trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 10,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.2,
      "grad_norm": 0.0990709736943245,
      "learning_rate": 4e-05,
      "loss": 0.711,
      "step": 1
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.120853953063488,
      "learning_rate": 8e-05,
      "loss": 0.7668,
      "step": 2
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.13565421104431152,
      "learning_rate": 0.00012,
      "loss": 0.9731,
      "step": 3
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.14744164049625397,
      "learning_rate": 0.00016,
      "loss": 0.8388,
      "step": 4
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.16943834722042084,
      "learning_rate": 0.0002,
      "loss": 0.8452,
      "step": 5
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.09526122361421585,
      "learning_rate": 0.00016,
      "loss": 0.6138,
      "step": 6
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.11567643284797668,
      "learning_rate": 0.00012,
      "loss": 0.5858,
      "step": 7
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.20003917813301086,
      "learning_rate": 8e-05,
      "loss": 0.7195,
      "step": 8
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.4086407721042633,
      "learning_rate": 4e-05,
      "loss": 1.0479,
      "step": 9
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.2869514226913452,
      "learning_rate": 0.0,
      "loss": 0.9978,
      "step": 10
    },
    {
      "epoch": 2.0,
      "step": 10,
      "total_flos": 1341465195577344.0,
      "train_loss": 0.8099669218063354,
      "train_runtime": 334.6224,
      "train_samples_per_second": 0.233,
      "train_steps_per_second": 0.03
    }
  ],
  "logging_steps": 1,
  "max_steps": 10,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "total_flos": 1341465195577344.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}