checkpoint-313/trainer_state.json

{
  "best_metric": 1.1438744068145752,
  "best_model_checkpoint": "./checkpoints/dpo-mix-7k/Qwen1.5-0.5B-dpo-mix-7k-lambda1.0-ORPO-29-9-49/checkpoint-313",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 313,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.16,
      "grad_norm": 616.0,
      "learning_rate": 2.5e-06,
      "loss": 15.2717,
      "step": 50
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.59375,
      "learning_rate": 5e-06,
      "loss": 4.0109,
      "step": 100
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.65625,
      "learning_rate": 4.889350784534168e-06,
      "loss": 1.2041,
      "step": 150
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.90625,
      "learning_rate": 4.5671977372432355e-06,
      "loss": 1.1597,
      "step": 200
    },
    {
      "epoch": 0.8,
      "grad_norm": 43.5,
      "learning_rate": 4.062057643681335e-06,
      "loss": 1.1542,
      "step": 250
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.8125,
      "learning_rate": 3.4186451878908393e-06,
      "loss": 1.1222,
      "step": 300
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.1438744068145752,
      "eval_runtime": 34.1359,
      "eval_samples_per_second": 16.551,
      "eval_steps_per_second": 1.055,
      "step": 313
    }
  ],
  "logging_steps": 50,
  "max_steps": 626,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 4744684177981440.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}