trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9874476987447699,
  "eval_steps": 500,
  "global_step": 59,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 8.333333333333333e-08,
      "logits/chosen": -2.828038215637207,
      "logits/rejected": -2.728235960006714,
      "logps/chosen": -198.4669189453125,
      "logps/pi_response": -116.54592895507812,
      "logps/ref_response": -116.54592895507812,
      "logps/rejected": -205.3916015625,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.930057285201027e-07,
      "logits/chosen": -2.744847297668457,
      "logits/rejected": -2.7069268226623535,
      "logps/chosen": -220.64466857910156,
      "logps/pi_response": -125.66951751708984,
      "logps/ref_response": -125.32001495361328,
      "logps/rejected": -271.061279296875,
      "loss": 0.685,
      "rewards/accuracies": 0.5833333134651184,
      "rewards/chosen": -0.02577850967645645,
      "rewards/margins": 0.024903353303670883,
      "rewards/rejected": -0.050681862980127335,
      "step": 10
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.187457503795526e-07,
      "logits/chosen": -2.7635605335235596,
      "logits/rejected": -2.719918727874756,
      "logps/chosen": -259.6158142089844,
      "logps/pi_response": -120.7113037109375,
      "logps/ref_response": -122.2673110961914,
      "logps/rejected": -314.64898681640625,
      "loss": 0.6284,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.23702558875083923,
      "rewards/margins": 0.28365251421928406,
      "rewards/rejected": -0.5206781625747681,
      "step": 20
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8691164100062034e-07,
      "logits/chosen": -2.7415413856506348,
      "logits/rejected": -2.70352840423584,
      "logps/chosen": -288.61761474609375,
      "logps/pi_response": -148.96580505371094,
      "logps/ref_response": -124.40946197509766,
      "logps/rejected": -359.23297119140625,
      "loss": 0.5876,
      "rewards/accuracies": 0.706250011920929,
      "rewards/chosen": -0.5176891684532166,
      "rewards/margins": 0.4622374176979065,
      "rewards/rejected": -0.9799267053604126,
      "step": 30
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.4248369943086995e-07,
      "logits/chosen": -2.661461353302002,
      "logits/rejected": -2.625527858734131,
      "logps/chosen": -286.5964660644531,
      "logps/pi_response": -156.09762573242188,
      "logps/ref_response": -118.28387451171875,
      "logps/rejected": -361.10809326171875,
      "loss": 0.5654,
      "rewards/accuracies": 0.653124988079071,
      "rewards/chosen": -0.5846043229103088,
      "rewards/margins": 0.482909619808197,
      "rewards/rejected": -1.0675138235092163,
      "step": 40
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.473909705816111e-08,
      "logits/chosen": -2.6372907161712646,
      "logits/rejected": -2.5853748321533203,
      "logps/chosen": -291.3328857421875,
      "logps/pi_response": -168.8394317626953,
      "logps/ref_response": -121.82298278808594,
      "logps/rejected": -364.73297119140625,
      "loss": 0.559,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6193948984146118,
      "rewards/margins": 0.5188055038452148,
      "rewards/rejected": -1.1382004022598267,
      "step": 50
    },
    {
      "epoch": 0.99,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.5958063157938295,
      "train_runtime": 3589.2721,
      "train_samples_per_second": 4.258,
      "train_steps_per_second": 0.016
    }
  ],
  "logging_steps": 10,
  "max_steps": 59,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}