trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 6603,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.23,
      "learning_rate": 4.62138421929426e-05,
      "loss": 3.7153,
      "step": 500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.242768438588521e-05,
      "loss": 2.7198,
      "step": 1000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.864152657882781e-05,
      "loss": 2.4168,
      "step": 1500
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.485536877177041e-05,
      "loss": 2.243,
      "step": 2000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.106921096471301e-05,
      "loss": 2.0121,
      "step": 2500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7283053157655614e-05,
      "loss": 1.8587,
      "step": 3000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3496895350598214e-05,
      "loss": 1.791,
      "step": 3500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9710737543540817e-05,
      "loss": 1.7424,
      "step": 4000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.5924579736483417e-05,
      "loss": 1.6664,
      "step": 4500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2138421929426019e-05,
      "loss": 1.4809,
      "step": 5000
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.35226412236862e-06,
      "loss": 1.4406,
      "step": 5500
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.566106315311222e-06,
      "loss": 1.4447,
      "step": 6000
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.79948508253824e-07,
      "loss": 1.4293,
      "step": 6500
    },
    {
      "epoch": 3.0,
      "step": 6603,
      "total_flos": 1.0540038816989184e+16,
      "train_loss": 1.9876760678924648,
      "train_runtime": 2038.2062,
      "train_samples_per_second": 32.384,
      "train_steps_per_second": 3.24
    }
  ],
  "max_steps": 6603,
  "num_train_epochs": 3,
  "total_flos": 1.0540038816989184e+16,
  "trial_name": null,
  "trial_params": null
}