      "step": 569
    },
    {
      "epoch": 1.8400646203554119,
      "grad_norm": 0.10875017940998077,
      "learning_rate": 1.833596392333709e-05,
      "loss": 0.0781,
      "step": 570
    },
    {
      "epoch": 1.8432956381260097,
      "grad_norm": 0.08669572323560715,
      "learning_rate": 1.8331454340473506e-05,
      "loss": 0.0587,
      "step": 571
    },
    {
      "epoch": 1.8465266558966076,
      "grad_norm": 0.09304548799991608,
      "learning_rate": 1.8326944757609923e-05,
      "loss": 0.0675,
      "step": 572
    },
    {
      "epoch": 1.849757673667205,
      "grad_norm": 0.09815046936273575,
      "learning_rate": 1.8322435174746338e-05,
      "loss": 0.0752,
      "step": 573
    },
    {
      "epoch": 1.8529886914378029,
      "grad_norm": 0.08040884137153625,
      "learning_rate": 1.8317925591882752e-05,
      "loss": 0.06,
      "step": 574
    },
    {
      "epoch": 1.8562197092084007,
      "grad_norm": 0.08228793740272522,
      "learning_rate": 1.8313416009019166e-05,
      "loss": 0.0547,
      "step": 575
    },
    {
      "epoch": 1.8594507269789984,
      "grad_norm": 0.07945281267166138,
      "learning_rate": 1.830890642615558e-05,
      "loss": 0.0467,
      "step": 576
    },
    {
      "epoch": 1.862681744749596,
      "grad_norm": 0.07659505307674408,
      "learning_rate": 1.8304396843291995e-05,
      "loss": 0.0571,
      "step": 577
    },
    {
      "epoch": 1.865912762520194,
      "grad_norm": 0.07296533137559891,
      "learning_rate": 1.8299887260428413e-05,
      "loss": 0.0567,
      "step": 578
    },
    {
      "epoch": 1.8691437802907918,
      "grad_norm": 0.10132135450839996,
      "learning_rate": 1.8295377677564827e-05,
      "loss": 0.0589,
      "step": 579
    },
    {
      "epoch": 1.8723747980613892,
      "grad_norm": 0.0985584482550621,
      "learning_rate": 1.8290868094701242e-05,
      "loss": 0.0707,
      "step": 580
    },
    {
      "epoch": 1.875605815831987,
      "grad_norm": 0.09575635194778442,
      "learning_rate": 1.8286358511837656e-05,
      "loss": 0.0687,
      "step": 581
    },
    {
      "epoch": 1.878836833602585,
      "grad_norm": 0.10488908737897873,
      "learning_rate": 1.828184892897407e-05,
      "loss": 0.0752,
      "step": 582
    },
    {
      "epoch": 1.8820678513731826,
      "grad_norm": 0.0739881619811058,
      "learning_rate": 1.8277339346110485e-05,
      "loss": 0.0537,
      "step": 583
    },
    {
      "epoch": 1.8852988691437802,
      "grad_norm": 0.06086435914039612,
      "learning_rate": 1.8272829763246903e-05,
      "loss": 0.051,
      "step": 584
    },
    {
      "epoch": 1.888529886914378,
      "grad_norm": 0.09947849065065384,
      "learning_rate": 1.8268320180383317e-05,
      "loss": 0.0705,
      "step": 585
    },
    {
      "epoch": 1.891760904684976,
      "grad_norm": 0.09509633481502533,
      "learning_rate": 1.826381059751973e-05,
      "loss": 0.0665,
      "step": 586
    },
    {
      "epoch": 1.8949919224555734,
      "grad_norm": 0.12639068067073822,
      "learning_rate": 1.8259301014656146e-05,
      "loss": 0.0538,
      "step": 587
    },
    {
      "epoch": 1.8982229402261712,
      "grad_norm": 0.09957147389650345,
      "learning_rate": 1.825479143179256e-05,
      "loss": 0.0649,
      "step": 588
    },
    {
      "epoch": 1.901453957996769,
      "grad_norm": 0.10096530616283417,
      "learning_rate": 1.8250281848928975e-05,
      "loss": 0.0607,
      "step": 589
    },
    {
      "epoch": 1.9046849757673667,
      "grad_norm": 0.10449621081352234,
      "learning_rate": 1.8245772266065392e-05,
      "loss": 0.0742,
      "step": 590
    },
    {
      "epoch": 1.9079159935379644,
      "grad_norm": 0.09589482843875885,
      "learning_rate": 1.8241262683201807e-05,
      "loss": 0.069,
      "step": 591
    },
    {
      "epoch": 1.9111470113085622,
      "grad_norm": 0.10444579273462296,
      "learning_rate": 1.823675310033822e-05,
      "loss": 0.0726,
      "step": 592
    },
    {
      "epoch": 1.9143780290791599,
      "grad_norm": 0.09007811546325684,
      "learning_rate": 1.8232243517474636e-05,
      "loss": 0.0668,
      "step": 593
    },
    {
      "epoch": 1.9176090468497575,
      "grad_norm": 0.093568354845047,
      "learning_rate": 1.822773393461105e-05,
      "loss": 0.0658,
      "step": 594
    },
    {
      "epoch": 1.9208400646203554,
      "grad_norm": 0.0825546383857727,
      "learning_rate": 1.8223224351747464e-05,
      "loss": 0.0586,
      "step": 595
    },
    {
      "epoch": 1.9240710823909533,
      "grad_norm": 0.08507188409566879,
      "learning_rate": 1.8218714768883882e-05,
      "loss": 0.0617,
      "step": 596
    },
    {
      "epoch": 1.927302100161551,
      "grad_norm": 0.0858079269528389,
      "learning_rate": 1.8214205186020297e-05,
      "loss": 0.0601,
      "step": 597
    },
    {
      "epoch": 1.9305331179321485,
      "grad_norm": 0.1205410435795784,
      "learning_rate": 1.820969560315671e-05,
      "loss": 0.0883,
      "step": 598
    },
    {
      "epoch": 1.9337641357027464,
      "grad_norm": 0.1289929449558258,
      "learning_rate": 1.8205186020293125e-05,
      "loss": 0.076,
      "step": 599
    },
    {
      "epoch": 1.936995153473344,
      "grad_norm": 0.11139614135026932,
      "learning_rate": 1.820067643742954e-05,
      "loss": 0.0846,
      "step": 600
    },
    {
      "epoch": 1.936995153473344,
      "eval_loss": 0.07442964613437653,
      "eval_runtime": 188.1343,
      "eval_samples_per_second": 1.047,
      "eval_steps_per_second": 1.047,
      "step": 600
    },
    {
      "epoch": 1.9402261712439417,
      "grad_norm": 0.0923788920044899,
      "learning_rate": 1.8196166854565954e-05,
      "loss": 0.0605,
      "step": 601
    },
    {
      "epoch": 1.9434571890145396,
      "grad_norm": 0.1251497119665146,
      "learning_rate": 1.819165727170237e-05,
      "loss": 0.0894,
      "step": 602
    },
    {
      "epoch": 1.9466882067851374,
      "grad_norm": 0.11357556283473969,
      "learning_rate": 1.8187147688838783e-05,
      "loss": 0.0819,
      "step": 603
    },
    {
      "epoch": 1.949919224555735,
      "grad_norm": 0.09567239135503769,
      "learning_rate": 1.8182638105975197e-05,
      "loss": 0.0624,
      "step": 604
    },
    {
      "epoch": 1.9531502423263327,
      "grad_norm": 0.09191922098398209,
      "learning_rate": 1.8178128523111612e-05,
      "loss": 0.0578,
      "step": 605
    },
    {
      "epoch": 1.9563812600969306,
      "grad_norm": 0.09503104537725449,
      "learning_rate": 1.817361894024803e-05,
      "loss": 0.0641,
      "step": 606
    },
    {
      "epoch": 1.9596122778675282,
      "grad_norm": 0.11036618053913116,
      "learning_rate": 1.8169109357384444e-05,
      "loss": 0.0666,
      "step": 607
    },
    {
      "epoch": 1.9628432956381259,
      "grad_norm": 0.09104762971401215,
      "learning_rate": 1.8164599774520858e-05,
      "loss": 0.0649,
      "step": 608
    },
    {
      "epoch": 1.9660743134087237,
      "grad_norm": 0.10882871598005295,
      "learning_rate": 1.8160090191657273e-05,
      "loss": 0.0824,
      "step": 609
    },
    {
      "epoch": 1.9693053311793216,
      "grad_norm": 0.09632111340761185,
      "learning_rate": 1.8155580608793687e-05,
      "loss": 0.069,
      "step": 610
    },
    {
      "epoch": 1.9725363489499192,
      "grad_norm": 0.07966237515211105,
      "learning_rate": 1.81510710259301e-05,
      "loss": 0.0557,
      "step": 611
    },
    {
      "epoch": 1.975767366720517,
      "grad_norm": 0.10335849225521088,
      "learning_rate": 1.8146561443066516e-05,
      "loss": 0.0787,
      "step": 612
    },
    {
      "epoch": 1.9789983844911148,
      "grad_norm": 0.12096443772315979,
      "learning_rate": 1.814205186020293e-05,
      "loss": 0.0731,
      "step": 613
    },
    {
      "epoch": 1.9822294022617124,
      "grad_norm": 0.09432677179574966,
      "learning_rate": 1.8137542277339348e-05,
      "loss": 0.0635,
      "step": 614
    },
    {
      "epoch": 1.98546042003231,
      "grad_norm": 0.11708611994981766,
      "learning_rate": 1.8133032694475762e-05,
      "loss": 0.0874,
      "step": 615
    },
    {
      "epoch": 1.988691437802908,
      "grad_norm": 0.1113506332039833,
      "learning_rate": 1.8128523111612177e-05,
      "loss": 0.0678,
      "step": 616
    },
    {
      "epoch": 1.9919224555735058,
      "grad_norm": 0.09246299415826797,
      "learning_rate": 1.812401352874859e-05,
      "loss": 0.0672,
      "step": 617
    },
    {
      "epoch": 1.9951534733441034,
      "grad_norm": 0.1115182563662529,
      "learning_rate": 1.8119503945885006e-05,
      "loss": 0.0758,
      "step": 618
    },
    {
      "epoch": 1.998384491114701,
      "grad_norm": 0.10478319972753525,
      "learning_rate": 1.811499436302142e-05,
      "loss": 0.0648,
      "step": 619
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.13386856019496918,
      "learning_rate": 1.8110484780157838e-05,
      "loss": 0.0624,
      "step": 620
    },
    {
      "epoch": 2.003231017770598,
      "grad_norm": 0.09262198954820633,
      "learning_rate": 1.8105975197294252e-05,
      "loss": 0.0608,
      "step": 621
    },
    {
      "epoch": 2.0064620355411953,
      "grad_norm": 0.07091473788022995,
      "learning_rate": 1.8101465614430667e-05,
      "loss": 0.0434,
      "step": 622
    },
    {
      "epoch": 2.009693053311793,
      "grad_norm": 0.10324624925851822,
      "learning_rate": 1.809695603156708e-05,
      "loss": 0.0764,
      "step": 623
    },
    {
      "epoch": 2.012924071082391,
      "grad_norm": 0.10515467822551727,
      "learning_rate": 1.8092446448703495e-05,
      "loss": 0.0609,
      "step": 624
    },
    {
      "epoch": 2.016155088852989,
      "grad_norm": 0.10561127960681915,
      "learning_rate": 1.8087936865839913e-05,
      "loss": 0.0719,
      "step": 625
    },
    {
      "epoch": 2.0193861066235863,
      "grad_norm": 0.1146024838089943,
      "learning_rate": 1.8083427282976327e-05,
      "loss": 0.0726,
      "step": 626
    },
    {
      "epoch": 2.022617124394184,
      "grad_norm": 0.07930684089660645,
      "learning_rate": 1.8078917700112742e-05,
      "loss": 0.0481,
      "step": 627
    },
    {
      "epoch": 2.025848142164782,
      "grad_norm": 0.09927454590797424,
      "learning_rate": 1.8074408117249156e-05,
      "loss": 0.0608,
      "step": 628
    },
    {
      "epoch": 2.0290791599353795,
      "grad_norm": 0.08592136949300766,
      "learning_rate": 1.806989853438557e-05,
      "loss": 0.0577,
      "step": 629
    },
    {
      "epoch": 2.0323101777059773,
      "grad_norm": 0.09232696890830994,
      "learning_rate": 1.8065388951521985e-05,
      "loss": 0.0528,
      "step": 630
    },
    {
      "epoch": 2.035541195476575,
      "grad_norm": 0.08930118381977081,
      "learning_rate": 1.8060879368658403e-05,
      "loss": 0.0633,
      "step": 631
    },
    {
      "epoch": 2.038772213247173,
      "grad_norm": 0.09835111349821091,
      "learning_rate": 1.8056369785794817e-05,
      "loss": 0.0648,
      "step": 632
    },
    {
      "epoch": 2.0420032310177705,
      "grad_norm": 0.10789217799901962,
      "learning_rate": 1.805186020293123e-05,
      "loss": 0.0695,
      "step": 633
    },
    {
      "epoch": 2.0452342487883683,
      "grad_norm": 0.10506349802017212,
      "learning_rate": 1.8047350620067646e-05,
      "loss": 0.0635,
      "step": 634
    },
    {
      "epoch": 2.048465266558966,
      "grad_norm": 0.13068322837352753,
      "learning_rate": 1.804284103720406e-05,
      "loss": 0.0802,
      "step": 635
    },
    {
      "epoch": 2.0516962843295636,
      "grad_norm": 0.09663469344377518,
      "learning_rate": 1.8038331454340475e-05,
      "loss": 0.062,
      "step": 636
    },
    {
      "epoch": 2.0549273021001615,
      "grad_norm": 0.07054325938224792,
      "learning_rate": 1.8033821871476893e-05,
      "loss": 0.0453,
      "step": 637
    },
    {
      "epoch": 2.0581583198707594,
      "grad_norm": 0.07739470899105072,
      "learning_rate": 1.8029312288613307e-05,
      "loss": 0.0463,
      "step": 638
    },
    {
      "epoch": 2.0613893376413572,
      "grad_norm": 0.09022580832242966,
      "learning_rate": 1.802480270574972e-05,
      "loss": 0.0587,
      "step": 639
    },
    {
      "epoch": 2.0646203554119547,
      "grad_norm": 0.09953221678733826,
      "learning_rate": 1.8020293122886136e-05,
      "loss": 0.0585,
      "step": 640
    },
    {
      "epoch": 2.0678513731825525,
      "grad_norm": 0.10069511830806732,
      "learning_rate": 1.801578354002255e-05,
      "loss": 0.0587,
      "step": 641
    },
    {
      "epoch": 2.0710823909531504,
      "grad_norm": 0.09177737683057785,
      "learning_rate": 1.8011273957158964e-05,
      "loss": 0.0523,
      "step": 642
    },
    {
      "epoch": 2.074313408723748,
      "grad_norm": 0.1010020524263382,
      "learning_rate": 1.800676437429538e-05,
      "loss": 0.0552,
      "step": 643
    },
    {
      "epoch": 2.0775444264943457,
      "grad_norm": 0.0997423455119133,
      "learning_rate": 1.8002254791431793e-05,
      "loss": 0.0662,
      "step": 644
    },
    {
      "epoch": 2.0807754442649435,
      "grad_norm": 0.09295801818370819,
      "learning_rate": 1.7997745208568208e-05,
      "loss": 0.0599,
      "step": 645
    },
    {
      "epoch": 2.0840064620355414,
      "grad_norm": 0.1053297147154808,
      "learning_rate": 1.7993235625704622e-05,
      "loss": 0.064,
      "step": 646
    },
    {
      "epoch": 2.087237479806139,
      "grad_norm": 0.11978495121002197,
      "learning_rate": 1.798872604284104e-05,
      "loss": 0.0727,
      "step": 647
    },
    {
      "epoch": 2.0904684975767367,
      "grad_norm": 0.07878235727548599,
      "learning_rate": 1.7984216459977454e-05,
      "loss": 0.0486,
      "step": 648
    },
    {
      "epoch": 2.0936995153473346,
      "grad_norm": 0.14993903040885925,
      "learning_rate": 1.797970687711387e-05,
      "loss": 0.0458,
      "step": 649
    },
    {
      "epoch": 2.096930533117932,
      "grad_norm": 0.0925765186548233,
      "learning_rate": 1.7975197294250283e-05,
      "loss": 0.0546,
      "step": 650
    },
    {
      "epoch": 2.10016155088853,
      "grad_norm": 0.09530377388000488,
      "learning_rate": 1.7970687711386697e-05,
      "loss": 0.0514,
      "step": 651
    },
    {
      "epoch": 2.1033925686591277,
      "grad_norm": 0.0945788025856018,
      "learning_rate": 1.7966178128523112e-05,
      "loss": 0.0604,
      "step": 652
    },
    {
      "epoch": 2.106623586429725,
      "grad_norm": 0.11486334353685379,
      "learning_rate": 1.7961668545659526e-05,
      "loss": 0.0633,
      "step": 653
    },
    {
      "epoch": 2.109854604200323,
      "grad_norm": 0.1077791377902031,
      "learning_rate": 1.795715896279594e-05,
      "loss": 0.0602,
      "step": 654
    },
    {
      "epoch": 2.113085621970921,
      "grad_norm": 0.10789015889167786,
      "learning_rate": 1.795264937993236e-05,
      "loss": 0.0541,
      "step": 655
    },
    {
      "epoch": 2.1163166397415187,
      "grad_norm": 0.10327862948179245,
      "learning_rate": 1.7948139797068773e-05,
      "loss": 0.0652,
      "step": 656
    },
    {
      "epoch": 2.119547657512116,
      "grad_norm": 0.10486488789319992,
      "learning_rate": 1.7943630214205187e-05,
      "loss": 0.0617,
      "step": 657
    },
    {
      "epoch": 2.122778675282714,
      "grad_norm": 0.0882355272769928,
      "learning_rate": 1.79391206313416e-05,
      "loss": 0.0553,
      "step": 658
    },
    {
      "epoch": 2.126009693053312,
      "grad_norm": 0.08177275210618973,
      "learning_rate": 1.7934611048478016e-05,
      "loss": 0.0522,
      "step": 659
    },
    {
      "epoch": 2.1292407108239093,
      "grad_norm": 0.1455976665019989,
      "learning_rate": 1.793010146561443e-05,
      "loss": 0.0909,
      "step": 660
    },
    {
      "epoch": 2.132471728594507,
      "grad_norm": 0.1145886555314064,
      "learning_rate": 1.7925591882750848e-05,
      "loss": 0.0733,
      "step": 661
    },
    {
      "epoch": 2.135702746365105,
      "grad_norm": 0.1092807874083519,
      "learning_rate": 1.7921082299887262e-05,
      "loss": 0.0581,
      "step": 662
    },
    {
      "epoch": 2.138933764135703,
      "grad_norm": 0.07647505402565002,
      "learning_rate": 1.7916572717023677e-05,
      "loss": 0.0461,
      "step": 663
    },
    {
      "epoch": 2.1421647819063003,
      "grad_norm": 0.09198980778455734,
      "learning_rate": 1.791206313416009e-05,
      "loss": 0.0549,
      "step": 664
    },
    {
      "epoch": 2.145395799676898,
      "grad_norm": 0.10971511900424957,
      "learning_rate": 1.7907553551296506e-05,
      "loss": 0.0643,
      "step": 665
    },
    {
      "epoch": 2.148626817447496,
      "grad_norm": 0.11374619603157043,
      "learning_rate": 1.790304396843292e-05,
      "loss": 0.0754,
      "step": 666
    },
    {
      "epoch": 2.1518578352180935,
      "grad_norm": 0.09252484142780304,
      "learning_rate": 1.7898534385569338e-05,
      "loss": 0.0502,
      "step": 667
    },
    {
      "epoch": 2.1550888529886914,
      "grad_norm": 0.09586004912853241,
      "learning_rate": 1.7894024802705752e-05,
      "loss": 0.0529,
      "step": 668
    },
    {
      "epoch": 2.158319870759289,
      "grad_norm": 0.10206209868192673,
      "learning_rate": 1.7889515219842167e-05,
      "loss": 0.0605,
      "step": 669
    },
    {
      "epoch": 2.161550888529887,
      "grad_norm": 0.15015992522239685,
      "learning_rate": 1.788500563697858e-05,
      "loss": 0.0926,
      "step": 670
    },
    {
      "epoch": 2.1647819063004845,
      "grad_norm": 0.10588102042675018,
      "learning_rate": 1.7880496054114995e-05,
      "loss": 0.0623,
      "step": 671
    },
    {
      "epoch": 2.1680129240710824,
      "grad_norm": 0.09418896585702896,
      "learning_rate": 1.787598647125141e-05,
      "loss": 0.0564,
      "step": 672
    },
    {
      "epoch": 2.1712439418416802,
      "grad_norm": 0.08213125914335251,
      "learning_rate": 1.7871476888387828e-05,
      "loss": 0.0464,
      "step": 673
    },
    {
      "epoch": 2.1744749596122777,
      "grad_norm": 0.09321248531341553,
      "learning_rate": 1.7866967305524242e-05,
      "loss": 0.0531,
      "step": 674
    },
    {
      "epoch": 2.1777059773828755,
      "grad_norm": 0.10642002522945404,
      "learning_rate": 1.7862457722660656e-05,
      "loss": 0.0669,
      "step": 675
    },
    {
      "epoch": 2.1777059773828755,
      "eval_loss": 0.0747215747833252,
      "eval_runtime": 188.1708,
      "eval_samples_per_second": 1.047,
      "eval_steps_per_second": 1.047,
      "step": 675
    },
    {
      "epoch": 2.1809369951534734,
      "grad_norm": 0.12946507334709167,
      "learning_rate": 1.785794813979707e-05,
      "loss": 0.072,
      "step": 676
    },
    {
      "epoch": 2.1841680129240713,
      "grad_norm": 0.10074260830879211,
      "learning_rate": 1.7853438556933485e-05,
      "loss": 0.0521,
      "step": 677
    },
    {
      "epoch": 2.1873990306946687,
      "grad_norm": 0.12798738479614258,
      "learning_rate": 1.78489289740699e-05,
      "loss": 0.0758,
      "step": 678
    },
    {
      "epoch": 2.1906300484652665,
      "grad_norm": 0.10193175077438354,
      "learning_rate": 1.7844419391206317e-05,
      "loss": 0.0632,
      "step": 679
    },
    {
      "epoch": 2.1938610662358644,
      "grad_norm": 0.12570485472679138,
      "learning_rate": 1.783990980834273e-05,
      "loss": 0.0724,
      "step": 680
    },
    {
      "epoch": 2.197092084006462,
      "grad_norm": 0.10074017196893692,
      "learning_rate": 1.7835400225479146e-05,
      "loss": 0.0563,
      "step": 681
    },
    {
      "epoch": 2.2003231017770597,
      "grad_norm": 0.08727949112653732,
      "learning_rate": 1.783089064261556e-05,
      "loss": 0.0524,
      "step": 682
    },
    {
      "epoch": 2.2035541195476576,
      "grad_norm": 0.11030570417642593,
      "learning_rate": 1.7826381059751975e-05,
      "loss": 0.0668,
      "step": 683
    },
    {
      "epoch": 2.2067851373182554,
      "grad_norm": 0.10606499761343002,
      "learning_rate": 1.782187147688839e-05,
      "loss": 0.0606,
      "step": 684
    },
    {
      "epoch": 2.210016155088853,
      "grad_norm": 0.11735937744379044,
      "learning_rate": 1.7817361894024804e-05,
      "loss": 0.0648,
      "step": 685
    },
    {
      "epoch": 2.2132471728594507,
      "grad_norm": 0.106626495718956,
      "learning_rate": 1.7812852311161218e-05,
      "loss": 0.0649,
      "step": 686
    },
    {
      "epoch": 2.2164781906300486,
      "grad_norm": 0.12231657654047012,
      "learning_rate": 1.7808342728297632e-05,
      "loss": 0.0702,
      "step": 687
    },
    {
      "epoch": 2.219709208400646,
      "grad_norm": 0.08800094574689865,
      "learning_rate": 1.780383314543405e-05,
      "loss": 0.0515,
      "step": 688
    },
    {
      "epoch": 2.222940226171244,
      "grad_norm": 0.08806774020195007,
      "learning_rate": 1.7799323562570465e-05,
      "loss": 0.0493,
      "step": 689
    },
    {
      "epoch": 2.2261712439418417,
      "grad_norm": 0.10804681479930878,
      "learning_rate": 1.779481397970688e-05,
      "loss": 0.0604,
      "step": 690
    },
    {
      "epoch": 2.2294022617124396,
      "grad_norm": 0.11405564099550247,
      "learning_rate": 1.7790304396843293e-05,
      "loss": 0.0597,
      "step": 691
    },
    {
      "epoch": 2.232633279483037,
      "grad_norm": 0.11010053753852844,
      "learning_rate": 1.7785794813979708e-05,
      "loss": 0.0634,
      "step": 692
    },
    {
      "epoch": 2.235864297253635,
      "grad_norm": 0.10657312721014023,
      "learning_rate": 1.7781285231116122e-05,
      "loss": 0.0539,
      "step": 693
    },
    {
      "epoch": 2.2390953150242328,
      "grad_norm": 0.08584710210561752,
      "learning_rate": 1.7776775648252536e-05,
      "loss": 0.0571,
      "step": 694
    },
    {
      "epoch": 2.24232633279483,
      "grad_norm": 0.10155533254146576,
      "learning_rate": 1.777226606538895e-05,
      "loss": 0.0597,
      "step": 695
    },
    {
      "epoch": 2.245557350565428,
      "grad_norm": 0.11395770311355591,
      "learning_rate": 1.7767756482525365e-05,
      "loss": 0.0675,
      "step": 696
    },
    {
      "epoch": 2.248788368336026,
      "grad_norm": 0.11109079420566559,
      "learning_rate": 1.7763246899661783e-05,
      "loss": 0.062,
      "step": 697
    },
    {
      "epoch": 2.2520193861066238,
      "grad_norm": 0.13479241728782654,
      "learning_rate": 1.7758737316798197e-05,
      "loss": 0.0782,
      "step": 698
    },
    {
      "epoch": 2.255250403877221,
      "grad_norm": 0.12003345042467117,
      "learning_rate": 1.7754227733934612e-05,
      "loss": 0.0683,
      "step": 699
    },
    {
      "epoch": 2.258481421647819,
      "grad_norm": 0.13395312428474426,
      "learning_rate": 1.7749718151071026e-05,
      "loss": 0.0764,
      "step": 700
    },
    {
      "epoch": 2.261712439418417,
      "grad_norm": 0.10561169683933258,
      "learning_rate": 1.774520856820744e-05,
      "loss": 0.0552,
      "step": 701
    },
    {
      "epoch": 2.2649434571890144,
      "grad_norm": 0.1412249207496643,
      "learning_rate": 1.7740698985343855e-05,
      "loss": 0.0812,
      "step": 702
    },
    {
      "epoch": 2.268174474959612,
      "grad_norm": 0.11307451128959656,
      "learning_rate": 1.7736189402480273e-05,
      "loss": 0.0671,
      "step": 703
    },
    {
      "epoch": 2.27140549273021,
      "grad_norm": 0.10989584773778915,
      "learning_rate": 1.7731679819616687e-05,
      "loss": 0.059,
      "step": 704
    },
    {
      "epoch": 2.274636510500808,
      "grad_norm": 0.0964912474155426,
      "learning_rate": 1.77271702367531e-05,
      "loss": 0.0515,
      "step": 705
    },
    {
      "epoch": 2.2778675282714054,
      "grad_norm": 0.09640849381685257,
      "learning_rate": 1.7722660653889516e-05,
      "loss": 0.0548,
      "step": 706
    },
    {
      "epoch": 2.2810985460420032,
      "grad_norm": 0.08393755555152893,
      "learning_rate": 1.771815107102593e-05,
      "loss": 0.0477,
      "step": 707
    },
    {
      "epoch": 2.284329563812601,
      "grad_norm": 0.08865144103765488,
      "learning_rate": 1.7713641488162348e-05,
      "loss": 0.0527,
      "step": 708
    },
    {
      "epoch": 2.2875605815831985,
      "grad_norm": 0.10840681195259094,
      "learning_rate": 1.7709131905298762e-05,
      "loss": 0.061,
      "step": 709
    },
    {
      "epoch": 2.2907915993537964,
      "grad_norm": 0.1336364448070526,
      "learning_rate": 1.7704622322435177e-05,
      "loss": 0.0655,
      "step": 710
    },
    {
      "epoch": 2.2940226171243943,
      "grad_norm": 0.10070191323757172,
      "learning_rate": 1.770011273957159e-05,
      "loss": 0.0599,
      "step": 711
    },
    {
      "epoch": 2.297253634894992,
      "grad_norm": 0.12378398329019547,
      "learning_rate": 1.7695603156708006e-05,
      "loss": 0.0712,
      "step": 712
    },
    {
      "epoch": 2.3004846526655895,
      "grad_norm": 0.08809908479452133,
      "learning_rate": 1.769109357384442e-05,
      "loss": 0.0419,
      "step": 713
    },
    {
      "epoch": 2.3037156704361874,
      "grad_norm": 0.10536797344684601,
      "learning_rate": 1.7686583990980838e-05,
      "loss": 0.062,
      "step": 714
    },
    {
      "epoch": 2.3069466882067853,
      "grad_norm": 0.09373629838228226,
      "learning_rate": 1.7682074408117252e-05,
      "loss": 0.0455,
      "step": 715
    },
    {
      "epoch": 2.3101777059773827,
      "grad_norm": 0.08422086387872696,
      "learning_rate": 1.7677564825253667e-05,
      "loss": 0.0481,
      "step": 716
    },
    {
      "epoch": 2.3134087237479806,
      "grad_norm": 0.12226711213588715,
      "learning_rate": 1.767305524239008e-05,
      "loss": 0.0713,
      "step": 717
    },
    {
      "epoch": 2.3166397415185784,
      "grad_norm": 0.11593876034021378,
      "learning_rate": 1.7668545659526495e-05,
      "loss": 0.065,
      "step": 718
    },
    {
      "epoch": 2.3198707592891763,
      "grad_norm": 0.10055369138717651,
      "learning_rate": 1.766403607666291e-05,
      "loss": 0.0515,
      "step": 719
    },
    {
      "epoch": 2.3231017770597737,
      "grad_norm": 0.1200050637125969,
      "learning_rate": 1.7659526493799328e-05,
      "loss": 0.0664,
      "step": 720
    },
    {
      "epoch": 2.3263327948303716,
      "grad_norm": 0.10001233220100403,
      "learning_rate": 1.7655016910935742e-05,
      "loss": 0.0578,
      "step": 721
    },
    {
      "epoch": 2.3295638126009695,
      "grad_norm": 0.08621415495872498,
      "learning_rate": 1.7650507328072156e-05,
      "loss": 0.0426,
      "step": 722
    },
    {
      "epoch": 2.332794830371567,
      "grad_norm": 0.08662088960409164,
      "learning_rate": 1.764599774520857e-05,
      "loss": 0.048,
      "step": 723
    },
    {
      "epoch": 2.3360258481421647,
      "grad_norm": 0.09761569648981094,
      "learning_rate": 1.7641488162344985e-05,
      "loss": 0.0585,
      "step": 724
    },
    {
      "epoch": 2.3392568659127626,
      "grad_norm": 0.1272287666797638,
      "learning_rate": 1.76369785794814e-05,
      "loss": 0.0722,
      "step": 725
    },
    {
      "epoch": 2.3424878836833605,
      "grad_norm": 0.0827430784702301,
      "learning_rate": 1.7632468996617814e-05,
      "loss": 0.0517,
      "step": 726
    },
    {
      "epoch": 2.345718901453958,
      "grad_norm": 0.08261015266180038,
      "learning_rate": 1.7627959413754228e-05,
      "loss": 0.0464,
      "step": 727
    },
    {
      "epoch": 2.3489499192245558,
      "grad_norm": 0.10019004344940186,
      "learning_rate": 1.7623449830890643e-05,
      "loss": 0.0539,
      "step": 728
    },
    {
      "epoch": 2.3521809369951536,
      "grad_norm": 0.11860109120607376,
      "learning_rate": 1.761894024802706e-05,
      "loss": 0.0604,
      "step": 729
    },
    {
      "epoch": 2.355411954765751,
      "grad_norm": 0.13498210906982422,
      "learning_rate": 1.7614430665163475e-05,
      "loss": 0.0497,
      "step": 730
    },
    {
      "epoch": 2.358642972536349,
      "grad_norm": 0.09867555648088455,
      "learning_rate": 1.760992108229989e-05,
      "loss": 0.0572,
      "step": 731
    },
    {
      "epoch": 2.361873990306947,
      "grad_norm": 0.10520780086517334,
      "learning_rate": 1.7605411499436304e-05,
      "loss": 0.0613,
      "step": 732
    },
    {
      "epoch": 2.3651050080775446,
      "grad_norm": 0.1396726369857788,
      "learning_rate": 1.7600901916572718e-05,
      "loss": 0.0808,
      "step": 733
    },
    {
      "epoch": 2.368336025848142,
      "grad_norm": 0.09852424263954163,
      "learning_rate": 1.7596392333709132e-05,
      "loss": 0.0602,
      "step": 734
    },
    {
      "epoch": 2.37156704361874,
      "grad_norm": 0.08897744864225388,
      "learning_rate": 1.7591882750845547e-05,
      "loss": 0.0463,
      "step": 735
    },
    {
      "epoch": 2.374798061389338,
      "grad_norm": 0.12664619088172913,
      "learning_rate": 1.758737316798196e-05,
      "loss": 0.0731,
      "step": 736
    },
    {
      "epoch": 2.378029079159935,
      "grad_norm": 0.0975411906838417,
      "learning_rate": 1.7582863585118376e-05,
      "loss": 0.0541,
      "step": 737
    },
    {
      "epoch": 2.381260096930533,
      "grad_norm": 0.10056427121162415,
      "learning_rate": 1.7578354002254793e-05,
      "loss": 0.0564,
      "step": 738
    },
    {
      "epoch": 2.384491114701131,
      "grad_norm": 0.0751773938536644,
      "learning_rate": 1.7573844419391208e-05,
      "loss": 0.043,
      "step": 739
    },
    {
      "epoch": 2.387722132471729,
      "grad_norm": 0.12571515142917633,
      "learning_rate": 1.7569334836527622e-05,
      "loss": 0.0715,
      "step": 740
    },
    {
      "epoch": 2.3909531502423262,
      "grad_norm": 0.09152042865753174,
      "learning_rate": 1.7564825253664037e-05,
      "loss": 0.0513,
      "step": 741
    },
    {
      "epoch": 2.394184168012924,
      "grad_norm": 0.16221173107624054,
      "learning_rate": 1.756031567080045e-05,
      "loss": 0.1003,
      "step": 742
    },
    {
      "epoch": 2.397415185783522,
      "grad_norm": 0.09910274296998978,
      "learning_rate": 1.7555806087936865e-05,
      "loss": 0.0543,
      "step": 743
    },
    {
      "epoch": 2.4006462035541194,
      "grad_norm": 0.10756971687078476,
      "learning_rate": 1.7551296505073283e-05,
      "loss": 0.0573,
      "step": 744
    },
    {
      "epoch": 2.4038772213247173,
      "grad_norm": 0.08702822029590607,
      "learning_rate": 1.7546786922209697e-05,
      "loss": 0.047,
      "step": 745
    },
    {
      "epoch": 2.407108239095315,
      "grad_norm": 0.14440016448497772,
      "learning_rate": 1.7542277339346112e-05,
      "loss": 0.0685,
      "step": 746
    },
    {
      "epoch": 2.410339256865913,
      "grad_norm": 0.09141986817121506,
      "learning_rate": 1.7537767756482526e-05,
      "loss": 0.0521,
      "step": 747
    },
    {
      "epoch": 2.4135702746365104,
      "grad_norm": 0.12515199184417725,
      "learning_rate": 1.753325817361894e-05,
      "loss": 0.0605,
      "step": 748
    },
    {
      "epoch": 2.4168012924071083,
      "grad_norm": 0.12448819726705551,
      "learning_rate": 1.7528748590755355e-05,
      "loss": 0.0723,
      "step": 749
    },
    {
      "epoch": 2.420032310177706,
      "grad_norm": 0.13118943572044373,
      "learning_rate": 1.7524239007891773e-05,
      "loss": 0.0654,
      "step": 750
    },
    {
      "epoch": 2.420032310177706,
      "eval_loss": 0.07308099418878555,
      "eval_runtime": 188.3306,
      "eval_samples_per_second": 1.046,
      "eval_steps_per_second": 1.046,
      "step": 750
    },
    {
      "epoch": 2.4232633279483036,
      "grad_norm": 0.1215345561504364,
      "learning_rate": 1.7519729425028187e-05,
      "loss": 0.0644,
      "step": 751
    },
    {
      "epoch": 2.4264943457189014,
      "grad_norm": 0.12855304777622223,
      "learning_rate": 1.75152198421646e-05,
      "loss": 0.0666,
      "step": 752
    },
    {
      "epoch": 2.4297253634894993,
      "grad_norm": 0.11538267880678177,
      "learning_rate": 1.7510710259301016e-05,
      "loss": 0.0545,
      "step": 753
    },
    {
      "epoch": 2.432956381260097,
      "grad_norm": 0.10273373872041702,
      "learning_rate": 1.750620067643743e-05,
      "loss": 0.0594,
      "step": 754
    },
    {
      "epoch": 2.4361873990306946,
      "grad_norm": 0.10953179746866226,
      "learning_rate": 1.7501691093573845e-05,
      "loss": 0.0587,
      "step": 755
    },
    {
      "epoch": 2.4394184168012925,
      "grad_norm": 0.09215240180492401,
      "learning_rate": 1.7497181510710263e-05,
      "loss": 0.0501,
      "step": 756
    },
    {
      "epoch": 2.4426494345718903,
      "grad_norm": 0.11669941246509552,
      "learning_rate": 1.7492671927846677e-05,
      "loss": 0.0585,
      "step": 757
    },
    {
      "epoch": 2.4458804523424877,
      "grad_norm": 0.11698901653289795,
      "learning_rate": 1.748816234498309e-05,
      "loss": 0.0551,
      "step": 758
    },
    {
      "epoch": 2.4491114701130856,
      "grad_norm": 0.1258348822593689,
      "learning_rate": 1.7483652762119506e-05,
      "loss": 0.0629,
      "step": 759
    },
    {
      "epoch": 2.4523424878836835,
      "grad_norm": 0.12607377767562866,
      "learning_rate": 1.747914317925592e-05,
      "loss": 0.0728,
      "step": 760
    },
    {
      "epoch": 2.4555735056542813,
      "grad_norm": 0.0982760339975357,
      "learning_rate": 1.7474633596392334e-05,
      "loss": 0.051,
      "step": 761
    },
    {
      "epoch": 2.4588045234248788,
      "grad_norm": 0.15601739287376404,
      "learning_rate": 1.7470124013528752e-05,
      "loss": 0.076,
      "step": 762
    },
    {
      "epoch": 2.4620355411954766,
      "grad_norm": 0.13090789318084717,
      "learning_rate": 1.7465614430665167e-05,
      "loss": 0.0739,
      "step": 763
    },
    {
      "epoch": 2.4652665589660745,
      "grad_norm": 0.10627159476280212,
      "learning_rate": 1.746110484780158e-05,
      "loss": 0.0509,
      "step": 764
    },
    {
      "epoch": 2.468497576736672,
      "grad_norm": 0.07108946144580841,
      "learning_rate": 1.7456595264937995e-05,
      "loss": 0.039,
      "step": 765
    },
    {
      "epoch": 2.47172859450727,
      "grad_norm": 0.14733023941516876,
      "learning_rate": 1.745208568207441e-05,
      "loss": 0.0788,
      "step": 766
    },
    {
      "epoch": 2.4749596122778676,
      "grad_norm": 0.10821715742349625,
      "learning_rate": 1.7447576099210824e-05,
      "loss": 0.0524,
      "step": 767
    },
    {
      "epoch": 2.4781906300484655,
      "grad_norm": 0.11846361309289932,
      "learning_rate": 1.744306651634724e-05,
      "loss": 0.0579,
      "step": 768
    },
    {
      "epoch": 2.481421647819063,
      "grad_norm": 0.10738200694322586,
      "learning_rate": 1.7438556933483653e-05,
      "loss": 0.0589,
      "step": 769
    },
    {
      "epoch": 2.484652665589661,
      "grad_norm": 0.1159975603222847,
      "learning_rate": 1.743404735062007e-05,
      "loss": 0.0565,
      "step": 770
    },
    {
      "epoch": 2.4878836833602587,
      "grad_norm": 0.10056610405445099,
      "learning_rate": 1.7429537767756485e-05,
      "loss": 0.0538,
      "step": 771
    },
    {
      "epoch": 2.491114701130856,
      "grad_norm": 0.10329104959964752,
      "learning_rate": 1.74250281848929e-05,
      "loss": 0.0566,
      "step": 772
    },
    {
      "epoch": 2.494345718901454,
      "grad_norm": 0.1422542929649353,
      "learning_rate": 1.7420518602029314e-05,
      "loss": 0.0747,
      "step": 773
    },
    {
      "epoch": 2.497576736672052,
      "grad_norm": 0.12898680567741394,
      "learning_rate": 1.741600901916573e-05,
      "loss": 0.0735,
      "step": 774
    },
    {
      "epoch": 2.5008077544426497,
      "grad_norm": 0.13066206872463226,
      "learning_rate": 1.7411499436302143e-05,
      "loss": 0.062,
      "step": 775
    },
    {
      "epoch": 2.504038772213247,
      "grad_norm": 0.12379497289657593,
      "learning_rate": 1.7406989853438557e-05,
      "loss": 0.0631,
      "step": 776
    },
    {
      "epoch": 2.507269789983845,
      "grad_norm": 0.1296347826719284,
      "learning_rate": 1.740248027057497e-05,
      "loss": 0.0634,
      "step": 777
    },
    {
      "epoch": 2.5105008077544424,
      "grad_norm": 0.10818596184253693,
      "learning_rate": 1.7397970687711386e-05,
      "loss": 0.0606,
      "step": 778
    },
    {
      "epoch": 2.5137318255250403,
      "grad_norm": 0.12639783322811127,
      "learning_rate": 1.73934611048478e-05,
      "loss": 0.0459,
      "step": 779
    },
    {
      "epoch": 2.516962843295638,
      "grad_norm": 0.1167321428656578,
      "learning_rate": 1.7388951521984218e-05,
      "loss": 0.0667,
      "step": 780
    },
    {
      "epoch": 2.520193861066236,
      "grad_norm": 0.10153870284557343,
      "learning_rate": 1.7384441939120632e-05,
      "loss": 0.0589,
      "step": 781
    },
    {
      "epoch": 2.523424878836834,
      "grad_norm": 0.09986142814159393,
      "learning_rate": 1.7379932356257047e-05,
      "loss": 0.045,
      "step": 782
    },
    {
      "epoch": 2.5266558966074313,
      "grad_norm": 0.10662157833576202,
      "learning_rate": 1.737542277339346e-05,
      "loss": 0.0586,
      "step": 783
    },
    {
      "epoch": 2.529886914378029,
      "grad_norm": 0.11709077656269073,
      "learning_rate": 1.7370913190529876e-05,
      "loss": 0.0613,
      "step": 784
    },
    {
      "epoch": 2.5331179321486266,
      "grad_norm": 0.13120310008525848,
      "learning_rate": 1.736640360766629e-05,
      "loss": 0.0664,
      "step": 785
    },
    {
      "epoch": 2.5363489499192244,
      "grad_norm": 0.13849826157093048,
      "learning_rate": 1.7361894024802708e-05,
      "loss": 0.0673,
      "step": 786
    },
    {
      "epoch": 2.5395799676898223,
      "grad_norm": 0.08833606541156769,
      "learning_rate": 1.7357384441939122e-05,
      "loss": 0.0459,
      "step": 787
    },
    {
      "epoch": 2.54281098546042,
      "grad_norm": 0.09421700984239578,
      "learning_rate": 1.7352874859075537e-05,
      "loss": 0.0481,
      "step": 788
    },
    {
      "epoch": 2.546042003231018,
      "grad_norm": 0.1201411634683609,
      "learning_rate": 1.734836527621195e-05,
      "loss": 0.0608,
      "step": 789
    },
    {
      "epoch": 2.5492730210016155,
      "grad_norm": 0.09896653145551682,
      "learning_rate": 1.7343855693348365e-05,
      "loss": 0.0465,
      "step": 790
    },
    {
      "epoch": 2.5525040387722133,
      "grad_norm": 0.12088964134454727,
      "learning_rate": 1.7339346110484783e-05,
      "loss": 0.0614,
      "step": 791
    },
    {
      "epoch": 2.5557350565428107,
      "grad_norm": 0.11183801293373108,
      "learning_rate": 1.7334836527621198e-05,
      "loss": 0.0545,
      "step": 792
    },
    {
      "epoch": 2.5589660743134086,
      "grad_norm": 0.11126703768968582,
      "learning_rate": 1.7330326944757612e-05,
      "loss": 0.0509,
      "step": 793
    },
    {
      "epoch": 2.5621970920840065,
      "grad_norm": 0.1374976634979248,
      "learning_rate": 1.7325817361894026e-05,
      "loss": 0.0664,
      "step": 794
    },
    {
      "epoch": 2.5654281098546043,
      "grad_norm": 0.16783633828163147,
      "learning_rate": 1.732130777903044e-05,
      "loss": 0.0868,
      "step": 795
    },
    {
      "epoch": 2.568659127625202,
      "grad_norm": 0.11534145474433899,
      "learning_rate": 1.7316798196166855e-05,
      "loss": 0.053,
      "step": 796
    },
    {
      "epoch": 2.5718901453957996,
      "grad_norm": 0.13769778609275818,
      "learning_rate": 1.7312288613303273e-05,
      "loss": 0.0756,
      "step": 797
    },
    {
      "epoch": 2.5751211631663975,
      "grad_norm": 0.09577422589063644,
      "learning_rate": 1.7307779030439687e-05,
      "loss": 0.0441,
      "step": 798
    },
    {
      "epoch": 2.578352180936995,
      "grad_norm": 0.11375096440315247,
      "learning_rate": 1.73032694475761e-05,
      "loss": 0.0524,
      "step": 799
    },
    {
      "epoch": 2.581583198707593,
      "grad_norm": 0.11465324461460114,
      "learning_rate": 1.7298759864712516e-05,
      "loss": 0.0526,
      "step": 800
    },
    {
      "epoch": 2.5848142164781907,
      "grad_norm": 0.11597500741481781,
      "learning_rate": 1.729425028184893e-05,
      "loss": 0.0591,
      "step": 801
    },
    {
      "epoch": 2.5880452342487885,
      "grad_norm": 0.09817709028720856,
      "learning_rate": 1.7289740698985345e-05,
      "loss": 0.0503,
      "step": 802
    },
    {
      "epoch": 2.5912762520193864,
      "grad_norm": 0.10352802276611328,
      "learning_rate": 1.7285231116121763e-05,
      "loss": 0.0461,
      "step": 803
    },
    {
      "epoch": 2.594507269789984,
      "grad_norm": 0.12035888433456421,
      "learning_rate": 1.7280721533258177e-05,
      "loss": 0.0576,
      "step": 804
    },
    {
      "epoch": 2.5977382875605817,
      "grad_norm": 0.12561960518360138,
      "learning_rate": 1.727621195039459e-05,
      "loss": 0.0599,
      "step": 805
    },
    {
      "epoch": 2.600969305331179,
      "grad_norm": 0.11348681896924973,
      "learning_rate": 1.7271702367531006e-05,
      "loss": 0.0503,
      "step": 806
    },
    {
      "epoch": 2.604200323101777,
      "grad_norm": 0.09772437810897827,
      "learning_rate": 1.726719278466742e-05,
      "loss": 0.0471,
      "step": 807
    },
    {
      "epoch": 2.607431340872375,
      "grad_norm": 0.10316241532564163,
      "learning_rate": 1.7262683201803835e-05,
      "loss": 0.0514,
      "step": 808
    },
    {
      "epoch": 2.6106623586429727,
      "grad_norm": 0.11204390227794647,
      "learning_rate": 1.725817361894025e-05,
      "loss": 0.0578,
      "step": 809
    },
    {
      "epoch": 2.61389337641357,
      "grad_norm": 0.10899617522954941,
      "learning_rate": 1.7253664036076663e-05,
      "loss": 0.0495,
      "step": 810
    },
    {
      "epoch": 2.617124394184168,
      "grad_norm": 0.1386026293039322,
      "learning_rate": 1.724915445321308e-05,
      "loss": 0.0542,
      "step": 811
    },
    {
      "epoch": 2.620355411954766,
      "grad_norm": 0.13927608728408813,
      "learning_rate": 1.7244644870349495e-05,
      "loss": 0.0628,
      "step": 812
    },
    {
      "epoch": 2.6235864297253633,
      "grad_norm": 0.099461629986763,
      "learning_rate": 1.724013528748591e-05,
      "loss": 0.0501,
      "step": 813
    },
    {
      "epoch": 2.626817447495961,
      "grad_norm": 0.09142296761274338,
      "learning_rate": 1.7235625704622324e-05,
      "loss": 0.0475,
      "step": 814
    },
    {
      "epoch": 2.630048465266559,
      "grad_norm": 0.12531687319278717,
      "learning_rate": 1.723111612175874e-05,
      "loss": 0.059,
      "step": 815
    },
    {
      "epoch": 2.633279483037157,
      "grad_norm": 0.1252615749835968,
      "learning_rate": 1.7226606538895153e-05,
      "loss": 0.0589,
      "step": 816
    },
    {
      "epoch": 2.6365105008077543,
      "grad_norm": 0.12725740671157837,
      "learning_rate": 1.7222096956031567e-05,
      "loss": 0.0522,
      "step": 817
    },
    {
      "epoch": 2.639741518578352,
      "grad_norm": 0.12746059894561768,
      "learning_rate": 1.7217587373167982e-05,
      "loss": 0.0661,
      "step": 818
    },
    {
      "epoch": 2.64297253634895,
      "grad_norm": 0.2133682370185852,
      "learning_rate": 1.7213077790304396e-05,
      "loss": 0.0639,
      "step": 819
    },
    {
      "epoch": 2.6462035541195474,
      "grad_norm": 0.11452341079711914,
      "learning_rate": 1.720856820744081e-05,
      "loss": 0.0512,
      "step": 820
    },
    {
      "epoch": 2.6494345718901453,
      "grad_norm": 0.12344635277986526,
      "learning_rate": 1.720405862457723e-05,
      "loss": 0.0503,
      "step": 821
    },
    {
      "epoch": 2.652665589660743,
      "grad_norm": 0.12654437124729156,
      "learning_rate": 1.7199549041713643e-05,
      "loss": 0.0543,
      "step": 822
    },
    {
      "epoch": 2.655896607431341,
      "grad_norm": 0.12805619835853577,
      "learning_rate": 1.7195039458850057e-05,
      "loss": 0.0646,
      "step": 823
    },
    {
      "epoch": 2.6591276252019385,
      "grad_norm": 0.11218256503343582,
      "learning_rate": 1.719052987598647e-05,
      "loss": 0.0575,
      "step": 824
    },
    {
      "epoch": 2.6623586429725363,
      "grad_norm": 0.12950399518013,
      "learning_rate": 1.7186020293122886e-05,
      "loss": 0.0673,
      "step": 825
    },
    {
      "epoch": 2.6623586429725363,
      "eval_loss": 0.07266557961702347,
      "eval_runtime": 187.9396,
      "eval_samples_per_second": 1.048,
      "eval_steps_per_second": 1.048,
      "step": 825
    },
    {
      "epoch": 2.665589660743134,
      "grad_norm": 0.11642561107873917,
      "learning_rate": 1.71815107102593e-05,
      "loss": 0.0664,
      "step": 826
    },
    {
      "epoch": 2.6688206785137316,
      "grad_norm": 0.09707733243703842,
      "learning_rate": 1.7177001127395718e-05,
      "loss": 0.0503,
      "step": 827
    },
    {
      "epoch": 2.6720516962843295,
      "grad_norm": 0.07535319775342941,
      "learning_rate": 1.7172491544532133e-05,
      "loss": 0.038,
      "step": 828
    },
    {
      "epoch": 2.6752827140549273,
      "grad_norm": 0.12683290243148804,
      "learning_rate": 1.7167981961668547e-05,
      "loss": 0.0614,
      "step": 829
    },
    {
      "epoch": 2.678513731825525,
      "grad_norm": 0.1531742513179779,
      "learning_rate": 1.716347237880496e-05,
      "loss": 0.069,
      "step": 830
    },
    {
      "epoch": 2.6817447495961226,
      "grad_norm": 0.13030219078063965,
      "learning_rate": 1.7158962795941376e-05,
      "loss": 0.0593,
      "step": 831
    },
    {
      "epoch": 2.6849757673667205,
      "grad_norm": 0.13288383185863495,
      "learning_rate": 1.715445321307779e-05,
      "loss": 0.0607,
      "step": 832
    },
    {
      "epoch": 2.6882067851373184,
      "grad_norm": 0.1245107427239418,
      "learning_rate": 1.7149943630214208e-05,
      "loss": 0.059,
      "step": 833
    },
    {
      "epoch": 2.691437802907916,
      "grad_norm": 0.10677826404571533,
      "learning_rate": 1.7145434047350622e-05,
      "loss": 0.0532,
      "step": 834
    },
    {
      "epoch": 2.6946688206785137,
      "grad_norm": 0.11118808388710022,
      "learning_rate": 1.7140924464487037e-05,
      "loss": 0.055,
      "step": 835
    },
    {
      "epoch": 2.6978998384491115,
      "grad_norm": 0.11494432389736176,
      "learning_rate": 1.713641488162345e-05,
      "loss": 0.0556,
      "step": 836
    },
    {
      "epoch": 2.7011308562197094,
      "grad_norm": 0.14139457046985626,
      "learning_rate": 1.7131905298759865e-05,
      "loss": 0.0687,
      "step": 837
    },
    {
      "epoch": 2.704361873990307,
      "grad_norm": 0.12973995506763458,
      "learning_rate": 1.712739571589628e-05,
      "loss": 0.0639,
      "step": 838
    },
    {
      "epoch": 2.7075928917609047,
      "grad_norm": 0.12217195332050323,
      "learning_rate": 1.7122886133032698e-05,
      "loss": 0.0636,
      "step": 839
    },
    {
      "epoch": 2.7108239095315025,
      "grad_norm": 0.08900095522403717,
      "learning_rate": 1.7118376550169112e-05,
      "loss": 0.0491,
      "step": 840
    },
    {
      "epoch": 2.7140549273021,
      "grad_norm": 0.12005368620157242,
      "learning_rate": 1.7113866967305526e-05,
      "loss": 0.0605,
      "step": 841
    },
    {
      "epoch": 2.717285945072698,
      "grad_norm": 0.1201101690530777,
      "learning_rate": 1.710935738444194e-05,
      "loss": 0.0597,
      "step": 842
    },
    {
      "epoch": 2.7205169628432957,
      "grad_norm": 0.12422793358564377,
      "learning_rate": 1.7104847801578355e-05,
      "loss": 0.0604,
      "step": 843
    },
    {
      "epoch": 2.7237479806138936,
      "grad_norm": 0.11504203826189041,
      "learning_rate": 1.710033821871477e-05,
      "loss": 0.0567,
      "step": 844
    },
    {
      "epoch": 2.726978998384491,
      "grad_norm": 0.13158410787582397,
      "learning_rate": 1.7095828635851187e-05,
      "loss": 0.0751,
      "step": 845
    },
    {
      "epoch": 2.730210016155089,
      "grad_norm": 0.13026569783687592,
      "learning_rate": 1.70913190529876e-05,
      "loss": 0.057,
      "step": 846
    },
    {
      "epoch": 2.7334410339256867,
      "grad_norm": 0.5227922201156616,
      "learning_rate": 1.7086809470124016e-05,
      "loss": 0.0567,
      "step": 847
    },
    {
      "epoch": 2.736672051696284,
      "grad_norm": 0.08213207870721817,
      "learning_rate": 1.708229988726043e-05,
      "loss": 0.0385,
      "step": 848
    },
    {
      "epoch": 2.739903069466882,
      "grad_norm": 0.14717501401901245,
      "learning_rate": 1.7077790304396845e-05,
      "loss": 0.0746,
      "step": 849
    },
    {
      "epoch": 2.74313408723748,
      "grad_norm": 0.11484480649232864,
      "learning_rate": 1.707328072153326e-05,
      "loss": 0.0513,
      "step": 850
    },
    {
      "epoch": 2.7463651050080777,
      "grad_norm": 0.13454794883728027,
      "learning_rate": 1.7068771138669674e-05,
      "loss": 0.0668,
      "step": 851
    },
    {
      "epoch": 2.749596122778675,
      "grad_norm": 0.16599783301353455,
      "learning_rate": 1.706426155580609e-05,
      "loss": 0.0659,
      "step": 852
    },
    {
      "epoch": 2.752827140549273,
      "grad_norm": 0.11365890502929688,
      "learning_rate": 1.7059751972942506e-05,
      "loss": 0.0591,
      "step": 853
    },
    {
      "epoch": 2.756058158319871,
      "grad_norm": 0.12964101135730743,
      "learning_rate": 1.705524239007892e-05,
      "loss": 0.0598,
      "step": 854
    },
    {
      "epoch": 2.7592891760904683,
      "grad_norm": 0.10415180772542953,
      "learning_rate": 1.7050732807215335e-05,
      "loss": 0.0499,
      "step": 855
    },
    {
      "epoch": 2.762520193861066,
      "grad_norm": 0.1433461755514145,
      "learning_rate": 1.704622322435175e-05,
      "loss": 0.0668,
      "step": 856
    },
    {
      "epoch": 2.765751211631664,
      "grad_norm": 0.12921610474586487,
      "learning_rate": 1.7041713641488163e-05,
      "loss": 0.0525,
      "step": 857
    },
    {
      "epoch": 2.768982229402262,
      "grad_norm": 0.11878325045108795,
      "learning_rate": 1.7037204058624578e-05,
      "loss": 0.06,
      "step": 858
    },
    {
      "epoch": 2.7722132471728593,
      "grad_norm": 0.09403812140226364,
      "learning_rate": 1.7032694475760992e-05,
      "loss": 0.0472,
      "step": 859
    },
    {
      "epoch": 2.775444264943457,
      "grad_norm": 0.09613174945116043,
      "learning_rate": 1.7028184892897407e-05,
      "loss": 0.0508,
      "step": 860
    },
    {
      "epoch": 2.778675282714055,
      "grad_norm": 0.11891157180070877,
      "learning_rate": 1.702367531003382e-05,
      "loss": 0.0589,
      "step": 861
    },
    {
      "epoch": 2.7819063004846525,
      "grad_norm": 0.1563875377178192,
      "learning_rate": 1.701916572717024e-05,
      "loss": 0.0794,
      "step": 862
    },
    {
      "epoch": 2.7851373182552503,
      "grad_norm": 0.13382849097251892,
      "learning_rate": 1.7014656144306653e-05,
      "loss": 0.0683,
      "step": 863
    },
    {
      "epoch": 2.788368336025848,
      "grad_norm": 0.15156099200248718,
      "learning_rate": 1.7010146561443067e-05,
      "loss": 0.0694,
      "step": 864
    },
    {
      "epoch": 2.791599353796446,
      "grad_norm": 0.12621727585792542,
      "learning_rate": 1.7005636978579482e-05,
      "loss": 0.0567,
      "step": 865
    },
    {
      "epoch": 2.7948303715670435,
      "grad_norm": 0.15388263761997223,
      "learning_rate": 1.7001127395715896e-05,
      "loss": 0.0738,
      "step": 866
    },
    {
      "epoch": 2.7980613893376414,
      "grad_norm": 0.1349688321352005,
      "learning_rate": 1.699661781285231e-05,
      "loss": 0.0546,
      "step": 867
    },
    {
      "epoch": 2.8012924071082392,
      "grad_norm": 0.11894084513187408,
      "learning_rate": 1.6992108229988725e-05,
      "loss": 0.0538,
      "step": 868
    },
    {
      "epoch": 2.8045234248788367,
      "grad_norm": 0.21414369344711304,
      "learning_rate": 1.6987598647125143e-05,
      "loss": 0.0511,
      "step": 869
    },
    {
      "epoch": 2.8077544426494345,
      "grad_norm": 0.1107967421412468,
      "learning_rate": 1.6983089064261557e-05,
      "loss": 0.0551,
      "step": 870
    },
    {
      "epoch": 2.8109854604200324,
      "grad_norm": 0.16710782051086426,
      "learning_rate": 1.697857948139797e-05,
      "loss": 0.0805,
      "step": 871
    },
    {
      "epoch": 2.8142164781906303,
      "grad_norm": 0.12345987558364868,
      "learning_rate": 1.6974069898534386e-05,
      "loss": 0.0576,
      "step": 872
    },
    {
      "epoch": 2.8174474959612277,
      "grad_norm": 0.11037538200616837,
      "learning_rate": 1.69695603156708e-05,
      "loss": 0.0486,
      "step": 873
    },
    {
      "epoch": 2.8206785137318255,
      "grad_norm": 0.10175740718841553,
      "learning_rate": 1.6965050732807218e-05,
      "loss": 0.0529,
      "step": 874
    },
    {
      "epoch": 2.8239095315024234,
      "grad_norm": 0.1126103326678276,
      "learning_rate": 1.6960541149943633e-05,
      "loss": 0.0546,
      "step": 875
    },
    {
      "epoch": 2.827140549273021,
      "grad_norm": 0.09911254793405533,
      "learning_rate": 1.6956031567080047e-05,
      "loss": 0.0455,
      "step": 876
    },
    {
      "epoch": 2.8303715670436187,
      "grad_norm": 0.1377602368593216,
      "learning_rate": 1.695152198421646e-05,
      "loss": 0.0599,
      "step": 877
    },
    {
      "epoch": 2.8336025848142166,
      "grad_norm": 0.11292906850576401,
      "learning_rate": 1.6947012401352876e-05,
      "loss": 0.053,
      "step": 878
    },
    {
      "epoch": 2.8368336025848144,
      "grad_norm": 0.13102898001670837,
      "learning_rate": 1.694250281848929e-05,
      "loss": 0.0629,
      "step": 879
    },
    {
      "epoch": 2.840064620355412,
      "grad_norm": 0.12573762238025665,
      "learning_rate": 1.6937993235625708e-05,
      "loss": 0.052,
      "step": 880
    },
    {
      "epoch": 2.8432956381260097,
      "grad_norm": 0.11562048643827438,
      "learning_rate": 1.6933483652762122e-05,
      "loss": 0.0594,
      "step": 881
    },
    {
      "epoch": 2.8465266558966076,
      "grad_norm": 0.15344659984111786,
      "learning_rate": 1.6928974069898537e-05,
      "loss": 0.0636,
      "step": 882
    },
    {
      "epoch": 2.849757673667205,
      "grad_norm": 0.11969706416130066,
      "learning_rate": 1.692446448703495e-05,
      "loss": 0.0549,
      "step": 883
    },
    {
      "epoch": 2.852988691437803,
      "grad_norm": 0.0930706337094307,
      "learning_rate": 1.6919954904171365e-05,
      "loss": 0.0434,
      "step": 884
    },
    {
      "epoch": 2.8562197092084007,
      "grad_norm": 0.1458914428949356,
      "learning_rate": 1.691544532130778e-05,
      "loss": 0.0707,
      "step": 885
    },
    {
      "epoch": 2.8594507269789986,
      "grad_norm": 0.11928731948137283,
      "learning_rate": 1.6910935738444198e-05,
      "loss": 0.0584,
      "step": 886
    },
    {
      "epoch": 2.862681744749596,
      "grad_norm": 0.1225530207157135,
      "learning_rate": 1.6906426155580612e-05,
      "loss": 0.0619,
      "step": 887
    },
    {
      "epoch": 2.865912762520194,
      "grad_norm": 0.13734394311904907,
      "learning_rate": 1.6901916572717026e-05,
      "loss": 0.065,
      "step": 888
    },
    {
      "epoch": 2.8691437802907918,
      "grad_norm": 0.13331110775470734,
      "learning_rate": 1.689740698985344e-05,
      "loss": 0.0652,
      "step": 889
    },
    {
      "epoch": 2.872374798061389,
      "grad_norm": 0.1149471327662468,
      "learning_rate": 1.6892897406989855e-05,
      "loss": 0.0495,
      "step": 890
    },
    {
      "epoch": 2.875605815831987,
      "grad_norm": 0.14575156569480896,
      "learning_rate": 1.688838782412627e-05,
      "loss": 0.069,
      "step": 891
    },
    {
      "epoch": 2.878836833602585,
      "grad_norm": 0.1205376535654068,
      "learning_rate": 1.6883878241262684e-05,
      "loss": 0.0595,
      "step": 892
    },
    {
      "epoch": 2.8820678513731828,
      "grad_norm": 0.13029593229293823,
      "learning_rate": 1.6879368658399102e-05,
      "loss": 0.0652,
      "step": 893
    },
    {
      "epoch": 2.88529886914378,
      "grad_norm": 0.1242680773139,
      "learning_rate": 1.6874859075535516e-05,
      "loss": 0.0628,
      "step": 894
    },
    {
      "epoch": 2.888529886914378,
      "grad_norm": 0.1066925972700119,
      "learning_rate": 1.687034949267193e-05,
      "loss": 0.0571,
      "step": 895
    },
    {
      "epoch": 2.891760904684976,
      "grad_norm": 0.09622512012720108,
      "learning_rate": 1.6865839909808345e-05,
      "loss": 0.0447,
      "step": 896
    },
    {
      "epoch": 2.8949919224555734,
      "grad_norm": 0.14432470500469208,
      "learning_rate": 1.686133032694476e-05,
      "loss": 0.0658,
      "step": 897
    },
    {
      "epoch": 2.898222940226171,
      "grad_norm": 0.1262982338666916,
      "learning_rate": 1.6856820744081174e-05,
      "loss": 0.057,
      "step": 898
    },
    {
      "epoch": 2.901453957996769,
      "grad_norm": 0.12278001755475998,
      "learning_rate": 1.6852311161217588e-05,
      "loss": 0.0533,
      "step": 899
    },
    {
      "epoch": 2.904684975767367,
      "grad_norm": 0.13526810705661774,
      "learning_rate": 1.6847801578354002e-05,
      "loss": 0.0536,
      "step": 900
    },
    {
      "epoch": 2.904684975767367,
      "eval_loss": 0.07217078655958176,
      "eval_runtime": 188.4791,
      "eval_samples_per_second": 1.045,
      "eval_steps_per_second": 1.045,
      "step": 900
    },
    {
      "epoch": 2.9079159935379644,
      "grad_norm": 0.07958260923624039,
      "learning_rate": 1.6843291995490417e-05,
      "loss": 0.0395,
      "step": 901
    },
    {
      "epoch": 2.9111470113085622,
      "grad_norm": 0.10193248093128204,
      "learning_rate": 1.683878241262683e-05,
      "loss": 0.0495,
      "step": 902
    },
    {
      "epoch": 2.9143780290791597,
      "grad_norm": 0.09985180199146271,
      "learning_rate": 1.683427282976325e-05,
      "loss": 0.0496,
      "step": 903
    },
    {
      "epoch": 2.9176090468497575,
      "grad_norm": 0.15160292387008667,
      "learning_rate": 1.6829763246899663e-05,
      "loss": 0.0668,
      "step": 904
    },
    {
      "epoch": 2.9208400646203554,
      "grad_norm": 0.13049964606761932,
      "learning_rate": 1.6825253664036078e-05,
      "loss": 0.067,
      "step": 905
    },
    {
      "epoch": 2.9240710823909533,
      "grad_norm": 0.13118034601211548,
      "learning_rate": 1.6820744081172492e-05,
      "loss": 0.06,
      "step": 906
    },
    {
      "epoch": 2.927302100161551,
      "grad_norm": 0.11038261651992798,
      "learning_rate": 1.6816234498308907e-05,
      "loss": 0.0542,
      "step": 907
    },
    {
      "epoch": 2.9305331179321485,
      "grad_norm": 0.11064022779464722,
      "learning_rate": 1.681172491544532e-05,
      "loss": 0.0535,
      "step": 908
    },
    {
      "epoch": 2.9337641357027464,
      "grad_norm": 0.10448973625898361,
      "learning_rate": 1.6807215332581735e-05,
      "loss": 0.0444,
      "step": 909
    },
    {
      "epoch": 2.936995153473344,
      "grad_norm": 0.09960347414016724,
      "learning_rate": 1.6802705749718153e-05,
      "loss": 0.0477,
      "step": 910
    },
    {
      "epoch": 2.9402261712439417,
      "grad_norm": 0.10175690799951553,
      "learning_rate": 1.6798196166854568e-05,
      "loss": 0.0521,
      "step": 911
    },
    {
      "epoch": 2.9434571890145396,
      "grad_norm": 0.11306945234537125,
      "learning_rate": 1.6793686583990982e-05,
      "loss": 0.0483,
      "step": 912
    },
    {
      "epoch": 2.9466882067851374,
      "grad_norm": 0.12629052996635437,
      "learning_rate": 1.6789177001127396e-05,
      "loss": 0.0594,
      "step": 913
    },
    {
      "epoch": 2.9499192245557353,
      "grad_norm": 0.16772274672985077,
      "learning_rate": 1.678466741826381e-05,
      "loss": 0.0808,
      "step": 914
    },
    {
      "epoch": 2.9531502423263327,
      "grad_norm": 0.14857324957847595,
      "learning_rate": 1.6780157835400225e-05,
      "loss": 0.0626,
      "step": 915
    },
    {
      "epoch": 2.9563812600969306,
      "grad_norm": 0.12077292054891586,
      "learning_rate": 1.6775648252536643e-05,
      "loss": 0.0547,
      "step": 916
    },
    {
      "epoch": 2.959612277867528,
      "grad_norm": 0.08560369163751602,
      "learning_rate": 1.6771138669673057e-05,
      "loss": 0.0402,
      "step": 917
    },
    {
      "epoch": 2.962843295638126,
      "grad_norm": 0.1328180432319641,
      "learning_rate": 1.676662908680947e-05,
      "loss": 0.0603,
      "step": 918
    },
    {
      "epoch": 2.9660743134087237,
      "grad_norm": 0.13140498101711273,
      "learning_rate": 1.6762119503945886e-05,
      "loss": 0.0541,
      "step": 919
    },
    {
      "epoch": 2.9693053311793216,
      "grad_norm": 0.13546602427959442,
      "learning_rate": 1.67576099210823e-05,
      "loss": 0.0644,
      "step": 920
    },
    {
      "epoch": 2.9725363489499195,
      "grad_norm": 0.13099107146263123,
      "learning_rate": 1.6753100338218715e-05,
      "loss": 0.0544,
      "step": 921
    },
    {
      "epoch": 2.975767366720517,
      "grad_norm": 0.12933450937271118,
      "learning_rate": 1.6748590755355133e-05,
      "loss": 0.0632,
      "step": 922
    },
    {
      "epoch": 2.9789983844911148,
      "grad_norm": 0.12769202888011932,
      "learning_rate": 1.6744081172491547e-05,
      "loss": 0.0591,
      "step": 923
    },
    {
      "epoch": 2.982229402261712,
      "grad_norm": 0.12964068353176117,
      "learning_rate": 1.673957158962796e-05,
      "loss": 0.0602,
      "step": 924
    },
    {
      "epoch": 2.98546042003231,
      "grad_norm": 0.1714252084493637,
      "learning_rate": 1.6735062006764376e-05,
      "loss": 0.076,
      "step": 925
    },
    {
      "epoch": 2.988691437802908,
      "grad_norm": 0.15382935106754303,
      "learning_rate": 1.673055242390079e-05,
      "loss": 0.0624,
      "step": 926
    },
    {
      "epoch": 2.991922455573506,
      "grad_norm": 0.15337888896465302,
      "learning_rate": 1.6726042841037205e-05,
      "loss": 0.0652,
      "step": 927
    },
    {
      "epoch": 2.9951534733441036,
      "grad_norm": 0.1587558090686798,
      "learning_rate": 1.6721533258173622e-05,
      "loss": 0.0754,
      "step": 928
    },
    {
      "epoch": 2.998384491114701,
      "grad_norm": 0.09836894273757935,
      "learning_rate": 1.6717023675310037e-05,
      "loss": 0.046,
      "step": 929
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.20606525242328644,
      "learning_rate": 1.671251409244645e-05,
      "loss": 0.0692,
      "step": 930
    },
    {
      "epoch": 3.003231017770598,
      "grad_norm": 0.12933456897735596,
      "learning_rate": 1.6708004509582866e-05,
      "loss": 0.0601,
      "step": 931
    },
    {
      "epoch": 3.0064620355411953,
      "grad_norm": 0.15069305896759033,
      "learning_rate": 1.670349492671928e-05,
      "loss": 0.0691,
      "step": 932
    },
    {
      "epoch": 3.009693053311793,
      "grad_norm": 0.12518665194511414,
      "learning_rate": 1.6698985343855694e-05,
      "loss": 0.0594,
      "step": 933
    },
    {
      "epoch": 3.012924071082391,
      "grad_norm": 0.13509726524353027,
      "learning_rate": 1.6694475760992112e-05,
      "loss": 0.0553,
      "step": 934
    },
    {
      "epoch": 3.016155088852989,
      "grad_norm": 0.18207424879074097,
      "learning_rate": 1.6689966178128526e-05,
      "loss": 0.0691,
      "step": 935
    },
    {
      "epoch": 3.0193861066235863,
      "grad_norm": 0.10155311226844788,
      "learning_rate": 1.668545659526494e-05,
      "loss": 0.0442,
      "step": 936
    },
    {
      "epoch": 3.022617124394184,
      "grad_norm": 0.08462440967559814,
      "learning_rate": 1.6680947012401355e-05,
      "loss": 0.0397,
      "step": 937
    },
    {
      "epoch": 3.025848142164782,
      "grad_norm": 0.1074729785323143,
      "learning_rate": 1.667643742953777e-05,
      "loss": 0.0476,
      "step": 938
    },
    {
      "epoch": 3.0290791599353795,
      "grad_norm": 0.10128747671842575,
      "learning_rate": 1.6671927846674184e-05,
      "loss": 0.044,
      "step": 939
    },
    {
      "epoch": 3.0323101777059773,
      "grad_norm": 0.10703253746032715,
      "learning_rate": 1.66674182638106e-05,
      "loss": 0.0434,
      "step": 940
    },
    {
      "epoch": 3.035541195476575,
      "grad_norm": 0.16827581822872162,
      "learning_rate": 1.6662908680947013e-05,
      "loss": 0.058,
      "step": 941
    },
    {
      "epoch": 3.038772213247173,
      "grad_norm": 0.12423544377088547,
      "learning_rate": 1.6658399098083427e-05,
      "loss": 0.0529,
      "step": 942
    },
    {
      "epoch": 3.0420032310177705,
      "grad_norm": 0.11421461403369904,
      "learning_rate": 1.665388951521984e-05,
      "loss": 0.0462,
      "step": 943
    },
    {
      "epoch": 3.0452342487883683,
      "grad_norm": 0.1504746377468109,
      "learning_rate": 1.664937993235626e-05,
      "loss": 0.0607,
      "step": 944
    },
    {
      "epoch": 3.048465266558966,
      "grad_norm": 0.1171237900853157,
      "learning_rate": 1.6644870349492674e-05,
      "loss": 0.0488,
      "step": 945
    },
    {
      "epoch": 3.0516962843295636,
      "grad_norm": 0.12751275300979614,
      "learning_rate": 1.6640360766629088e-05,
      "loss": 0.0566,
      "step": 946
    },
    {
      "epoch": 3.0549273021001615,
      "grad_norm": 0.10137461870908737,
      "learning_rate": 1.6635851183765503e-05,
      "loss": 0.042,
      "step": 947
    },
    {
      "epoch": 3.0581583198707594,
      "grad_norm": 0.10805993527173996,
      "learning_rate": 1.6631341600901917e-05,
      "loss": 0.0436,
      "step": 948
    },
    {
      "epoch": 3.0613893376413572,
      "grad_norm": 0.15429779887199402,
      "learning_rate": 1.662683201803833e-05,
      "loss": 0.0611,
      "step": 949
    },
    {
      "epoch": 3.0646203554119547,
      "grad_norm": 0.15192106366157532,
      "learning_rate": 1.6622322435174746e-05,
      "loss": 0.0558,
      "step": 950
    },
    {
      "epoch": 3.0678513731825525,
      "grad_norm": 0.14291639626026154,
      "learning_rate": 1.661781285231116e-05,
      "loss": 0.0582,
      "step": 951
    },
    {
      "epoch": 3.0710823909531504,
      "grad_norm": 0.11516083776950836,
      "learning_rate": 1.6613303269447578e-05,
      "loss": 0.0455,
      "step": 952
    },
    {
      "epoch": 3.074313408723748,
      "grad_norm": 0.11716248095035553,
      "learning_rate": 1.6608793686583992e-05,
      "loss": 0.0419,
      "step": 953
    },
    {
      "epoch": 3.0775444264943457,
      "grad_norm": 0.13777975738048553,
      "learning_rate": 1.6604284103720407e-05,
      "loss": 0.0587,
      "step": 954
    },
    {
      "epoch": 3.0807754442649435,
      "grad_norm": 0.15481697022914886,
      "learning_rate": 1.659977452085682e-05,
      "loss": 0.058,
      "step": 955
    },
    {
      "epoch": 3.0840064620355414,
      "grad_norm": 0.11290151625871658,
      "learning_rate": 1.6595264937993235e-05,
      "loss": 0.0428,
      "step": 956
    },
    {
      "epoch": 3.087237479806139,
      "grad_norm": 0.11138515174388885,
      "learning_rate": 1.6590755355129653e-05,
      "loss": 0.0445,
      "step": 957
    },
    {
      "epoch": 3.0904684975767367,
      "grad_norm": 0.13892598450183868,
      "learning_rate": 1.6586245772266068e-05,
      "loss": 0.0532,
      "step": 958
    },
    {
      "epoch": 3.0936995153473346,
      "grad_norm": 0.14099125564098358,
      "learning_rate": 1.6581736189402482e-05,
      "loss": 0.0532,
      "step": 959
    },
    {
      "epoch": 3.096930533117932,
      "grad_norm": 0.1620667278766632,
      "learning_rate": 1.6577226606538896e-05,
      "loss": 0.0716,
      "step": 960
    },
    {
      "epoch": 3.10016155088853,
      "grad_norm": 0.1435079723596573,
      "learning_rate": 1.657271702367531e-05,
      "loss": 0.0587,
      "step": 961
    },
    {
      "epoch": 3.1033925686591277,
      "grad_norm": 0.1412099003791809,
      "learning_rate": 1.6568207440811725e-05,
      "loss": 0.0599,
      "step": 962
    },
    {
      "epoch": 3.106623586429725,
      "grad_norm": 0.16996391117572784,
      "learning_rate": 1.6563697857948143e-05,
      "loss": 0.0577,
      "step": 963
    },
    {
      "epoch": 3.109854604200323,
      "grad_norm": 0.14544463157653809,
      "learning_rate": 1.6559188275084557e-05,
      "loss": 0.0595,
      "step": 964
    },
    {
      "epoch": 3.113085621970921,
      "grad_norm": 0.12646666169166565,
      "learning_rate": 1.6554678692220972e-05,
      "loss": 0.0496,
      "step": 965
    },
    {
      "epoch": 3.1163166397415187,
      "grad_norm": 0.16260091960430145,
      "learning_rate": 1.6550169109357386e-05,
      "loss": 0.0588,
      "step": 966
    },
    {
      "epoch": 3.119547657512116,
      "grad_norm": 0.14531579613685608,
      "learning_rate": 1.65456595264938e-05,
      "loss": 0.0654,
      "step": 967
    },
    {
      "epoch": 3.122778675282714,
      "grad_norm": 0.13838277757167816,
      "learning_rate": 1.6541149943630215e-05,
      "loss": 0.058,
      "step": 968
    },
    {
      "epoch": 3.126009693053312,
      "grad_norm": 0.10179346054792404,
      "learning_rate": 1.6536640360766633e-05,
      "loss": 0.0394,
      "step": 969
    },
    {
      "epoch": 3.1292407108239093,
      "grad_norm": 0.14759835600852966,
      "learning_rate": 1.6532130777903047e-05,
      "loss": 0.0616,
      "step": 970
    },
    {
      "epoch": 3.132471728594507,
      "grad_norm": 0.12317320704460144,
      "learning_rate": 1.652762119503946e-05,
      "loss": 0.0457,
      "step": 971
    },
    {
      "epoch": 3.135702746365105,
      "grad_norm": 0.14770029485225677,
      "learning_rate": 1.6523111612175876e-05,
      "loss": 0.0606,
      "step": 972
    },
    {
      "epoch": 3.138933764135703,
      "grad_norm": 0.14644749462604523,
      "learning_rate": 1.651860202931229e-05,
      "loss": 0.0576,
      "step": 973
    },
    {
      "epoch": 3.1421647819063003,
      "grad_norm": 0.15745016932487488,
      "learning_rate": 1.6514092446448705e-05,
      "loss": 0.0658,
      "step": 974
    },
    {
      "epoch": 3.145395799676898,
      "grad_norm": 0.15281431376934052,
      "learning_rate": 1.6509582863585122e-05,
      "loss": 0.0579,
      "step": 975
    },
    {
      "epoch": 3.145395799676898,
      "eval_loss": 0.07240297645330429,
      "eval_runtime": 188.3995,
      "eval_samples_per_second": 1.046,
      "eval_steps_per_second": 1.046,
      "step": 975
    },
    {
      "epoch": 3.148626817447496,
      "grad_norm": 0.186857670545578,
      "learning_rate": 1.6505073280721537e-05,
      "loss": 0.0712,
      "step": 976
    },
    {
      "epoch": 3.1518578352180935,
      "grad_norm": 0.11668923497200012,
      "learning_rate": 1.650056369785795e-05,
      "loss": 0.0432,
      "step": 977
    },
    {
      "epoch": 3.1550888529886914,
      "grad_norm": 0.1078757792711258,
      "learning_rate": 1.6496054114994366e-05,
      "loss": 0.0438,
      "step": 978
    },
    {
      "epoch": 3.158319870759289,
      "grad_norm": 0.10889827460050583,
      "learning_rate": 1.649154453213078e-05,
      "loss": 0.0404,
      "step": 979
    },
    {
      "epoch": 3.161550888529887,
      "grad_norm": 0.11770477145910263,
      "learning_rate": 1.6487034949267194e-05,
      "loss": 0.0379,
      "step": 980
    },
    {
      "epoch": 3.1647819063004845,
      "grad_norm": 0.1730085015296936,
      "learning_rate": 1.648252536640361e-05,
      "loss": 0.0527,
      "step": 981
    },
    {
      "epoch": 3.1680129240710824,
      "grad_norm": 0.17005400359630585,
      "learning_rate": 1.6478015783540023e-05,
      "loss": 0.0608,
      "step": 982
    },
    {
      "epoch": 3.1712439418416802,
      "grad_norm": 0.15987516939640045,
      "learning_rate": 1.6473506200676438e-05,
      "loss": 0.0657,
      "step": 983
    },
    {
      "epoch": 3.1744749596122777,
      "grad_norm": 0.12467172741889954,
      "learning_rate": 1.6468996617812852e-05,
      "loss": 0.0532,
      "step": 984
    },
    {
      "epoch": 3.1777059773828755,
      "grad_norm": 0.13436008989810944,
      "learning_rate": 1.646448703494927e-05,
      "loss": 0.0464,
      "step": 985
    },
    {
      "epoch": 3.1809369951534734,
      "grad_norm": 0.15260566771030426,
      "learning_rate": 1.6459977452085684e-05,
      "loss": 0.0585,
      "step": 986
    },
    {
      "epoch": 3.1841680129240713,
      "grad_norm": 0.1228412613272667,
      "learning_rate": 1.64554678692221e-05,
      "loss": 0.042,
      "step": 987
    },
    {
      "epoch": 3.1873990306946687,
      "grad_norm": 0.1621600091457367,
      "learning_rate": 1.6450958286358513e-05,
      "loss": 0.0705,
      "step": 988
    },
    {
      "epoch": 3.1906300484652665,
      "grad_norm": 0.14798057079315186,
      "learning_rate": 1.6446448703494927e-05,
      "loss": 0.0536,
      "step": 989
    },
    {
      "epoch": 3.1938610662358644,
      "grad_norm": 0.17002591490745544,
      "learning_rate": 1.644193912063134e-05,
      "loss": 0.0624,
      "step": 990
    },
    {
      "epoch": 3.197092084006462,
      "grad_norm": 0.11882289499044418,
      "learning_rate": 1.6437429537767756e-05,
      "loss": 0.0386,
      "step": 991
    },
    {
      "epoch": 3.2003231017770597,
      "grad_norm": 0.177546426653862,
      "learning_rate": 1.643291995490417e-05,
      "loss": 0.0684,
      "step": 992
    },
    {
      "epoch": 3.2035541195476576,
      "grad_norm": 0.1515907645225525,
      "learning_rate": 1.6428410372040588e-05,
      "loss": 0.0588,
      "step": 993
    },
    {
      "epoch": 3.2067851373182554,
      "grad_norm": 0.13172346353530884,
      "learning_rate": 1.6423900789177003e-05,
      "loss": 0.0502,
      "step": 994
    },
    {
      "epoch": 3.210016155088853,
      "grad_norm": 0.1430046707391739,
      "learning_rate": 1.6419391206313417e-05,
      "loss": 0.0538,
      "step": 995
    },
    {
      "epoch": 3.2132471728594507,
      "grad_norm": 0.10192380100488663,
      "learning_rate": 1.641488162344983e-05,
      "loss": 0.0414,
      "step": 996
    },
    {
      "epoch": 3.2164781906300486,
      "grad_norm": 0.12296223640441895,
      "learning_rate": 1.6410372040586246e-05,
      "loss": 0.0466,
      "step": 997
    },
    {
      "epoch": 3.219709208400646,
      "grad_norm": 0.1641893982887268,
      "learning_rate": 1.640586245772266e-05,
      "loss": 0.0606,
      "step": 998
    },
    {
      "epoch": 3.222940226171244,
      "grad_norm": 0.109470434486866,
      "learning_rate": 1.6401352874859078e-05,
      "loss": 0.0424,
      "step": 999
    },
    {
      "epoch": 3.2261712439418417,
      "grad_norm": 0.10068835318088531,
      "learning_rate": 1.6396843291995492e-05,
      "loss": 0.0364,
      "step": 1000
    },
    {
      "epoch": 3.2294022617124396,
      "grad_norm": 0.09672326594591141,
      "learning_rate": 1.6392333709131907e-05,
      "loss": 0.0403,
      "step": 1001
    },
    {
      "epoch": 3.232633279483037,
      "grad_norm": 0.21638123691082,
      "learning_rate": 1.638782412626832e-05,
      "loss": 0.0785,
      "step": 1002
    },
    {
      "epoch": 3.235864297253635,
      "grad_norm": 0.15015633404254913,
      "learning_rate": 1.6383314543404735e-05,
      "loss": 0.0513,
      "step": 1003
    },
    {
      "epoch": 3.2390953150242328,
      "grad_norm": 0.11658553779125214,
      "learning_rate": 1.637880496054115e-05,
      "loss": 0.0444,
      "step": 1004
    },
    {
      "epoch": 3.24232633279483,
      "grad_norm": 0.142287939786911,
      "learning_rate": 1.6374295377677568e-05,
      "loss": 0.0576,
      "step": 1005
    },
    {
      "epoch": 3.245557350565428,
      "grad_norm": 0.11885146051645279,
      "learning_rate": 1.6369785794813982e-05,
      "loss": 0.0442,
      "step": 1006
    },
    {
      "epoch": 3.248788368336026,
      "grad_norm": 0.1423695832490921,
      "learning_rate": 1.6365276211950396e-05,
      "loss": 0.0539,
      "step": 1007
    },
    {
      "epoch": 3.2520193861066238,
      "grad_norm": 0.14337588846683502,
      "learning_rate": 1.636076662908681e-05,
      "loss": 0.056,
      "step": 1008
    },
    {
      "epoch": 3.255250403877221,
      "grad_norm": 0.15875791013240814,
      "learning_rate": 1.6356257046223225e-05,
      "loss": 0.0672,
      "step": 1009
    },
    {
      "epoch": 3.258481421647819,
      "grad_norm": 0.1158808171749115,
      "learning_rate": 1.635174746335964e-05,
      "loss": 0.0464,
      "step": 1010
    },
    {
      "epoch": 3.261712439418417,
      "grad_norm": 0.12882259488105774,
      "learning_rate": 1.6347237880496057e-05,
      "loss": 0.0497,
      "step": 1011
    },
    {
      "epoch": 3.2649434571890144,
      "grad_norm": 0.15846951305866241,
      "learning_rate": 1.6342728297632472e-05,
      "loss": 0.0656,
      "step": 1012
    },
    {
      "epoch": 3.268174474959612,
      "grad_norm": 0.16542094945907593,
      "learning_rate": 1.6338218714768886e-05,
      "loss": 0.06,
      "step": 1013
    },
    {
      "epoch": 3.27140549273021,
      "grad_norm": 0.1318148821592331,
      "learning_rate": 1.63337091319053e-05,
      "loss": 0.0475,
      "step": 1014
    },
    {
      "epoch": 3.274636510500808,
      "grad_norm": 0.110074482858181,
      "learning_rate": 1.6329199549041715e-05,
      "loss": 0.0411,
      "step": 1015
    },
    {
      "epoch": 3.2778675282714054,
      "grad_norm": 0.14870020747184753,
      "learning_rate": 1.632468996617813e-05,
      "loss": 0.0491,
      "step": 1016
    },
    {
      "epoch": 3.2810985460420032,
      "grad_norm": 0.16925957798957825,
      "learning_rate": 1.6320180383314547e-05,
      "loss": 0.0648,
      "step": 1017
    },
    {
      "epoch": 3.284329563812601,
      "grad_norm": 0.11573273688554764,
      "learning_rate": 1.631567080045096e-05,
      "loss": 0.0457,
      "step": 1018
    },
    {
      "epoch": 3.2875605815831985,
      "grad_norm": 0.16456320881843567,
      "learning_rate": 1.6311161217587376e-05,
      "loss": 0.0608,
      "step": 1019
    },
    {
      "epoch": 3.2907915993537964,
      "grad_norm": 0.14626148343086243,
      "learning_rate": 1.630665163472379e-05,
      "loss": 0.0543,
      "step": 1020
    },
    {
      "epoch": 3.2940226171243943,
      "grad_norm": 0.15516629815101624,
      "learning_rate": 1.6302142051860205e-05,
      "loss": 0.0628,
      "step": 1021
    },
    {
      "epoch": 3.297253634894992,
      "grad_norm": 0.18173731863498688,
      "learning_rate": 1.629763246899662e-05,
      "loss": 0.0677,
      "step": 1022
    },
    {
      "epoch": 3.3004846526655895,
      "grad_norm": 0.146285742521286,
      "learning_rate": 1.6293122886133033e-05,
      "loss": 0.052,
      "step": 1023
    },
    {
      "epoch": 3.3037156704361874,
      "grad_norm": 0.19001832604408264,
      "learning_rate": 1.6288613303269448e-05,
      "loss": 0.0693,
      "step": 1024
    },
    {
      "epoch": 3.3069466882067853,
      "grad_norm": 0.09959913045167923,
      "learning_rate": 1.6284103720405862e-05,
      "loss": 0.0339,
      "step": 1025
    },
    {
      "epoch": 3.3101777059773827,
      "grad_norm": 0.16015265882015228,
      "learning_rate": 1.627959413754228e-05,
      "loss": 0.0615,
      "step": 1026
    },
    {
      "epoch": 3.3134087237479806,
      "grad_norm": 0.13552674651145935,
      "learning_rate": 1.6275084554678694e-05,
      "loss": 0.0475,
      "step": 1027
    },
    {
      "epoch": 3.3166397415185784,
      "grad_norm": 0.16153255105018616,
      "learning_rate": 1.627057497181511e-05,
      "loss": 0.0601,
      "step": 1028
    },
    {
      "epoch": 3.3198707592891763,
      "grad_norm": 0.14683452248573303,
      "learning_rate": 1.6266065388951523e-05,
      "loss": 0.0526,
      "step": 1029
    },
    {
      "epoch": 3.3231017770597737,
      "grad_norm": 0.11259462684392929,
      "learning_rate": 1.6261555806087938e-05,
      "loss": 0.0429,
      "step": 1030
    },
    {
      "epoch": 3.3263327948303716,
      "grad_norm": 0.1688949465751648,
      "learning_rate": 1.6257046223224352e-05,
      "loss": 0.0624,
      "step": 1031
    },
    {
      "epoch": 3.3295638126009695,
      "grad_norm": 0.10584679245948792,
      "learning_rate": 1.6252536640360766e-05,
      "loss": 0.0376,
      "step": 1032
    },
    {
      "epoch": 3.332794830371567,
      "grad_norm": 0.16506820917129517,
      "learning_rate": 1.624802705749718e-05,
      "loss": 0.0585,
      "step": 1033
    },
    {
      "epoch": 3.3360258481421647,
      "grad_norm": 0.11264611780643463,
      "learning_rate": 1.6243517474633595e-05,
      "loss": 0.0432,
      "step": 1034
    },
    {
      "epoch": 3.3392568659127626,
      "grad_norm": 0.17402660846710205,
      "learning_rate": 1.6239007891770013e-05,
      "loss": 0.0566,
      "step": 1035
    },
    {
      "epoch": 3.3424878836833605,
      "grad_norm": 0.15407973527908325,
      "learning_rate": 1.6234498308906427e-05,
      "loss": 0.0519,
      "step": 1036
    },
    {
      "epoch": 3.345718901453958,
      "grad_norm": 0.1423128843307495,
      "learning_rate": 1.622998872604284e-05,
      "loss": 0.0511,
      "step": 1037
    },
    {
      "epoch": 3.3489499192245558,
      "grad_norm": 0.11291830986738205,
      "learning_rate": 1.6225479143179256e-05,
      "loss": 0.0413,
      "step": 1038
    },
    {
      "epoch": 3.3521809369951536,
      "grad_norm": 0.17067734897136688,
      "learning_rate": 1.622096956031567e-05,
      "loss": 0.0587,
      "step": 1039
    },
    {
      "epoch": 3.355411954765751,
      "grad_norm": 0.17072725296020508,
      "learning_rate": 1.6216459977452088e-05,
      "loss": 0.0606,
      "step": 1040
    },
    {
      "epoch": 3.358642972536349,
      "grad_norm": 0.13390378654003143,
      "learning_rate": 1.6211950394588503e-05,
      "loss": 0.0411,
      "step": 1041
    },
    {
      "epoch": 3.361873990306947,
      "grad_norm": 0.10424879193305969,
      "learning_rate": 1.6207440811724917e-05,
      "loss": 0.0356,
      "step": 1042
    },
    {
      "epoch": 3.3651050080775446,
      "grad_norm": 0.26167649030685425,
      "learning_rate": 1.620293122886133e-05,
      "loss": 0.0569,
      "step": 1043
    },
    {
      "epoch": 3.368336025848142,
      "grad_norm": 0.1589985489845276,
      "learning_rate": 1.6198421645997746e-05,
      "loss": 0.0539,
      "step": 1044
    },
    {
      "epoch": 3.37156704361874,
      "grad_norm": 0.14946100115776062,
      "learning_rate": 1.619391206313416e-05,
      "loss": 0.0531,
      "step": 1045
    },
    {
      "epoch": 3.374798061389338,
      "grad_norm": 0.1740565448999405,
      "learning_rate": 1.6189402480270578e-05,
      "loss": 0.0626,
      "step": 1046
    },
    {
      "epoch": 3.378029079159935,
      "grad_norm": 0.16527672111988068,
      "learning_rate": 1.6184892897406992e-05,
      "loss": 0.061,
      "step": 1047
    },
    {
      "epoch": 3.381260096930533,
      "grad_norm": 0.11640432476997375,
      "learning_rate": 1.6180383314543407e-05,
      "loss": 0.0433,
      "step": 1048
    },
    {
      "epoch": 3.384491114701131,
      "grad_norm": 0.11857109516859055,
      "learning_rate": 1.617587373167982e-05,
      "loss": 0.0374,
      "step": 1049
    },
    {
      "epoch": 3.387722132471729,
      "grad_norm": 0.15555702149868011,
      "learning_rate": 1.6171364148816236e-05,
      "loss": 0.0505,
      "step": 1050
    },
    {
      "epoch": 3.387722132471729,
      "eval_loss": 0.07340462505817413,
      "eval_runtime": 188.3173,
      "eval_samples_per_second": 1.046,
      "eval_steps_per_second": 1.046,
      "step": 1050
    },
    {
      "epoch": 3.3909531502423262,
      "grad_norm": 0.1614767611026764,
      "learning_rate": 1.616685456595265e-05,
      "loss": 0.0616,
      "step": 1051
    },
    {
      "epoch": 3.394184168012924,
      "grad_norm": 0.13632138073444366,
      "learning_rate": 1.6162344983089068e-05,
      "loss": 0.0495,
      "step": 1052
    },
    {
      "epoch": 3.397415185783522,
      "grad_norm": 0.1372632533311844,
      "learning_rate": 1.6157835400225482e-05,
      "loss": 0.0433,
      "step": 1053
    },
    {
      "epoch": 3.4006462035541194,
      "grad_norm": 0.17442293465137482,
      "learning_rate": 1.6153325817361896e-05,
      "loss": 0.0578,
      "step": 1054
    },
    {
      "epoch": 3.4038772213247173,
      "grad_norm": 0.18292829394340515,
      "learning_rate": 1.614881623449831e-05,
      "loss": 0.0648,
      "step": 1055
    },
    {
      "epoch": 3.407108239095315,
      "grad_norm": 0.15629050135612488,
      "learning_rate": 1.6144306651634725e-05,
      "loss": 0.0504,
      "step": 1056
    },
    {
      "epoch": 3.410339256865913,
      "grad_norm": 0.1325637549161911,
      "learning_rate": 1.613979706877114e-05,
      "loss": 0.039,
      "step": 1057
    },
    {
      "epoch": 3.4135702746365104,
      "grad_norm": 0.17488038539886475,
      "learning_rate": 1.6135287485907557e-05,
      "loss": 0.0557,
      "step": 1058
    },
    {
      "epoch": 3.4168012924071083,
      "grad_norm": 0.1492920219898224,
      "learning_rate": 1.6130777903043972e-05,
      "loss": 0.0537,
      "step": 1059
    },
    {
      "epoch": 3.420032310177706,
      "grad_norm": 0.14212766289710999,
      "learning_rate": 1.6126268320180386e-05,
      "loss": 0.0422,
      "step": 1060
    },
    {
      "epoch": 3.4232633279483036,
      "grad_norm": 0.1526009440422058,
      "learning_rate": 1.61217587373168e-05,
      "loss": 0.0572,
      "step": 1061
    },
    {
      "epoch": 3.4264943457189014,
      "grad_norm": 0.11353015899658203,
      "learning_rate": 1.6117249154453215e-05,
      "loss": 0.0383,
      "step": 1062
    },
    {
      "epoch": 3.4297253634894993,
      "grad_norm": 0.18253804743289948,
      "learning_rate": 1.611273957158963e-05,
      "loss": 0.0657,
      "step": 1063
    },
    {
      "epoch": 3.432956381260097,
      "grad_norm": 0.1422210931777954,
      "learning_rate": 1.6108229988726044e-05,
      "loss": 0.0528,
      "step": 1064
    },
    {
      "epoch": 3.4361873990306946,
      "grad_norm": 0.15058425068855286,
      "learning_rate": 1.6103720405862458e-05,
      "loss": 0.052,
      "step": 1065
    },
    {
      "epoch": 3.4394184168012925,
      "grad_norm": 0.13702097535133362,
      "learning_rate": 1.6099210822998873e-05,
      "loss": 0.0478,
      "step": 1066
    },
    {
      "epoch": 3.4426494345718903,
      "grad_norm": 0.19859325885772705,
      "learning_rate": 1.609470124013529e-05,
      "loss": 0.0512,
      "step": 1067
    },
    {
      "epoch": 3.4458804523424877,
      "grad_norm": 0.1942027509212494,
      "learning_rate": 1.6090191657271705e-05,
      "loss": 0.0711,
      "step": 1068
    },
    {
      "epoch": 3.4491114701130856,
      "grad_norm": 0.1615460216999054,
      "learning_rate": 1.608568207440812e-05,
      "loss": 0.0532,
      "step": 1069
    },
    {
      "epoch": 3.4523424878836835,
      "grad_norm": 0.15197354555130005,
      "learning_rate": 1.6081172491544533e-05,
      "loss": 0.0566,
      "step": 1070
    },
    {
      "epoch": 3.4555735056542813,
      "grad_norm": 0.131342813372612,
      "learning_rate": 1.6076662908680948e-05,
      "loss": 0.0448,
      "step": 1071
    },
    {
      "epoch": 3.4588045234248788,
      "grad_norm": 0.1431513875722885,
      "learning_rate": 1.6072153325817362e-05,
      "loss": 0.0502,
      "step": 1072
    },
    {
      "epoch": 3.4620355411954766,
      "grad_norm": 0.1926133781671524,
      "learning_rate": 1.6067643742953777e-05,
      "loss": 0.0625,
      "step": 1073
    },
    {
      "epoch": 3.4652665589660745,
      "grad_norm": 0.13775743544101715,
      "learning_rate": 1.606313416009019e-05,
      "loss": 0.0529,
      "step": 1074
    },
    {
      "epoch": 3.468497576736672,
      "grad_norm": 0.1370486617088318,
      "learning_rate": 1.6058624577226605e-05,
      "loss": 0.0489,
      "step": 1075
    },
    {
      "epoch": 3.47172859450727,
      "grad_norm": 0.18667002022266388,
      "learning_rate": 1.6054114994363023e-05,
      "loss": 0.0734,
      "step": 1076
    },
    {
      "epoch": 3.4749596122778676,
      "grad_norm": 0.16826723515987396,
      "learning_rate": 1.6049605411499438e-05,
      "loss": 0.0568,
      "step": 1077
    },
    {
      "epoch": 3.4781906300484655,
      "grad_norm": 0.1706121861934662,
      "learning_rate": 1.6045095828635852e-05,
      "loss": 0.064,
      "step": 1078
    },
    {
      "epoch": 3.481421647819063,
      "grad_norm": 0.12642326951026917,
      "learning_rate": 1.6040586245772266e-05,
      "loss": 0.0441,
      "step": 1079
    },
    {
      "epoch": 3.484652665589661,
      "grad_norm": 0.14685548841953278,
      "learning_rate": 1.603607666290868e-05,
      "loss": 0.0484,
      "step": 1080
    },
    {
      "epoch": 3.4878836833602587,
      "grad_norm": 0.13969610631465912,
      "learning_rate": 1.6031567080045095e-05,
      "loss": 0.0467,
      "step": 1081
    },
    {
      "epoch": 3.491114701130856,
      "grad_norm": 0.18751631677150726,
      "learning_rate": 1.6027057497181513e-05,
      "loss": 0.0653,
      "step": 1082
    },
    {
      "epoch": 3.494345718901454,
      "grad_norm": 0.14187730848789215,
      "learning_rate": 1.6022547914317927e-05,
      "loss": 0.0485,
      "step": 1083
    },
    {
      "epoch": 3.497576736672052,
      "grad_norm": 0.13812421262264252,
      "learning_rate": 1.6018038331454342e-05,
      "loss": 0.0494,
      "step": 1084
    },
    {
      "epoch": 3.5008077544426497,
      "grad_norm": 0.13007357716560364,
      "learning_rate": 1.6013528748590756e-05,
      "loss": 0.0454,
      "step": 1085
    },
    {
      "epoch": 3.504038772213247,
      "grad_norm": 0.16555847227573395,
      "learning_rate": 1.600901916572717e-05,
      "loss": 0.0474,
      "step": 1086
    },
    {
      "epoch": 3.507269789983845,
      "grad_norm": 0.11033131182193756,
      "learning_rate": 1.6004509582863585e-05,
      "loss": 0.0367,
      "step": 1087
    },
    {
      "epoch": 3.5105008077544424,
      "grad_norm": 0.14931395649909973,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.0489,
      "step": 1088
    },
    {
      "epoch": 3.5137318255250403,
      "grad_norm": 0.16848234832286835,
      "learning_rate": 1.5995490417136417e-05,
      "loss": 0.0578,
      "step": 1089
    },
    {
      "epoch": 3.516962843295638,
      "grad_norm": 0.15877306461334229,
      "learning_rate": 1.599098083427283e-05,
      "loss": 0.0524,
      "step": 1090
    },
    {
      "epoch": 3.520193861066236,
      "grad_norm": 0.16530410945415497,
      "learning_rate": 1.5986471251409246e-05,
      "loss": 0.0619,
      "step": 1091
    },
    {
      "epoch": 3.523424878836834,
      "grad_norm": 0.14331963658332825,
      "learning_rate": 1.598196166854566e-05,
      "loss": 0.0486,
      "step": 1092
    },
    {
      "epoch": 3.5266558966074313,
      "grad_norm": 0.15027795732021332,
      "learning_rate": 1.5977452085682075e-05,
      "loss": 0.048,
      "step": 1093
    },
    {
      "epoch": 3.529886914378029,
      "grad_norm": 0.15376311540603638,
      "learning_rate": 1.5972942502818492e-05,
      "loss": 0.0559,
      "step": 1094
    },
    {
      "epoch": 3.5331179321486266,
      "grad_norm": 0.1315869837999344,
      "learning_rate": 1.5968432919954907e-05,
      "loss": 0.0451,
      "step": 1095
    },
    {
      "epoch": 3.5363489499192244,
      "grad_norm": 0.13606947660446167,
      "learning_rate": 1.596392333709132e-05,
      "loss": 0.0433,
      "step": 1096
    },
    {
      "epoch": 3.5395799676898223,
      "grad_norm": 0.20028483867645264,
      "learning_rate": 1.5959413754227736e-05,
      "loss": 0.0593,
      "step": 1097
    },
    {
      "epoch": 3.54281098546042,
      "grad_norm": 0.15004722774028778,
      "learning_rate": 1.595490417136415e-05,
      "loss": 0.0501,
      "step": 1098
    },
    {
      "epoch": 3.546042003231018,
      "grad_norm": 0.14318561553955078,
      "learning_rate": 1.5950394588500564e-05,
      "loss": 0.0463,
      "step": 1099
    },
    {
      "epoch": 3.5492730210016155,
      "grad_norm": 0.11907773464918137,
      "learning_rate": 1.5945885005636982e-05,
      "loss": 0.0367,
      "step": 1100
    },
    {
      "epoch": 3.5525040387722133,
      "grad_norm": 0.19116652011871338,
      "learning_rate": 1.5941375422773397e-05,
      "loss": 0.0741,
      "step": 1101
    },
    {
      "epoch": 3.5557350565428107,
      "grad_norm": 0.14904284477233887,
      "learning_rate": 1.593686583990981e-05,
      "loss": 0.0493,
      "step": 1102
    },
    {
      "epoch": 3.5589660743134086,
      "grad_norm": 0.134224995970726,
      "learning_rate": 1.5932356257046225e-05,
      "loss": 0.0519,
      "step": 1103
    },
    {
      "epoch": 3.5621970920840065,
      "grad_norm": 0.16584910452365875,
      "learning_rate": 1.592784667418264e-05,
      "loss": 0.0604,
      "step": 1104
    },
    {
      "epoch": 3.5654281098546043,
      "grad_norm": 0.19957157969474792,
      "learning_rate": 1.5923337091319054e-05,
      "loss": 0.0874,
      "step": 1105
    },
    {
      "epoch": 3.568659127625202,
      "grad_norm": 0.16732187569141388,
      "learning_rate": 1.591882750845547e-05,
      "loss": 0.0521,
      "step": 1106
    },
    {
      "epoch": 3.5718901453957996,
      "grad_norm": 0.10181103646755219,
      "learning_rate": 1.5914317925591883e-05,
      "loss": 0.0356,
      "step": 1107
    },
    {
      "epoch": 3.5751211631663975,
      "grad_norm": 0.1692725121974945,
      "learning_rate": 1.59098083427283e-05,
      "loss": 0.0546,
      "step": 1108
    },
    {
      "epoch": 3.578352180936995,
      "grad_norm": 0.15010833740234375,
      "learning_rate": 1.5905298759864715e-05,
      "loss": 0.0525,
      "step": 1109
    },
    {
      "epoch": 3.581583198707593,
      "grad_norm": 0.14599505066871643,
      "learning_rate": 1.590078917700113e-05,
      "loss": 0.0503,
      "step": 1110
    },
    {
      "epoch": 3.5848142164781907,
      "grad_norm": 0.1455962210893631,
      "learning_rate": 1.5896279594137544e-05,
      "loss": 0.0538,
      "step": 1111
    },
    {
      "epoch": 3.5880452342487885,
      "grad_norm": 0.16955074667930603,
      "learning_rate": 1.5891770011273958e-05,
      "loss": 0.0562,
      "step": 1112
    },
    {
      "epoch": 3.5912762520193864,
      "grad_norm": 0.12441671639680862,
      "learning_rate": 1.5887260428410373e-05,
      "loss": 0.0423,
      "step": 1113
    },
    {
      "epoch": 3.594507269789984,
      "grad_norm": 0.1585661768913269,
      "learning_rate": 1.5882750845546787e-05,
      "loss": 0.0585,
      "step": 1114
    },
    {
      "epoch": 3.5977382875605817,
      "grad_norm": 0.1167236939072609,
      "learning_rate": 1.58782412626832e-05,
      "loss": 0.0396,
      "step": 1115
    },
    {
      "epoch": 3.600969305331179,
      "grad_norm": 0.11918371170759201,
      "learning_rate": 1.5873731679819616e-05,
      "loss": 0.0443,
      "step": 1116
    },
    {
      "epoch": 3.604200323101777,
      "grad_norm": 0.1533348709344864,
      "learning_rate": 1.586922209695603e-05,
      "loss": 0.0554,
      "step": 1117
    },
    {
      "epoch": 3.607431340872375,
      "grad_norm": 0.14366954565048218,
      "learning_rate": 1.5864712514092448e-05,
      "loss": 0.046,
      "step": 1118
    },
    {
      "epoch": 3.6106623586429727,
      "grad_norm": 0.17497192323207855,
      "learning_rate": 1.5860202931228862e-05,
      "loss": 0.0572,
      "step": 1119
    },
    {
      "epoch": 3.61389337641357,
      "grad_norm": 0.13877834379673004,
      "learning_rate": 1.5855693348365277e-05,
      "loss": 0.0493,
      "step": 1120
    },
    {
      "epoch": 3.617124394184168,
      "grad_norm": 0.19728821516036987,
      "learning_rate": 1.585118376550169e-05,
      "loss": 0.0712,
      "step": 1121
    },
    {
      "epoch": 3.620355411954766,
      "grad_norm": 0.11815082281827927,
      "learning_rate": 1.5846674182638105e-05,
      "loss": 0.0431,
      "step": 1122
    },
    {
      "epoch": 3.6235864297253633,
      "grad_norm": 0.1730748862028122,
      "learning_rate": 1.5842164599774523e-05,
      "loss": 0.0604,
      "step": 1123
    },
    {
      "epoch": 3.626817447495961,
      "grad_norm": 0.10923890024423599,
      "learning_rate": 1.5837655016910938e-05,
      "loss": 0.0355,
      "step": 1124
    },
    {
      "epoch": 3.630048465266559,
      "grad_norm": 0.12295415252447128,
      "learning_rate": 1.5833145434047352e-05,
      "loss": 0.0412,
      "step": 1125
    },
    {
      "epoch": 3.630048465266559,
      "eval_loss": 0.0719488188624382,
      "eval_runtime": 188.3414,
      "eval_samples_per_second": 1.046,
      "eval_steps_per_second": 1.046,
      "step": 1125
    }
  ],
  "logging_steps": 1,
  "max_steps": 4635,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 75,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.3264284514179097e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}