trainer_state.json

{
  "best_metric": 0.9842931937172775,
  "best_model_checkpoint": "model/checkpoint-94",
  "epoch": 7.703703703703704,
  "eval_steps": 500,
  "global_step": 104,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.59,
      "learning_rate": 0.00036363636363636367,
      "loss": 1.0999,
      "step": 8
    },
    {
      "epoch": 0.96,
      "eval_accuracy": 0.643979057591623,
      "eval_loss": 0.7986512780189514,
      "eval_runtime": 28.581,
      "eval_samples_per_second": 6.683,
      "eval_steps_per_second": 0.21,
      "step": 13
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0004731182795698925,
      "loss": 1.0029,
      "step": 16
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00043010752688172043,
      "loss": 0.6342,
      "step": 24
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9424083769633508,
      "eval_loss": 0.24137574434280396,
      "eval_runtime": 26.3088,
      "eval_samples_per_second": 7.26,
      "eval_steps_per_second": 0.228,
      "step": 27
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0003870967741935484,
      "loss": 0.5732,
      "step": 32
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0003440860215053764,
      "loss": 0.4882,
      "step": 40
    },
    {
      "epoch": 2.96,
      "eval_accuracy": 0.9633507853403142,
      "eval_loss": 0.16461053490638733,
      "eval_runtime": 26.318,
      "eval_samples_per_second": 7.257,
      "eval_steps_per_second": 0.228,
      "step": 40
    },
    {
      "epoch": 3.56,
      "learning_rate": 0.0003010752688172043,
      "loss": 0.463,
      "step": 48
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9424083769633508,
      "eval_loss": 0.25284695625305176,
      "eval_runtime": 26.2832,
      "eval_samples_per_second": 7.267,
      "eval_steps_per_second": 0.228,
      "step": 54
    },
    {
      "epoch": 4.15,
      "learning_rate": 0.00025806451612903227,
      "loss": 0.4286,
      "step": 56
    },
    {
      "epoch": 4.74,
      "learning_rate": 0.00021505376344086021,
      "loss": 0.4609,
      "step": 64
    },
    {
      "epoch": 4.96,
      "eval_accuracy": 0.9790575916230366,
      "eval_loss": 0.11301162838935852,
      "eval_runtime": 26.3727,
      "eval_samples_per_second": 7.242,
      "eval_steps_per_second": 0.228,
      "step": 67
    },
    {
      "epoch": 5.33,
      "learning_rate": 0.0001720430107526882,
      "loss": 0.4636,
      "step": 72
    },
    {
      "epoch": 5.93,
      "learning_rate": 0.00012903225806451613,
      "loss": 0.4251,
      "step": 80
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9633507853403142,
      "eval_loss": 0.13039104640483856,
      "eval_runtime": 26.3002,
      "eval_samples_per_second": 7.262,
      "eval_steps_per_second": 0.228,
      "step": 81
    },
    {
      "epoch": 6.52,
      "learning_rate": 8.60215053763441e-05,
      "loss": 0.3802,
      "step": 88
    },
    {
      "epoch": 6.96,
      "eval_accuracy": 0.9842931937172775,
      "eval_loss": 0.07386188954114914,
      "eval_runtime": 26.267,
      "eval_samples_per_second": 7.271,
      "eval_steps_per_second": 0.228,
      "step": 94
    },
    {
      "epoch": 7.11,
      "learning_rate": 4.301075268817205e-05,
      "loss": 0.378,
      "step": 96
    },
    {
      "epoch": 7.7,
      "learning_rate": 0.0,
      "loss": 0.4147,
      "step": 104
    },
    {
      "epoch": 7.7,
      "eval_accuracy": 0.9842931937172775,
      "eval_loss": 0.06754852086305618,
      "eval_runtime": 26.3749,
      "eval_samples_per_second": 7.242,
      "eval_steps_per_second": 0.227,
      "step": 104
    },
    {
      "epoch": 7.7,
      "step": 104,
      "total_flos": 1.0251773186064077e+18,
      "train_loss": 0.5548124955250666,
      "train_runtime": 4572.4853,
      "train_samples_per_second": 3.004,
      "train_steps_per_second": 0.023
    }
  ],
  "logging_steps": 8,
  "max_steps": 104,
  "num_train_epochs": 8,
  "save_steps": 500,
  "total_flos": 1.0251773186064077e+18,
  "trial_name": null,
  "trial_params": null
}