{
  "best_metric": 0.3642527461051941,
  "best_model_checkpoint": "./working/checkpoint-9000",
  "epoch": 107.14285714285714,
  "eval_steps": 1000,
  "global_step": 9000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.19,
      "grad_norm": 5.874335765838623,
      "learning_rate": 1.5e-06,
      "loss": 12.5691,
      "step": 100
    },
    {
      "epoch": 2.38,
      "grad_norm": 11.808794975280762,
      "learning_rate": 2.9850000000000002e-06,
      "loss": 9.8251,
      "step": 200
    },
    {
      "epoch": 3.57,
      "grad_norm": 10.97357177734375,
      "learning_rate": 4.485e-06,
      "loss": 6.5329,
      "step": 300
    },
    {
      "epoch": 4.76,
      "grad_norm": 10.752958297729492,
      "learning_rate": 5.985e-06,
      "loss": 5.2405,
      "step": 400
    },
    {
      "epoch": 5.95,
      "grad_norm": 5.865973949432373,
      "learning_rate": 7.485e-06,
      "loss": 4.629,
      "step": 500
    },
    {
      "epoch": 7.14,
      "grad_norm": 4.192840099334717,
      "learning_rate": 8.985e-06,
      "loss": 4.2437,
      "step": 600
    },
    {
      "epoch": 8.33,
      "grad_norm": 3.2876224517822266,
      "learning_rate": 1.0485e-05,
      "loss": 3.9784,
      "step": 700
    },
    {
      "epoch": 9.52,
      "grad_norm": 1.2407236099243164,
      "learning_rate": 1.1985000000000001e-05,
      "loss": 3.7904,
      "step": 800
    },
    {
      "epoch": 10.71,
      "grad_norm": 0.6437963247299194,
      "learning_rate": 1.3485e-05,
      "loss": 3.6682,
      "step": 900
    },
    {
      "epoch": 11.9,
      "grad_norm": 0.3295372724533081,
      "learning_rate": 1.4985e-05,
      "loss": 3.5325,
      "step": 1000
    },
    {
      "epoch": 11.9,
      "eval_cer": 0.926601333602748,
      "eval_loss": 3.4897494316101074,
      "eval_runtime": 10.3726,
      "eval_samples_per_second": 32.393,
      "eval_steps_per_second": 4.049,
      "eval_wer": 1.0,
      "step": 1000
    },
    {
      "epoch": 13.1,
      "grad_norm": 0.61485356092453,
      "learning_rate": 1.6485e-05,
      "loss": 3.4934,
      "step": 1100
    },
    {
      "epoch": 14.29,
      "grad_norm": 0.4778901934623718,
      "learning_rate": 1.7985e-05,
      "loss": 3.4662,
      "step": 1200
    },
    {
      "epoch": 15.48,
      "grad_norm": 0.56944739818573,
      "learning_rate": 1.9485e-05,
      "loss": 3.4413,
      "step": 1300
    },
    {
      "epoch": 16.67,
      "grad_norm": 0.37235376238822937,
      "learning_rate": 2.0985e-05,
      "loss": 3.4089,
      "step": 1400
    },
    {
      "epoch": 17.86,
      "grad_norm": 0.4277956187725067,
      "learning_rate": 2.2485000000000002e-05,
      "loss": 3.3384,
      "step": 1500
    },
    {
      "epoch": 19.05,
      "grad_norm": 0.50371253490448,
      "learning_rate": 2.3985e-05,
      "loss": 3.2455,
      "step": 1600
    },
    {
      "epoch": 20.24,
      "grad_norm": 1.1383576393127441,
      "learning_rate": 2.5485e-05,
      "loss": 3.0681,
      "step": 1700
    },
    {
      "epoch": 21.43,
      "grad_norm": 0.8668686747550964,
      "learning_rate": 2.6985e-05,
      "loss": 2.7949,
      "step": 1800
    },
    {
      "epoch": 22.62,
      "grad_norm": 1.0731563568115234,
      "learning_rate": 2.8485000000000003e-05,
      "loss": 2.4866,
      "step": 1900
    },
    {
      "epoch": 23.81,
      "grad_norm": 1.3317248821258545,
      "learning_rate": 2.9985000000000002e-05,
      "loss": 2.1973,
      "step": 2000
    },
    {
      "epoch": 23.81,
      "eval_cer": 0.24030107092341887,
      "eval_loss": 1.1350404024124146,
      "eval_runtime": 10.3527,
      "eval_samples_per_second": 32.455,
      "eval_steps_per_second": 4.057,
      "eval_wer": 0.839647119875454,
      "step": 2000
    },
    {
      "epoch": 25.0,
      "grad_norm": 2.0239453315734863,
      "learning_rate": 2.962875e-05,
      "loss": 1.9821,
      "step": 2100
    },
    {
      "epoch": 26.19,
      "grad_norm": 1.464921236038208,
      "learning_rate": 2.925375e-05,
      "loss": 1.853,
      "step": 2200
    },
    {
      "epoch": 27.38,
      "grad_norm": 1.6508703231811523,
      "learning_rate": 2.887875e-05,
      "loss": 1.7547,
      "step": 2300
    },
    {
      "epoch": 28.57,
      "grad_norm": 1.3476407527923584,
      "learning_rate": 2.850375e-05,
      "loss": 1.7171,
      "step": 2400
    },
    {
      "epoch": 29.76,
      "grad_norm": 1.2977994680404663,
      "learning_rate": 2.812875e-05,
      "loss": 1.6498,
      "step": 2500
    },
    {
      "epoch": 30.95,
      "grad_norm": 1.8536533117294312,
      "learning_rate": 2.775375e-05,
      "loss": 1.5965,
      "step": 2600
    },
    {
      "epoch": 32.14,
      "grad_norm": 1.7063647508621216,
      "learning_rate": 2.7378750000000003e-05,
      "loss": 1.5744,
      "step": 2700
    },
    {
      "epoch": 33.33,
      "grad_norm": 1.613274097442627,
      "learning_rate": 2.700375e-05,
      "loss": 1.5483,
      "step": 2800
    },
    {
      "epoch": 34.52,
      "grad_norm": 1.6182752847671509,
      "learning_rate": 2.662875e-05,
      "loss": 1.5076,
      "step": 2900
    },
    {
      "epoch": 35.71,
      "grad_norm": 1.9291083812713623,
      "learning_rate": 2.6253750000000003e-05,
      "loss": 1.4762,
      "step": 3000
    },
    {
      "epoch": 35.71,
      "eval_cer": 0.15634471610426348,
      "eval_loss": 0.527005672454834,
      "eval_runtime": 10.3175,
      "eval_samples_per_second": 32.566,
      "eval_steps_per_second": 4.071,
      "eval_wer": 0.6844836533471718,
      "step": 3000
    },
    {
      "epoch": 36.9,
      "grad_norm": 1.7794642448425293,
      "learning_rate": 2.587875e-05,
      "loss": 1.4752,
      "step": 3100
    },
    {
      "epoch": 38.1,
      "grad_norm": 1.7410004138946533,
      "learning_rate": 2.550375e-05,
      "loss": 1.4586,
      "step": 3200
    },
    {
      "epoch": 39.29,
      "grad_norm": 1.6410831212997437,
      "learning_rate": 2.512875e-05,
      "loss": 1.4172,
      "step": 3300
    },
    {
      "epoch": 40.48,
      "grad_norm": 2.350106954574585,
      "learning_rate": 2.475375e-05,
      "loss": 1.3751,
      "step": 3400
    },
    {
      "epoch": 41.67,
      "grad_norm": 1.7794309854507446,
      "learning_rate": 2.437875e-05,
      "loss": 1.3516,
      "step": 3500
    },
    {
      "epoch": 42.86,
      "grad_norm": 1.8536804914474487,
      "learning_rate": 2.400375e-05,
      "loss": 1.338,
      "step": 3600
    },
    {
      "epoch": 44.05,
      "grad_norm": 2.043091058731079,
      "learning_rate": 2.362875e-05,
      "loss": 1.31,
      "step": 3700
    },
    {
      "epoch": 45.24,
      "grad_norm": 2.556605577468872,
      "learning_rate": 2.325375e-05,
      "loss": 1.2737,
      "step": 3800
    },
    {
      "epoch": 46.43,
      "grad_norm": 2.167360544204712,
      "learning_rate": 2.2878750000000002e-05,
      "loss": 1.2706,
      "step": 3900
    },
    {
      "epoch": 47.62,
      "grad_norm": 1.933358907699585,
      "learning_rate": 2.2503750000000003e-05,
      "loss": 1.2409,
      "step": 4000
    },
    {
      "epoch": 47.62,
      "eval_cer": 0.140280864821176,
      "eval_loss": 0.41946256160736084,
      "eval_runtime": 10.4076,
      "eval_samples_per_second": 32.284,
      "eval_steps_per_second": 4.036,
      "eval_wer": 0.6331084587441619,
      "step": 4000
    },
    {
      "epoch": 48.81,
      "grad_norm": 2.1927788257598877,
      "learning_rate": 2.212875e-05,
      "loss": 1.2476,
      "step": 4100
    },
    {
      "epoch": 50.0,
      "grad_norm": 3.7042958736419678,
      "learning_rate": 2.175375e-05,
      "loss": 1.2211,
      "step": 4200
    },
    {
      "epoch": 51.19,
      "grad_norm": 2.503298282623291,
      "learning_rate": 2.13825e-05,
      "loss": 1.1974,
      "step": 4300
    },
    {
      "epoch": 52.38,
      "grad_norm": 2.378753423690796,
      "learning_rate": 2.101125e-05,
      "loss": 1.19,
      "step": 4400
    },
    {
      "epoch": 53.57,
      "grad_norm": 6.145068645477295,
      "learning_rate": 2.063625e-05,
      "loss": 1.1734,
      "step": 4500
    },
    {
      "epoch": 54.76,
      "grad_norm": 2.2741551399230957,
      "learning_rate": 2.026125e-05,
      "loss": 1.1664,
      "step": 4600
    },
    {
      "epoch": 55.95,
      "grad_norm": 3.21976900100708,
      "learning_rate": 1.988625e-05,
      "loss": 1.1555,
      "step": 4700
    },
    {
      "epoch": 57.14,
      "grad_norm": 3.06923508644104,
      "learning_rate": 1.951125e-05,
      "loss": 1.1391,
      "step": 4800
    },
    {
      "epoch": 58.33,
      "grad_norm": 1.8809341192245483,
      "learning_rate": 1.9136249999999998e-05,
      "loss": 1.1271,
      "step": 4900
    },
    {
      "epoch": 59.52,
      "grad_norm": 2.041844367980957,
      "learning_rate": 1.876125e-05,
      "loss": 1.1241,
      "step": 5000
    },
    {
      "epoch": 59.52,
      "eval_cer": 0.13785613255203072,
      "eval_loss": 0.38446417450904846,
      "eval_runtime": 10.417,
      "eval_samples_per_second": 32.255,
      "eval_steps_per_second": 4.032,
      "eval_wer": 0.63622210690192,
      "step": 5000
    },
    {
      "epoch": 60.71,
      "grad_norm": 2.3043465614318848,
      "learning_rate": 1.838625e-05,
      "loss": 1.1042,
      "step": 5100
    },
    {
      "epoch": 61.9,
      "grad_norm": 3.682835340499878,
      "learning_rate": 1.801125e-05,
      "loss": 1.0921,
      "step": 5200
    },
    {
      "epoch": 63.1,
      "grad_norm": 4.466809272766113,
      "learning_rate": 1.7636250000000002e-05,
      "loss": 1.095,
      "step": 5300
    },
    {
      "epoch": 64.29,
      "grad_norm": 2.615339994430542,
      "learning_rate": 1.726125e-05,
      "loss": 1.1057,
      "step": 5400
    },
    {
      "epoch": 65.48,
      "grad_norm": 3.483346700668335,
      "learning_rate": 1.688625e-05,
      "loss": 1.0553,
      "step": 5500
    },
    {
      "epoch": 66.67,
      "grad_norm": 2.141965866088867,
      "learning_rate": 1.651125e-05,
      "loss": 1.0656,
      "step": 5600
    },
    {
      "epoch": 67.86,
      "grad_norm": 2.2111611366271973,
      "learning_rate": 1.613625e-05,
      "loss": 1.0673,
      "step": 5700
    },
    {
      "epoch": 69.05,
      "grad_norm": 2.071429491043091,
      "learning_rate": 1.576125e-05,
      "loss": 1.0632,
      "step": 5800
    },
    {
      "epoch": 70.24,
      "grad_norm": 4.86116886138916,
      "learning_rate": 1.538625e-05,
      "loss": 1.0447,
      "step": 5900
    },
    {
      "epoch": 71.43,
      "grad_norm": 3.3076369762420654,
      "learning_rate": 1.5011250000000001e-05,
      "loss": 1.024,
      "step": 6000
    },
    {
      "epoch": 71.43,
      "eval_cer": 0.13548191553849262,
      "eval_loss": 0.3715837895870209,
      "eval_runtime": 10.3955,
      "eval_samples_per_second": 32.322,
      "eval_steps_per_second": 4.04,
      "eval_wer": 0.6320705760249092,
      "step": 6000
    },
    {
      "epoch": 72.62,
      "grad_norm": 2.3496508598327637,
      "learning_rate": 1.463625e-05,
      "loss": 1.0379,
      "step": 6100
    },
    {
      "epoch": 73.81,
      "grad_norm": 2.621004343032837,
      "learning_rate": 1.426125e-05,
      "loss": 1.0515,
      "step": 6200
    },
    {
      "epoch": 75.0,
      "grad_norm": 5.240926742553711,
      "learning_rate": 1.388625e-05,
      "loss": 1.0253,
      "step": 6300
    },
    {
      "epoch": 76.19,
      "grad_norm": 2.9943532943725586,
      "learning_rate": 1.351125e-05,
      "loss": 1.0131,
      "step": 6400
    },
    {
      "epoch": 77.38,
      "grad_norm": 2.475804328918457,
      "learning_rate": 1.3136250000000001e-05,
      "loss": 1.0227,
      "step": 6500
    },
    {
      "epoch": 78.57,
      "grad_norm": 2.5056631565093994,
      "learning_rate": 1.2761250000000001e-05,
      "loss": 1.0025,
      "step": 6600
    },
    {
      "epoch": 79.76,
      "grad_norm": 3.9102323055267334,
      "learning_rate": 1.238625e-05,
      "loss": 1.0181,
      "step": 6700
    },
    {
      "epoch": 80.95,
      "grad_norm": 3.3800106048583984,
      "learning_rate": 1.201125e-05,
      "loss": 0.9892,
      "step": 6800
    },
    {
      "epoch": 82.14,
      "grad_norm": 2.165987014770508,
      "learning_rate": 1.164e-05,
      "loss": 0.9822,
      "step": 6900
    },
    {
      "epoch": 83.33,
      "grad_norm": 2.426816463470459,
      "learning_rate": 1.1265e-05,
      "loss": 0.9922,
      "step": 7000
    },
    {
      "epoch": 83.33,
      "eval_cer": 0.13310769852495455,
      "eval_loss": 0.3727741539478302,
      "eval_runtime": 10.3942,
      "eval_samples_per_second": 32.326,
      "eval_steps_per_second": 4.041,
      "eval_wer": 0.628956927867151,
      "step": 7000
    },
    {
      "epoch": 84.52,
      "grad_norm": 3.203552007675171,
      "learning_rate": 1.089e-05,
      "loss": 0.9995,
      "step": 7100
    },
    {
      "epoch": 85.71,
      "grad_norm": 2.827246904373169,
      "learning_rate": 1.0515e-05,
      "loss": 0.9683,
      "step": 7200
    },
    {
      "epoch": 86.9,
      "grad_norm": 2.9927895069122314,
      "learning_rate": 1.0140000000000001e-05,
      "loss": 0.9864,
      "step": 7300
    },
    {
      "epoch": 88.1,
      "grad_norm": 2.151737928390503,
      "learning_rate": 9.765e-06,
      "loss": 0.9744,
      "step": 7400
    },
    {
      "epoch": 89.29,
      "grad_norm": 2.5920581817626953,
      "learning_rate": 9.39e-06,
      "loss": 0.9794,
      "step": 7500
    },
    {
      "epoch": 90.48,
      "grad_norm": 3.2127621173858643,
      "learning_rate": 9.015e-06,
      "loss": 0.9622,
      "step": 7600
    },
    {
      "epoch": 91.67,
      "grad_norm": 3.541879892349243,
      "learning_rate": 8.64e-06,
      "loss": 0.9624,
      "step": 7700
    },
    {
      "epoch": 92.86,
      "grad_norm": 2.827958345413208,
      "learning_rate": 8.265000000000001e-06,
      "loss": 0.9637,
      "step": 7800
    },
    {
      "epoch": 94.05,
      "grad_norm": 2.648591995239258,
      "learning_rate": 7.89e-06,
      "loss": 0.9684,
      "step": 7900
    },
    {
      "epoch": 95.24,
      "grad_norm": 4.264640808105469,
      "learning_rate": 7.515e-06,
      "loss": 0.9432,
      "step": 8000
    },
    {
      "epoch": 95.24,
      "eval_cer": 0.13209739341281068,
      "eval_loss": 0.3648131787776947,
      "eval_runtime": 10.4236,
      "eval_samples_per_second": 32.234,
      "eval_steps_per_second": 4.029,
      "eval_wer": 0.6170212765957447,
      "step": 8000
    },
    {
      "epoch": 96.43,
      "grad_norm": 2.8603055477142334,
      "learning_rate": 7.14e-06,
      "loss": 0.9576,
      "step": 8100
    },
    {
      "epoch": 97.62,
      "grad_norm": 2.931117296218872,
      "learning_rate": 6.7650000000000005e-06,
      "loss": 0.9579,
      "step": 8200
    },
    {
      "epoch": 98.81,
      "grad_norm": 3.449780225753784,
      "learning_rate": 6.39e-06,
      "loss": 0.9535,
      "step": 8300
    },
    {
      "epoch": 100.0,
      "grad_norm": 4.3435468673706055,
      "learning_rate": 6.015000000000001e-06,
      "loss": 0.9463,
      "step": 8400
    },
    {
      "epoch": 101.19,
      "grad_norm": 2.2839837074279785,
      "learning_rate": 5.64e-06,
      "loss": 0.9413,
      "step": 8500
    },
    {
      "epoch": 102.38,
      "grad_norm": 3.1021485328674316,
      "learning_rate": 5.2649999999999996e-06,
      "loss": 0.9436,
      "step": 8600
    },
    {
      "epoch": 103.57,
      "grad_norm": 2.9421229362487793,
      "learning_rate": 4.890000000000001e-06,
      "loss": 0.939,
      "step": 8700
    },
    {
      "epoch": 104.76,
      "grad_norm": 2.0578436851501465,
      "learning_rate": 4.515e-06,
      "loss": 0.9338,
      "step": 8800
    },
    {
      "epoch": 105.95,
      "grad_norm": 3.5860297679901123,
      "learning_rate": 4.14e-06,
      "loss": 0.9315,
      "step": 8900
    },
    {
      "epoch": 107.14,
      "grad_norm": 2.1002159118652344,
      "learning_rate": 3.765e-06,
      "loss": 0.9279,
      "step": 9000
    },
    {
      "epoch": 107.14,
      "eval_cer": 0.13245100020206102,
      "eval_loss": 0.3642527461051941,
      "eval_runtime": 10.5077,
      "eval_samples_per_second": 31.976,
      "eval_steps_per_second": 3.997,
      "eval_wer": 0.6248053969901401,
      "step": 9000
    }
  ],
  "logging_steps": 100,
  "max_steps": 10000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 120,
  "save_steps": 1000,
  "total_flos": 2.7115316842745717e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}