trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.985781990521327,
  "eval_steps": 100,
  "global_step": 52,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 132.98895459124606,
      "learning_rate": 8.333333333333333e-08,
      "logits/chosen": 123.11854553222656,
      "logits/rejected": 97.00198364257812,
      "logps/chosen": -425.18585205078125,
      "logps/rejected": -424.1869201660156,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.19,
      "grad_norm": 170.32108935367978,
      "learning_rate": 4.907293218369498e-07,
      "logits/chosen": 117.15108489990234,
      "logits/rejected": 135.9984588623047,
      "logps/chosen": -441.28948974609375,
      "logps/rejected": -524.968505859375,
      "loss": 0.707,
      "rewards/accuracies": 0.5347222089767456,
      "rewards/chosen": 0.10494841635227203,
      "rewards/margins": 0.14762566983699799,
      "rewards/rejected": -0.04267726466059685,
      "step": 10
    },
    {
      "epoch": 0.38,
      "grad_norm": 101.98831849682922,
      "learning_rate": 3.941700805287168e-07,
      "logits/chosen": 125.30863952636719,
      "logits/rejected": 132.74871826171875,
      "logps/chosen": -423.9988708496094,
      "logps/rejected": -499.4864196777344,
      "loss": 0.5939,
      "rewards/accuracies": 0.6625000238418579,
      "rewards/chosen": 0.11328498274087906,
      "rewards/margins": 0.7120789289474487,
      "rewards/rejected": -0.5987939238548279,
      "step": 20
    },
    {
      "epoch": 0.57,
      "grad_norm": 94.48847661999667,
      "learning_rate": 2.3293939665883228e-07,
      "logits/chosen": 123.69091796875,
      "logits/rejected": 127.02400207519531,
      "logps/chosen": -473.69482421875,
      "logps/rejected": -546.0104370117188,
      "loss": 0.5604,
      "rewards/accuracies": 0.7437499761581421,
      "rewards/chosen": -0.9287363290786743,
      "rewards/margins": 1.0110455751419067,
      "rewards/rejected": -1.9397817850112915,
      "step": 30
    },
    {
      "epoch": 0.76,
      "grad_norm": 108.45257424394471,
      "learning_rate": 7.936171419533652e-08,
      "logits/chosen": 124.60306549072266,
      "logits/rejected": 119.5762939453125,
      "logps/chosen": -499.6136779785156,
      "logps/rejected": -527.9450073242188,
      "loss": 0.5394,
      "rewards/accuracies": 0.737500011920929,
      "rewards/chosen": -1.3216161727905273,
      "rewards/margins": 0.8374040722846985,
      "rewards/rejected": -2.159019947052002,
      "step": 40
    },
    {
      "epoch": 0.95,
      "grad_norm": 116.27759622068336,
      "learning_rate": 2.328513490917311e-09,
      "logits/chosen": 133.5895233154297,
      "logits/rejected": 134.02474975585938,
      "logps/chosen": -508.62506103515625,
      "logps/rejected": -535.1971435546875,
      "loss": 0.5317,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": -1.2901136875152588,
      "rewards/margins": 0.9073255658149719,
      "rewards/rejected": -2.197439432144165,
      "step": 50
    },
    {
      "epoch": 0.99,
      "step": 52,
      "total_flos": 0.0,
      "train_loss": 0.5840168778712933,
      "train_runtime": 572.7421,
      "train_samples_per_second": 11.785,
      "train_steps_per_second": 0.091
    }
  ],
  "logging_steps": 10,
  "max_steps": 52,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}