trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.38476337052712584,
  "eval_steps": 500,
  "global_step": 6000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03206361421059382,
      "grad_norm": 10.316516876220703,
      "learning_rate": 0.0002465,
      "loss": 4.7051,
      "step": 500
    },
    {
      "epoch": 0.03206361421059382,
      "eval_loss": 1.6561740636825562,
      "eval_runtime": 191.0416,
      "eval_samples_per_second": 36.756,
      "eval_steps_per_second": 2.298,
      "eval_wer": 0.9354894431230816,
      "step": 500
    },
    {
      "epoch": 0.06412722842118763,
      "grad_norm": 4.912766933441162,
      "learning_rate": 0.00027816666666666663,
      "loss": 1.0362,
      "step": 1000
    },
    {
      "epoch": 0.06412722842118763,
      "eval_loss": 1.1839842796325684,
      "eval_runtime": 175.5788,
      "eval_samples_per_second": 39.993,
      "eval_steps_per_second": 2.5,
      "eval_wer": 0.7977517639086356,
      "step": 1000
    },
    {
      "epoch": 0.09619084263178146,
      "grad_norm": 4.178534030914307,
      "learning_rate": 0.00025038888888888886,
      "loss": 0.811,
      "step": 1500
    },
    {
      "epoch": 0.09619084263178146,
      "eval_loss": 1.0081175565719604,
      "eval_runtime": 175.5534,
      "eval_samples_per_second": 39.999,
      "eval_steps_per_second": 2.501,
      "eval_wer": 0.7174690070290597,
      "step": 1500
    },
    {
      "epoch": 0.12825445684237527,
      "grad_norm": 7.957609176635742,
      "learning_rate": 0.0002226111111111111,
      "loss": 0.6903,
      "step": 2000
    },
    {
      "epoch": 0.12825445684237527,
      "eval_loss": 0.8935067057609558,
      "eval_runtime": 176.1028,
      "eval_samples_per_second": 39.874,
      "eval_steps_per_second": 2.493,
      "eval_wer": 0.6401094885661516,
      "step": 2000
    },
    {
      "epoch": 0.16031807105296908,
      "grad_norm": 6.541042327880859,
      "learning_rate": 0.00019483333333333332,
      "loss": 0.6238,
      "step": 2500
    },
    {
      "epoch": 0.16031807105296908,
      "eval_loss": 0.8059775233268738,
      "eval_runtime": 177.0587,
      "eval_samples_per_second": 39.659,
      "eval_steps_per_second": 2.479,
      "eval_wer": 0.5848868573858276,
      "step": 2500
    },
    {
      "epoch": 0.19238168526356292,
      "grad_norm": 5.036514759063721,
      "learning_rate": 0.00016705555555555554,
      "loss": 0.5649,
      "step": 3000
    },
    {
      "epoch": 0.19238168526356292,
      "eval_loss": 0.7770201563835144,
      "eval_runtime": 178.1454,
      "eval_samples_per_second": 39.417,
      "eval_steps_per_second": 2.464,
      "eval_wer": 0.5589231852668783,
      "step": 3000
    },
    {
      "epoch": 0.22444529947415673,
      "grad_norm": 4.140881538391113,
      "learning_rate": 0.00013927777777777777,
      "loss": 0.5309,
      "step": 3500
    },
    {
      "epoch": 0.22444529947415673,
      "eval_loss": 0.7264481782913208,
      "eval_runtime": 176.7301,
      "eval_samples_per_second": 39.733,
      "eval_steps_per_second": 2.484,
      "eval_wer": 0.5326539018589139,
      "step": 3500
    },
    {
      "epoch": 0.25650891368475054,
      "grad_norm": 6.205782413482666,
      "learning_rate": 0.00011155555555555555,
      "loss": 0.4892,
      "step": 4000
    },
    {
      "epoch": 0.25650891368475054,
      "eval_loss": 0.6865007877349854,
      "eval_runtime": 178.7785,
      "eval_samples_per_second": 39.278,
      "eval_steps_per_second": 2.456,
      "eval_wer": 0.5106366016024662,
      "step": 4000
    },
    {
      "epoch": 0.2885725278953444,
      "grad_norm": 9.618443489074707,
      "learning_rate": 8.388888888888888e-05,
      "loss": 0.4521,
      "step": 4500
    },
    {
      "epoch": 0.2885725278953444,
      "eval_loss": 0.6477864980697632,
      "eval_runtime": 177.7072,
      "eval_samples_per_second": 39.514,
      "eval_steps_per_second": 2.47,
      "eval_wer": 0.4861345486918508,
      "step": 4500
    },
    {
      "epoch": 0.32063614210593816,
      "grad_norm": 8.743291854858398,
      "learning_rate": 5.61111111111111e-05,
      "loss": 0.4309,
      "step": 5000
    },
    {
      "epoch": 0.32063614210593816,
      "eval_loss": 0.6221807599067688,
      "eval_runtime": 177.9115,
      "eval_samples_per_second": 39.469,
      "eval_steps_per_second": 2.468,
      "eval_wer": 0.4763416999960138,
      "step": 5000
    },
    {
      "epoch": 0.352699756316532,
      "grad_norm": 8.870036125183105,
      "learning_rate": 2.833333333333333e-05,
      "loss": 0.4055,
      "step": 5500
    },
    {
      "epoch": 0.352699756316532,
      "eval_loss": 0.5988152027130127,
      "eval_runtime": 178.6515,
      "eval_samples_per_second": 39.306,
      "eval_steps_per_second": 2.457,
      "eval_wer": 0.45257045668956536,
      "step": 5500
    },
    {
      "epoch": 0.38476337052712584,
      "grad_norm": 14.018965721130371,
      "learning_rate": 6.11111111111111e-07,
      "loss": 0.3896,
      "step": 6000
    },
    {
      "epoch": 0.38476337052712584,
      "eval_loss": 0.5865370631217957,
      "eval_runtime": 176.8507,
      "eval_samples_per_second": 39.706,
      "eval_steps_per_second": 2.482,
      "eval_wer": 0.44523578575319894,
      "step": 6000
    },
    {
      "epoch": 0.38476337052712584,
      "step": 6000,
      "total_flos": 6.835515215409967e+18,
      "train_loss": 0.9274428431193034,
      "train_runtime": 4544.4955,
      "train_samples_per_second": 10.562,
      "train_steps_per_second": 1.32
    }
  ],
  "logging_steps": 500,
  "max_steps": 6000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.835515215409967e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}