{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.080321285140563,
  "eval_steps": 1000,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.10040160642570281,
      "grad_norm": 4.897975444793701,
      "learning_rate": 5.000000000000001e-07,
      "loss": 1.0304,
      "step": 25
    },
    {
      "epoch": 0.20080321285140562,
      "grad_norm": 4.204061031341553,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.8001,
      "step": 50
    },
    {
      "epoch": 0.30120481927710846,
      "grad_norm": 3.303119659423828,
      "learning_rate": 1.5e-06,
      "loss": 0.5902,
      "step": 75
    },
    {
      "epoch": 0.40160642570281124,
      "grad_norm": 3.5852315425872803,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.5539,
      "step": 100
    },
    {
      "epoch": 0.5020080321285141,
      "grad_norm": 3.452212333679199,
      "learning_rate": 2.5e-06,
      "loss": 0.5258,
      "step": 125
    },
    {
      "epoch": 0.6024096385542169,
      "grad_norm": 3.3276026248931885,
      "learning_rate": 3e-06,
      "loss": 0.49,
      "step": 150
    },
    {
      "epoch": 0.7028112449799196,
      "grad_norm": 3.3051962852478027,
      "learning_rate": 3.5e-06,
      "loss": 0.4907,
      "step": 175
    },
    {
      "epoch": 0.8032128514056225,
      "grad_norm": 3.1331520080566406,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.4923,
      "step": 200
    },
    {
      "epoch": 0.9036144578313253,
      "grad_norm": 3.6547775268554688,
      "learning_rate": 4.5e-06,
      "loss": 0.4883,
      "step": 225
    },
    {
      "epoch": 1.0040160642570282,
      "grad_norm": 2.569526433944702,
      "learning_rate": 5e-06,
      "loss": 0.4599,
      "step": 250
    },
    {
      "epoch": 1.104417670682731,
      "grad_norm": 2.9575791358947754,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.3833,
      "step": 275
    },
    {
      "epoch": 1.2048192771084336,
      "grad_norm": 3.450204372406006,
      "learning_rate": 6e-06,
      "loss": 0.3799,
      "step": 300
    },
    {
      "epoch": 1.3052208835341366,
      "grad_norm": 3.037454843521118,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.3795,
      "step": 325
    },
    {
      "epoch": 1.4056224899598393,
      "grad_norm": 2.9566783905029297,
      "learning_rate": 7e-06,
      "loss": 0.3825,
      "step": 350
    },
    {
      "epoch": 1.5060240963855422,
      "grad_norm": 2.975231647491455,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.3744,
      "step": 375
    },
    {
      "epoch": 1.606425702811245,
      "grad_norm": 3.3231241703033447,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.3759,
      "step": 400
    },
    {
      "epoch": 1.7068273092369477,
      "grad_norm": 2.9671576023101807,
      "learning_rate": 8.5e-06,
      "loss": 0.3669,
      "step": 425
    },
    {
      "epoch": 1.8072289156626506,
      "grad_norm": 3.1648316383361816,
      "learning_rate": 9e-06,
      "loss": 0.3448,
      "step": 450
    },
    {
      "epoch": 1.9076305220883534,
      "grad_norm": 2.88740611076355,
      "learning_rate": 9.5e-06,
      "loss": 0.3453,
      "step": 475
    },
    {
      "epoch": 2.0080321285140563,
      "grad_norm": 2.326314687728882,
      "learning_rate": 1e-05,
      "loss": 0.3381,
      "step": 500
    },
    {
      "epoch": 2.108433734939759,
      "grad_norm": 2.6187126636505127,
      "learning_rate": 9.944444444444445e-06,
      "loss": 0.2405,
      "step": 525
    },
    {
      "epoch": 2.208835341365462,
      "grad_norm": 2.2135109901428223,
      "learning_rate": 9.88888888888889e-06,
      "loss": 0.2255,
      "step": 550
    },
    {
      "epoch": 2.3092369477911645,
      "grad_norm": 2.8001041412353516,
      "learning_rate": 9.833333333333333e-06,
      "loss": 0.2205,
      "step": 575
    },
    {
      "epoch": 2.4096385542168672,
      "grad_norm": 2.499293565750122,
      "learning_rate": 9.777777777777779e-06,
      "loss": 0.2284,
      "step": 600
    },
    {
      "epoch": 2.5100401606425704,
      "grad_norm": 2.735288381576538,
      "learning_rate": 9.722222222222223e-06,
      "loss": 0.2314,
      "step": 625
    },
    {
      "epoch": 2.610441767068273,
      "grad_norm": 2.3216466903686523,
      "learning_rate": 9.666666666666667e-06,
      "loss": 0.2234,
      "step": 650
    },
    {
      "epoch": 2.710843373493976,
      "grad_norm": 3.042780637741089,
      "learning_rate": 9.611111111111112e-06,
      "loss": 0.2315,
      "step": 675
    },
    {
      "epoch": 2.8112449799196786,
      "grad_norm": 2.8229448795318604,
      "learning_rate": 9.555555555555556e-06,
      "loss": 0.2292,
      "step": 700
    },
    {
      "epoch": 2.9116465863453813,
      "grad_norm": 2.255049228668213,
      "learning_rate": 9.5e-06,
      "loss": 0.2226,
      "step": 725
    },
    {
      "epoch": 3.0120481927710845,
      "grad_norm": 2.1830315589904785,
      "learning_rate": 9.444444444444445e-06,
      "loss": 0.2049,
      "step": 750
    },
    {
      "epoch": 3.112449799196787,
      "grad_norm": 2.056739330291748,
      "learning_rate": 9.38888888888889e-06,
      "loss": 0.1114,
      "step": 775
    },
    {
      "epoch": 3.21285140562249,
      "grad_norm": 2.144355535507202,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.1084,
      "step": 800
    },
    {
      "epoch": 3.3132530120481927,
      "grad_norm": 2.115220069885254,
      "learning_rate": 9.277777777777778e-06,
      "loss": 0.1072,
      "step": 825
    },
    {
      "epoch": 3.4136546184738954,
      "grad_norm": 1.9734667539596558,
      "learning_rate": 9.222222222222224e-06,
      "loss": 0.1155,
      "step": 850
    },
    {
      "epoch": 3.5140562248995986,
      "grad_norm": 2.23511004447937,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.1145,
      "step": 875
    },
    {
      "epoch": 3.6144578313253013,
      "grad_norm": 2.7143516540527344,
      "learning_rate": 9.111111111111112e-06,
      "loss": 0.1168,
      "step": 900
    },
    {
      "epoch": 3.714859437751004,
      "grad_norm": 1.7502648830413818,
      "learning_rate": 9.055555555555556e-06,
      "loss": 0.1097,
      "step": 925
    },
    {
      "epoch": 3.8152610441767068,
      "grad_norm": 1.9246796369552612,
      "learning_rate": 9e-06,
      "loss": 0.1136,
      "step": 950
    },
    {
      "epoch": 3.9156626506024095,
      "grad_norm": 2.3665595054626465,
      "learning_rate": 8.944444444444446e-06,
      "loss": 0.1147,
      "step": 975
    },
    {
      "epoch": 4.016064257028113,
      "grad_norm": 1.1922588348388672,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.1024,
      "step": 1000
    },
    {
      "epoch": 4.016064257028113,
      "eval_loss": 0.35915979743003845,
      "eval_runtime": 2109.2153,
      "eval_samples_per_second": 2.551,
      "eval_steps_per_second": 0.16,
      "eval_wer": 0.2913833552029902,
      "step": 1000
    },
    {
      "epoch": 4.116465863453815,
      "grad_norm": 1.0898044109344482,
      "learning_rate": 8.833333333333334e-06,
      "loss": 0.0517,
      "step": 1025
    },
    {
      "epoch": 4.216867469879518,
      "grad_norm": 1.187383770942688,
      "learning_rate": 8.777777777777778e-06,
      "loss": 0.0486,
      "step": 1050
    },
    {
      "epoch": 4.317269076305221,
      "grad_norm": 1.626198649406433,
      "learning_rate": 8.722222222222224e-06,
      "loss": 0.0478,
      "step": 1075
    },
    {
      "epoch": 4.417670682730924,
      "grad_norm": 1.0097136497497559,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.0541,
      "step": 1100
    },
    {
      "epoch": 4.518072289156627,
      "grad_norm": 2.22725510597229,
      "learning_rate": 8.611111111111112e-06,
      "loss": 0.05,
      "step": 1125
    },
    {
      "epoch": 4.618473895582329,
      "grad_norm": 1.5070220232009888,
      "learning_rate": 8.555555555555556e-06,
      "loss": 0.0508,
      "step": 1150
    },
    {
      "epoch": 4.718875502008032,
      "grad_norm": 1.563808560371399,
      "learning_rate": 8.5e-06,
      "loss": 0.0549,
      "step": 1175
    },
    {
      "epoch": 4.8192771084337345,
      "grad_norm": 2.125509023666382,
      "learning_rate": 8.444444444444446e-06,
      "loss": 0.0536,
      "step": 1200
    },
    {
      "epoch": 4.919678714859438,
      "grad_norm": 1.5022045373916626,
      "learning_rate": 8.38888888888889e-06,
      "loss": 0.0593,
      "step": 1225
    },
    {
      "epoch": 5.020080321285141,
      "grad_norm": 1.0924606323242188,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.0508,
      "step": 1250
    },
    {
      "epoch": 5.120481927710843,
      "grad_norm": 1.493357539176941,
      "learning_rate": 8.277777777777778e-06,
      "loss": 0.0243,
      "step": 1275
    },
    {
      "epoch": 5.220883534136546,
      "grad_norm": 3.102536916732788,
      "learning_rate": 8.222222222222222e-06,
      "loss": 0.0263,
      "step": 1300
    },
    {
      "epoch": 5.321285140562249,
      "grad_norm": 1.6360998153686523,
      "learning_rate": 8.166666666666668e-06,
      "loss": 0.0248,
      "step": 1325
    },
    {
      "epoch": 5.421686746987952,
      "grad_norm": 1.239522099494934,
      "learning_rate": 8.111111111111112e-06,
      "loss": 0.0232,
      "step": 1350
    },
    {
      "epoch": 5.522088353413655,
      "grad_norm": 1.0750267505645752,
      "learning_rate": 8.055555555555557e-06,
      "loss": 0.0274,
      "step": 1375
    },
    {
      "epoch": 5.622489959839357,
      "grad_norm": 0.8077370524406433,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0242,
      "step": 1400
    },
    {
      "epoch": 5.72289156626506,
      "grad_norm": 0.8543096780776978,
      "learning_rate": 7.944444444444445e-06,
      "loss": 0.0259,
      "step": 1425
    },
    {
      "epoch": 5.823293172690763,
      "grad_norm": 1.09683096408844,
      "learning_rate": 7.88888888888889e-06,
      "loss": 0.0224,
      "step": 1450
    },
    {
      "epoch": 5.923694779116466,
      "grad_norm": 1.1407488584518433,
      "learning_rate": 7.833333333333333e-06,
      "loss": 0.0261,
      "step": 1475
    },
    {
      "epoch": 6.024096385542169,
      "grad_norm": 1.050767421722412,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.0222,
      "step": 1500
    },
    {
      "epoch": 6.124497991967871,
      "grad_norm": 0.547297477722168,
      "learning_rate": 7.722222222222223e-06,
      "loss": 0.015,
      "step": 1525
    },
    {
      "epoch": 6.224899598393574,
      "grad_norm": 0.6439629793167114,
      "learning_rate": 7.666666666666667e-06,
      "loss": 0.013,
      "step": 1550
    },
    {
      "epoch": 6.325301204819277,
      "grad_norm": 0.43862244486808777,
      "learning_rate": 7.611111111111111e-06,
      "loss": 0.012,
      "step": 1575
    },
    {
      "epoch": 6.42570281124498,
      "grad_norm": 0.27668318152427673,
      "learning_rate": 7.555555555555556e-06,
      "loss": 0.0096,
      "step": 1600
    },
    {
      "epoch": 6.526104417670683,
      "grad_norm": 0.33968040347099304,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.0095,
      "step": 1625
    },
    {
      "epoch": 6.626506024096385,
      "grad_norm": 1.3353443145751953,
      "learning_rate": 7.444444444444445e-06,
      "loss": 0.0104,
      "step": 1650
    },
    {
      "epoch": 6.7269076305220885,
      "grad_norm": 0.7162081599235535,
      "learning_rate": 7.38888888888889e-06,
      "loss": 0.0135,
      "step": 1675
    },
    {
      "epoch": 6.827309236947791,
      "grad_norm": 0.9529353976249695,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.0128,
      "step": 1700
    },
    {
      "epoch": 6.927710843373494,
      "grad_norm": 0.8639408349990845,
      "learning_rate": 7.277777777777778e-06,
      "loss": 0.0142,
      "step": 1725
    },
    {
      "epoch": 7.028112449799197,
      "grad_norm": 0.6225258708000183,
      "learning_rate": 7.222222222222223e-06,
      "loss": 0.0137,
      "step": 1750
    },
    {
      "epoch": 7.128514056224899,
      "grad_norm": 0.5014215707778931,
      "learning_rate": 7.166666666666667e-06,
      "loss": 0.0074,
      "step": 1775
    },
    {
      "epoch": 7.228915662650603,
      "grad_norm": 0.5766475796699524,
      "learning_rate": 7.111111111111112e-06,
      "loss": 0.0069,
      "step": 1800
    },
    {
      "epoch": 7.329317269076305,
      "grad_norm": 0.48634541034698486,
      "learning_rate": 7.055555555555557e-06,
      "loss": 0.0064,
      "step": 1825
    },
    {
      "epoch": 7.429718875502008,
      "grad_norm": 1.4947255849838257,
      "learning_rate": 7e-06,
      "loss": 0.0056,
      "step": 1850
    },
    {
      "epoch": 7.530120481927711,
      "grad_norm": 0.29385507106781006,
      "learning_rate": 6.944444444444445e-06,
      "loss": 0.0065,
      "step": 1875
    },
    {
      "epoch": 7.6305220883534135,
      "grad_norm": 0.2936380207538605,
      "learning_rate": 6.88888888888889e-06,
      "loss": 0.0058,
      "step": 1900
    },
    {
      "epoch": 7.730923694779117,
      "grad_norm": 0.5639839768409729,
      "learning_rate": 6.833333333333334e-06,
      "loss": 0.0074,
      "step": 1925
    },
    {
      "epoch": 7.831325301204819,
      "grad_norm": 0.6243617534637451,
      "learning_rate": 6.777777777777779e-06,
      "loss": 0.0078,
      "step": 1950
    },
    {
      "epoch": 7.931726907630522,
      "grad_norm": 0.3415360748767853,
      "learning_rate": 6.7222222222222235e-06,
      "loss": 0.0057,
      "step": 1975
    },
    {
      "epoch": 8.032128514056225,
      "grad_norm": 0.2595039904117584,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.0052,
      "step": 2000
    },
    {
      "epoch": 8.032128514056225,
      "eval_loss": 0.43360185623168945,
      "eval_runtime": 2091.0326,
      "eval_samples_per_second": 2.573,
      "eval_steps_per_second": 0.161,
      "eval_wer": 0.26667064296791126,
      "step": 2000
    },
    {
      "epoch": 8.132530120481928,
      "grad_norm": 0.6806650757789612,
      "learning_rate": 6.6111111111111115e-06,
      "loss": 0.0037,
      "step": 2025
    },
    {
      "epoch": 8.23293172690763,
      "grad_norm": 0.505061149597168,
      "learning_rate": 6.555555555555556e-06,
      "loss": 0.0038,
      "step": 2050
    },
    {
      "epoch": 8.333333333333334,
      "grad_norm": 0.1494867503643036,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.0037,
      "step": 2075
    },
    {
      "epoch": 8.433734939759036,
      "grad_norm": 1.245992660522461,
      "learning_rate": 6.444444444444445e-06,
      "loss": 0.0043,
      "step": 2100
    },
    {
      "epoch": 8.534136546184738,
      "grad_norm": 0.4061618745326996,
      "learning_rate": 6.3888888888888885e-06,
      "loss": 0.0033,
      "step": 2125
    },
    {
      "epoch": 8.634538152610443,
      "grad_norm": 0.22515226900577545,
      "learning_rate": 6.333333333333333e-06,
      "loss": 0.0035,
      "step": 2150
    },
    {
      "epoch": 8.734939759036145,
      "grad_norm": 0.11906741559505463,
      "learning_rate": 6.277777777777778e-06,
      "loss": 0.0036,
      "step": 2175
    },
    {
      "epoch": 8.835341365461847,
      "grad_norm": 0.41508978605270386,
      "learning_rate": 6.222222222222223e-06,
      "loss": 0.0037,
      "step": 2200
    },
    {
      "epoch": 8.93574297188755,
      "grad_norm": 0.5366353988647461,
      "learning_rate": 6.166666666666667e-06,
      "loss": 0.0042,
      "step": 2225
    },
    {
      "epoch": 9.036144578313253,
      "grad_norm": 0.17406271398067474,
      "learning_rate": 6.111111111111112e-06,
      "loss": 0.0039,
      "step": 2250
    },
    {
      "epoch": 9.136546184738956,
      "grad_norm": 0.07782778143882751,
      "learning_rate": 6.055555555555555e-06,
      "loss": 0.0029,
      "step": 2275
    },
    {
      "epoch": 9.236947791164658,
      "grad_norm": 0.6227321028709412,
      "learning_rate": 6e-06,
      "loss": 0.003,
      "step": 2300
    },
    {
      "epoch": 9.337349397590362,
      "grad_norm": 0.13376328349113464,
      "learning_rate": 5.944444444444445e-06,
      "loss": 0.0031,
      "step": 2325
    },
    {
      "epoch": 9.437751004016064,
      "grad_norm": 0.2563847303390503,
      "learning_rate": 5.88888888888889e-06,
      "loss": 0.0032,
      "step": 2350
    },
    {
      "epoch": 9.538152610441767,
      "grad_norm": 0.6586058139801025,
      "learning_rate": 5.833333333333334e-06,
      "loss": 0.0037,
      "step": 2375
    },
    {
      "epoch": 9.638554216867469,
      "grad_norm": 0.11519382148981094,
      "learning_rate": 5.777777777777778e-06,
      "loss": 0.0029,
      "step": 2400
    },
    {
      "epoch": 9.738955823293173,
      "grad_norm": 1.0726128816604614,
      "learning_rate": 5.722222222222222e-06,
      "loss": 0.0046,
      "step": 2425
    },
    {
      "epoch": 9.839357429718875,
      "grad_norm": 0.2474745362997055,
      "learning_rate": 5.666666666666667e-06,
      "loss": 0.0033,
      "step": 2450
    },
    {
      "epoch": 9.939759036144578,
      "grad_norm": 0.29594412446022034,
      "learning_rate": 5.611111111111112e-06,
      "loss": 0.0032,
      "step": 2475
    },
    {
      "epoch": 10.040160642570282,
      "grad_norm": 0.15369091928005219,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.0039,
      "step": 2500
    },
    {
      "epoch": 10.140562248995984,
      "grad_norm": 0.1087050512433052,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.0029,
      "step": 2525
    },
    {
      "epoch": 10.240963855421686,
      "grad_norm": 0.6390794515609741,
      "learning_rate": 5.444444444444445e-06,
      "loss": 0.0021,
      "step": 2550
    },
    {
      "epoch": 10.34136546184739,
      "grad_norm": 0.12335559725761414,
      "learning_rate": 5.388888888888889e-06,
      "loss": 0.002,
      "step": 2575
    },
    {
      "epoch": 10.441767068273093,
      "grad_norm": 0.10374537110328674,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.0019,
      "step": 2600
    },
    {
      "epoch": 10.542168674698795,
      "grad_norm": 0.21587899327278137,
      "learning_rate": 5.2777777777777785e-06,
      "loss": 0.0025,
      "step": 2625
    },
    {
      "epoch": 10.642570281124499,
      "grad_norm": 0.479494571685791,
      "learning_rate": 5.2222222222222226e-06,
      "loss": 0.0032,
      "step": 2650
    },
    {
      "epoch": 10.742971887550201,
      "grad_norm": 0.09270445257425308,
      "learning_rate": 5.1666666666666675e-06,
      "loss": 0.0025,
      "step": 2675
    },
    {
      "epoch": 10.843373493975903,
      "grad_norm": 0.10902594029903412,
      "learning_rate": 5.1111111111111115e-06,
      "loss": 0.0021,
      "step": 2700
    },
    {
      "epoch": 10.943775100401606,
      "grad_norm": 0.08857084810733795,
      "learning_rate": 5.0555555555555555e-06,
      "loss": 0.0024,
      "step": 2725
    },
    {
      "epoch": 11.04417670682731,
      "grad_norm": 0.06791070848703384,
      "learning_rate": 5e-06,
      "loss": 0.0022,
      "step": 2750
    },
    {
      "epoch": 11.144578313253012,
      "grad_norm": 0.07958344370126724,
      "learning_rate": 4.944444444444445e-06,
      "loss": 0.0013,
      "step": 2775
    },
    {
      "epoch": 11.244979919678714,
      "grad_norm": 0.0524766705930233,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.0017,
      "step": 2800
    },
    {
      "epoch": 11.345381526104418,
      "grad_norm": 0.07896359264850616,
      "learning_rate": 4.833333333333333e-06,
      "loss": 0.002,
      "step": 2825
    },
    {
      "epoch": 11.44578313253012,
      "grad_norm": 0.07483222335577011,
      "learning_rate": 4.777777777777778e-06,
      "loss": 0.0012,
      "step": 2850
    },
    {
      "epoch": 11.546184738955823,
      "grad_norm": 0.04400847479701042,
      "learning_rate": 4.722222222222222e-06,
      "loss": 0.0013,
      "step": 2875
    },
    {
      "epoch": 11.646586345381525,
      "grad_norm": 0.2449353188276291,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.0014,
      "step": 2900
    },
    {
      "epoch": 11.74698795180723,
      "grad_norm": 0.04744872450828552,
      "learning_rate": 4.611111111111112e-06,
      "loss": 0.0018,
      "step": 2925
    },
    {
      "epoch": 11.847389558232932,
      "grad_norm": 0.07577396929264069,
      "learning_rate": 4.555555555555556e-06,
      "loss": 0.002,
      "step": 2950
    },
    {
      "epoch": 11.947791164658634,
      "grad_norm": 0.05065092816948891,
      "learning_rate": 4.5e-06,
      "loss": 0.002,
      "step": 2975
    },
    {
      "epoch": 12.048192771084338,
      "grad_norm": 0.047723036259412766,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.0014,
      "step": 3000
    },
    {
      "epoch": 12.048192771084338,
      "eval_loss": 0.47207939624786377,
      "eval_runtime": 2099.5139,
      "eval_samples_per_second": 2.563,
      "eval_steps_per_second": 0.161,
      "eval_wer": 0.2708059962622768,
      "step": 3000
    },
    {
      "epoch": 12.14859437751004,
      "grad_norm": 0.028108999133110046,
      "learning_rate": 4.388888888888889e-06,
      "loss": 0.0012,
      "step": 3025
    },
    {
      "epoch": 12.248995983935743,
      "grad_norm": 0.2570669651031494,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.0017,
      "step": 3050
    },
    {
      "epoch": 12.349397590361447,
      "grad_norm": 0.17378300428390503,
      "learning_rate": 4.277777777777778e-06,
      "loss": 0.0013,
      "step": 3075
    },
    {
      "epoch": 12.449799196787149,
      "grad_norm": 0.02885044366121292,
      "learning_rate": 4.222222222222223e-06,
      "loss": 0.0012,
      "step": 3100
    },
    {
      "epoch": 12.550200803212851,
      "grad_norm": 0.03352699801325798,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.001,
      "step": 3125
    },
    {
      "epoch": 12.650602409638553,
      "grad_norm": 0.03538672626018524,
      "learning_rate": 4.111111111111111e-06,
      "loss": 0.001,
      "step": 3150
    },
    {
      "epoch": 12.751004016064257,
      "grad_norm": 0.3616931736469269,
      "learning_rate": 4.055555555555556e-06,
      "loss": 0.0012,
      "step": 3175
    },
    {
      "epoch": 12.85140562248996,
      "grad_norm": 0.0438653789460659,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0012,
      "step": 3200
    },
    {
      "epoch": 12.951807228915662,
      "grad_norm": 0.03480483964085579,
      "learning_rate": 3.944444444444445e-06,
      "loss": 0.001,
      "step": 3225
    },
    {
      "epoch": 13.052208835341366,
      "grad_norm": 0.023823482915759087,
      "learning_rate": 3.88888888888889e-06,
      "loss": 0.0009,
      "step": 3250
    },
    {
      "epoch": 13.152610441767068,
      "grad_norm": 0.02707427553832531,
      "learning_rate": 3.833333333333334e-06,
      "loss": 0.0008,
      "step": 3275
    },
    {
      "epoch": 13.25301204819277,
      "grad_norm": 0.027042578905820847,
      "learning_rate": 3.777777777777778e-06,
      "loss": 0.0008,
      "step": 3300
    },
    {
      "epoch": 13.353413654618475,
      "grad_norm": 0.030249129980802536,
      "learning_rate": 3.7222222222222225e-06,
      "loss": 0.0008,
      "step": 3325
    },
    {
      "epoch": 13.453815261044177,
      "grad_norm": 0.024524545297026634,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 0.0008,
      "step": 3350
    },
    {
      "epoch": 13.55421686746988,
      "grad_norm": 0.025434421375393867,
      "learning_rate": 3.6111111111111115e-06,
      "loss": 0.0008,
      "step": 3375
    },
    {
      "epoch": 13.654618473895582,
      "grad_norm": 0.24862855672836304,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0009,
      "step": 3400
    },
    {
      "epoch": 13.755020080321286,
      "grad_norm": 0.025703459978103638,
      "learning_rate": 3.5e-06,
      "loss": 0.0008,
      "step": 3425
    },
    {
      "epoch": 13.855421686746988,
      "grad_norm": 0.02714374288916588,
      "learning_rate": 3.444444444444445e-06,
      "loss": 0.0008,
      "step": 3450
    },
    {
      "epoch": 13.95582329317269,
      "grad_norm": 0.02454632706940174,
      "learning_rate": 3.3888888888888893e-06,
      "loss": 0.0008,
      "step": 3475
    },
    {
      "epoch": 14.056224899598394,
      "grad_norm": 0.019607126712799072,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.0007,
      "step": 3500
    },
    {
      "epoch": 14.156626506024097,
      "grad_norm": 0.020296042785048485,
      "learning_rate": 3.277777777777778e-06,
      "loss": 0.0007,
      "step": 3525
    },
    {
      "epoch": 14.257028112449799,
      "grad_norm": 0.02188483253121376,
      "learning_rate": 3.2222222222222227e-06,
      "loss": 0.0007,
      "step": 3550
    },
    {
      "epoch": 14.357429718875501,
      "grad_norm": 0.021025992929935455,
      "learning_rate": 3.1666666666666667e-06,
      "loss": 0.0007,
      "step": 3575
    },
    {
      "epoch": 14.457831325301205,
      "grad_norm": 0.021677756682038307,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.0007,
      "step": 3600
    },
    {
      "epoch": 14.558232931726907,
      "grad_norm": 0.02050989866256714,
      "learning_rate": 3.055555555555556e-06,
      "loss": 0.0007,
      "step": 3625
    },
    {
      "epoch": 14.65863453815261,
      "grad_norm": 0.02166111022233963,
      "learning_rate": 3e-06,
      "loss": 0.0007,
      "step": 3650
    },
    {
      "epoch": 14.759036144578314,
      "grad_norm": 0.020434729754924774,
      "learning_rate": 2.944444444444445e-06,
      "loss": 0.0008,
      "step": 3675
    },
    {
      "epoch": 14.859437751004016,
      "grad_norm": 0.022142188623547554,
      "learning_rate": 2.888888888888889e-06,
      "loss": 0.0007,
      "step": 3700
    },
    {
      "epoch": 14.959839357429718,
      "grad_norm": 0.02277183160185814,
      "learning_rate": 2.8333333333333335e-06,
      "loss": 0.0007,
      "step": 3725
    },
    {
      "epoch": 15.060240963855422,
      "grad_norm": 0.020305529236793518,
      "learning_rate": 2.7777777777777783e-06,
      "loss": 0.0007,
      "step": 3750
    },
    {
      "epoch": 15.160642570281125,
      "grad_norm": 0.018854940310120583,
      "learning_rate": 2.7222222222222224e-06,
      "loss": 0.0007,
      "step": 3775
    },
    {
      "epoch": 15.261044176706827,
      "grad_norm": 0.02211603708565235,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.0006,
      "step": 3800
    },
    {
      "epoch": 15.36144578313253,
      "grad_norm": 0.019066350534558296,
      "learning_rate": 2.6111111111111113e-06,
      "loss": 0.0006,
      "step": 3825
    },
    {
      "epoch": 15.461847389558233,
      "grad_norm": 0.01600765995681286,
      "learning_rate": 2.5555555555555557e-06,
      "loss": 0.0006,
      "step": 3850
    },
    {
      "epoch": 15.562248995983936,
      "grad_norm": 0.022780191153287888,
      "learning_rate": 2.5e-06,
      "loss": 0.0006,
      "step": 3875
    },
    {
      "epoch": 15.662650602409638,
      "grad_norm": 0.020977351814508438,
      "learning_rate": 2.4444444444444447e-06,
      "loss": 0.0006,
      "step": 3900
    },
    {
      "epoch": 15.763052208835342,
      "grad_norm": 0.022907493636012077,
      "learning_rate": 2.388888888888889e-06,
      "loss": 0.0006,
      "step": 3925
    },
    {
      "epoch": 15.863453815261044,
      "grad_norm": 0.019752057269215584,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 0.0006,
      "step": 3950
    },
    {
      "epoch": 15.963855421686747,
      "grad_norm": 0.01936071738600731,
      "learning_rate": 2.277777777777778e-06,
      "loss": 0.0006,
      "step": 3975
    },
    {
      "epoch": 16.06425702811245,
      "grad_norm": 0.01969081535935402,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.0006,
      "step": 4000
    },
    {
      "epoch": 16.06425702811245,
      "eval_loss": 0.49721163511276245,
      "eval_runtime": 2085.4766,
      "eval_samples_per_second": 2.58,
      "eval_steps_per_second": 0.162,
      "eval_wer": 0.2700306175195833,
      "step": 4000
    },
    {
      "epoch": 16.164658634538153,
      "grad_norm": 0.018238188698887825,
      "learning_rate": 2.166666666666667e-06,
      "loss": 0.0006,
      "step": 4025
    },
    {
      "epoch": 16.265060240963855,
      "grad_norm": 0.01677379012107849,
      "learning_rate": 2.1111111111111114e-06,
      "loss": 0.0006,
      "step": 4050
    },
    {
      "epoch": 16.365461847389557,
      "grad_norm": 0.015731429681181908,
      "learning_rate": 2.0555555555555555e-06,
      "loss": 0.0006,
      "step": 4075
    },
    {
      "epoch": 16.46586345381526,
      "grad_norm": 0.01889980211853981,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0006,
      "step": 4100
    },
    {
      "epoch": 16.566265060240966,
      "grad_norm": 0.017136871814727783,
      "learning_rate": 1.944444444444445e-06,
      "loss": 0.0006,
      "step": 4125
    },
    {
      "epoch": 16.666666666666668,
      "grad_norm": 0.01828804425895214,
      "learning_rate": 1.888888888888889e-06,
      "loss": 0.0006,
      "step": 4150
    },
    {
      "epoch": 16.76706827309237,
      "grad_norm": 0.019622551277279854,
      "learning_rate": 1.8333333333333333e-06,
      "loss": 0.0007,
      "step": 4175
    },
    {
      "epoch": 16.867469879518072,
      "grad_norm": 0.01913955993950367,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.0006,
      "step": 4200
    },
    {
      "epoch": 16.967871485943775,
      "grad_norm": 0.020911525934934616,
      "learning_rate": 1.7222222222222224e-06,
      "loss": 0.0006,
      "step": 4225
    },
    {
      "epoch": 17.068273092369477,
      "grad_norm": 0.016923176124691963,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.0005,
      "step": 4250
    },
    {
      "epoch": 17.16867469879518,
      "grad_norm": 0.01572297513484955,
      "learning_rate": 1.6111111111111113e-06,
      "loss": 0.0005,
      "step": 4275
    },
    {
      "epoch": 17.269076305220885,
      "grad_norm": 0.016946347430348396,
      "learning_rate": 1.5555555555555558e-06,
      "loss": 0.0006,
      "step": 4300
    },
    {
      "epoch": 17.369477911646587,
      "grad_norm": 0.015353444032371044,
      "learning_rate": 1.5e-06,
      "loss": 0.0005,
      "step": 4325
    },
    {
      "epoch": 17.46987951807229,
      "grad_norm": 0.017800990492105484,
      "learning_rate": 1.4444444444444445e-06,
      "loss": 0.0005,
      "step": 4350
    },
    {
      "epoch": 17.570281124497992,
      "grad_norm": 0.016689756885170937,
      "learning_rate": 1.3888888888888892e-06,
      "loss": 0.0005,
      "step": 4375
    },
    {
      "epoch": 17.670682730923694,
      "grad_norm": 0.019084421917796135,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0005,
      "step": 4400
    },
    {
      "epoch": 17.771084337349397,
      "grad_norm": 0.01737227849662304,
      "learning_rate": 1.2777777777777779e-06,
      "loss": 0.0005,
      "step": 4425
    },
    {
      "epoch": 17.8714859437751,
      "grad_norm": 0.019846035167574883,
      "learning_rate": 1.2222222222222223e-06,
      "loss": 0.0006,
      "step": 4450
    },
    {
      "epoch": 17.971887550200805,
      "grad_norm": 0.016533395275473595,
      "learning_rate": 1.1666666666666668e-06,
      "loss": 0.0005,
      "step": 4475
    },
    {
      "epoch": 18.072289156626507,
      "grad_norm": 0.01615770161151886,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.0005,
      "step": 4500
    },
    {
      "epoch": 18.17269076305221,
      "grad_norm": 0.01649695262312889,
      "learning_rate": 1.0555555555555557e-06,
      "loss": 0.0005,
      "step": 4525
    },
    {
      "epoch": 18.27309236947791,
      "grad_norm": 0.015866920351982117,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0005,
      "step": 4550
    },
    {
      "epoch": 18.373493975903614,
      "grad_norm": 0.01560311671346426,
      "learning_rate": 9.444444444444445e-07,
      "loss": 0.0005,
      "step": 4575
    },
    {
      "epoch": 18.473895582329316,
      "grad_norm": 0.014042481780052185,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.0005,
      "step": 4600
    },
    {
      "epoch": 18.57429718875502,
      "grad_norm": 0.017769090831279755,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.0005,
      "step": 4625
    },
    {
      "epoch": 18.674698795180724,
      "grad_norm": 0.016878824681043625,
      "learning_rate": 7.777777777777779e-07,
      "loss": 0.0005,
      "step": 4650
    },
    {
      "epoch": 18.775100401606426,
      "grad_norm": 0.017039382830262184,
      "learning_rate": 7.222222222222222e-07,
      "loss": 0.0005,
      "step": 4675
    },
    {
      "epoch": 18.87550200803213,
      "grad_norm": 0.016136417165398598,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.0005,
      "step": 4700
    },
    {
      "epoch": 18.97590361445783,
      "grad_norm": 0.015085269697010517,
      "learning_rate": 6.111111111111112e-07,
      "loss": 0.0005,
      "step": 4725
    },
    {
      "epoch": 19.076305220883533,
      "grad_norm": 0.015243038535118103,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.0005,
      "step": 4750
    },
    {
      "epoch": 19.176706827309236,
      "grad_norm": 0.014165305532515049,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.0005,
      "step": 4775
    },
    {
      "epoch": 19.27710843373494,
      "grad_norm": 0.014705569483339787,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.0005,
      "step": 4800
    },
    {
      "epoch": 19.377510040160644,
      "grad_norm": 0.014895942993462086,
      "learning_rate": 3.8888888888888895e-07,
      "loss": 0.0005,
      "step": 4825
    },
    {
      "epoch": 19.477911646586346,
      "grad_norm": 0.01476211566478014,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 0.0005,
      "step": 4850
    },
    {
      "epoch": 19.57831325301205,
      "grad_norm": 0.015145435929298401,
      "learning_rate": 2.7777777777777776e-07,
      "loss": 0.0005,
      "step": 4875
    },
    {
      "epoch": 19.67871485943775,
      "grad_norm": 0.016057029366493225,
      "learning_rate": 2.2222222222222224e-07,
      "loss": 0.0005,
      "step": 4900
    },
    {
      "epoch": 19.779116465863453,
      "grad_norm": 0.01526793371886015,
      "learning_rate": 1.6666666666666668e-07,
      "loss": 0.0005,
      "step": 4925
    },
    {
      "epoch": 19.879518072289155,
      "grad_norm": 0.014341037720441818,
      "learning_rate": 1.1111111111111112e-07,
      "loss": 0.0005,
      "step": 4950
    },
    {
      "epoch": 19.97991967871486,
      "grad_norm": 0.01495534647256136,
      "learning_rate": 5.555555555555556e-08,
      "loss": 0.0005,
      "step": 4975
    },
    {
      "epoch": 20.080321285140563,
      "grad_norm": 0.015628041699528694,
      "learning_rate": 0.0,
      "loss": 0.0005,
      "step": 5000
    },
    {
      "epoch": 20.080321285140563,
      "eval_loss": 0.506655752658844,
      "eval_runtime": 2090.8018,
      "eval_samples_per_second": 2.574,
      "eval_steps_per_second": 0.161,
      "eval_wer": 0.2715416119925246,
      "step": 5000
    },
    {
      "epoch": 20.080321285140563,
      "step": 5000,
      "total_flos": 5.435997290496e+20,
      "train_loss": 0.07043341839555652,
      "train_runtime": 38891.7587,
      "train_samples_per_second": 4.114,
      "train_steps_per_second": 0.129
    }
  ],
  "logging_steps": 25,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 21,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.435997290496e+20,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}