trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.888888888888889,
  "eval_steps": 50,
  "global_step": 39,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.37037037037037035,
      "grad_norm": 70.16926770164845,
      "learning_rate": 5e-07,
      "logits/chosen": -2.7230114936828613,
      "logits/rejected": -2.7195019721984863,
      "logps/chosen": -305.3332824707031,
      "logps/rejected": -237.0945281982422,
      "loss": 0.6888,
      "rewards/accuracies": 0.41874998807907104,
      "rewards/chosen": 0.02525922656059265,
      "rewards/margins": 0.01608438417315483,
      "rewards/rejected": 0.009174839593470097,
      "step": 5
    },
    {
      "epoch": 0.7407407407407407,
      "grad_norm": 74.09095204023392,
      "learning_rate": 1e-06,
      "logits/chosen": -2.690033435821533,
      "logits/rejected": -2.684277057647705,
      "logps/chosen": -300.15606689453125,
      "logps/rejected": -242.4936981201172,
      "loss": 0.6262,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": 0.6478828191757202,
      "rewards/margins": 0.29922136664390564,
      "rewards/rejected": 0.3486614227294922,
      "step": 10
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 31.36667250284913,
      "learning_rate": 9.284285880837946e-07,
      "logits/chosen": -2.531611680984497,
      "logits/rejected": -2.559107542037964,
      "logps/chosen": -275.9391784667969,
      "logps/rejected": -219.6202850341797,
      "loss": 0.5025,
      "rewards/accuracies": 0.7875000238418579,
      "rewards/chosen": 2.089695453643799,
      "rewards/margins": 1.2278673648834229,
      "rewards/rejected": 0.8618279695510864,
      "step": 15
    },
    {
      "epoch": 1.4814814814814814,
      "grad_norm": 31.816132043252477,
      "learning_rate": 7.342042203498951e-07,
      "logits/chosen": -2.470438003540039,
      "logits/rejected": -2.4644124507904053,
      "logps/chosen": -283.08331298828125,
      "logps/rejected": -224.30355834960938,
      "loss": 0.2912,
      "rewards/accuracies": 0.875,
      "rewards/chosen": 2.9861273765563965,
      "rewards/margins": 2.8242931365966797,
      "rewards/rejected": 0.16183419525623322,
      "step": 20
    },
    {
      "epoch": 1.8518518518518519,
      "grad_norm": 28.50967013170727,
      "learning_rate": 4.7293054570729126e-07,
      "logits/chosen": -2.4437544345855713,
      "logits/rejected": -2.4410457611083984,
      "logps/chosen": -279.73944091796875,
      "logps/rejected": -233.590576171875,
      "loss": 0.2731,
      "rewards/accuracies": 0.887499988079071,
      "rewards/chosen": 3.013139486312866,
      "rewards/margins": 3.5092759132385254,
      "rewards/rejected": -0.4961363673210144,
      "step": 25
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 17.254950263421055,
      "learning_rate": 2.1940646731880885e-07,
      "logits/chosen": -2.4052348136901855,
      "logits/rejected": -2.3916208744049072,
      "logps/chosen": -267.53753662109375,
      "logps/rejected": -228.9844207763672,
      "loss": 0.1943,
      "rewards/accuracies": 0.9437500238418579,
      "rewards/chosen": 2.7877602577209473,
      "rewards/margins": 3.3208701610565186,
      "rewards/rejected": -0.5331100225448608,
      "step": 30
    },
    {
      "epoch": 2.5925925925925926,
      "grad_norm": 21.18382552424417,
      "learning_rate": 4.621229016452155e-08,
      "logits/chosen": -2.4393057823181152,
      "logits/rejected": -2.405693292617798,
      "logps/chosen": -260.98291015625,
      "logps/rejected": -263.63232421875,
      "loss": 0.1714,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": 3.206425905227661,
      "rewards/margins": 4.12821626663208,
      "rewards/rejected": -0.9217900037765503,
      "step": 35
    },
    {
      "epoch": 2.888888888888889,
      "step": 39,
      "total_flos": 459455468666880.0,
      "train_loss": 0.36858611076306075,
      "train_runtime": 1333.6786,
      "train_samples_per_second": 7.691,
      "train_steps_per_second": 0.029
    }
  ],
  "logging_steps": 5,
  "max_steps": 39,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 459455468666880.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}