checkpoint-300/trainer_state.json

{
  "best_global_step": 300,
  "best_metric": 56.034321954859166,
  "best_model_checkpoint": "./JUDICM/checkpoint-300",
  "epoch": 3.5294117647058822,
  "eval_steps": 100,
  "global_step": 300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 9.61864185333252,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 2.9117,
      "step": 20
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 8.609731674194336,
      "learning_rate": 3.7e-06,
      "loss": 2.0419,
      "step": 40
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 7.390444278717041,
      "learning_rate": 5.7e-06,
      "loss": 1.5565,
      "step": 60
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 5.834730625152588,
      "learning_rate": 7.7e-06,
      "loss": 0.9792,
      "step": 80
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 5.353567123413086,
      "learning_rate": 9.7e-06,
      "loss": 0.8465,
      "step": 100
    },
    {
      "epoch": 1.1764705882352942,
      "eval_loss": 0.681628942489624,
      "eval_runtime": 544.1503,
      "eval_samples_per_second": 1.251,
      "eval_steps_per_second": 0.158,
      "eval_wer": 74.07821923770442,
      "step": 100
    },
    {
      "epoch": 1.4117647058823528,
      "grad_norm": 4.50194787979126,
      "learning_rate": 9.773333333333335e-06,
      "loss": 0.6137,
      "step": 120
    },
    {
      "epoch": 1.6470588235294117,
      "grad_norm": 4.959797382354736,
      "learning_rate": 9.506666666666667e-06,
      "loss": 0.6733,
      "step": 140
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 5.115293502807617,
      "learning_rate": 9.240000000000001e-06,
      "loss": 0.7026,
      "step": 160
    },
    {
      "epoch": 2.1176470588235294,
      "grad_norm": 3.1620101928710938,
      "learning_rate": 8.973333333333334e-06,
      "loss": 0.5185,
      "step": 180
    },
    {
      "epoch": 2.3529411764705883,
      "grad_norm": 4.74745512008667,
      "learning_rate": 8.706666666666667e-06,
      "loss": 0.4669,
      "step": 200
    },
    {
      "epoch": 2.3529411764705883,
      "eval_loss": 0.47813770174980164,
      "eval_runtime": 481.7363,
      "eval_samples_per_second": 1.414,
      "eval_steps_per_second": 0.179,
      "eval_wer": 61.49350245600945,
      "step": 200
    },
    {
      "epoch": 2.588235294117647,
      "grad_norm": 3.0848231315612793,
      "learning_rate": 8.44e-06,
      "loss": 0.5316,
      "step": 220
    },
    {
      "epoch": 2.8235294117647056,
      "grad_norm": 3.566438674926758,
      "learning_rate": 8.173333333333334e-06,
      "loss": 0.4629,
      "step": 240
    },
    {
      "epoch": 3.0588235294117645,
      "grad_norm": 2.9106101989746094,
      "learning_rate": 7.906666666666667e-06,
      "loss": 0.4565,
      "step": 260
    },
    {
      "epoch": 3.2941176470588234,
      "grad_norm": 3.077847957611084,
      "learning_rate": 7.640000000000001e-06,
      "loss": 0.345,
      "step": 280
    },
    {
      "epoch": 3.5294117647058822,
      "grad_norm": 3.5237925052642822,
      "learning_rate": 7.373333333333334e-06,
      "loss": 0.3376,
      "step": 300
    },
    {
      "epoch": 3.5294117647058822,
      "eval_loss": 0.367522269487381,
      "eval_runtime": 520.1482,
      "eval_samples_per_second": 1.309,
      "eval_steps_per_second": 0.165,
      "eval_wer": 56.034321954859166,
      "step": 300
    }
  ],
  "logging_steps": 20,
  "max_steps": 850,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 9.797807112192e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}