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": -1.778551697731018,
      "logits/rejected": -1.5408724546432495,
      "logps/chosen": -277.6687927246094,
      "logps/pi_response": -231.2880859375,
      "logps/ref_response": -231.2880859375,
      "logps/rejected": -499.2101135253906,
      "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": -1.5808309316635132,
      "logits/rejected": -1.382399320602417,
      "logps/chosen": -353.8753662109375,
      "logps/pi_response": -214.94216918945312,
      "logps/ref_response": -209.04591369628906,
      "logps/rejected": -626.6241455078125,
      "loss": 0.6343,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.19698534905910492,
      "rewards/margins": 0.29289910197257996,
      "rewards/rejected": -0.4898844361305237,
      "step": 10
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.187457503795526e-07,
      "logits/chosen": -0.6259251236915588,
      "logits/rejected": -0.24522070586681366,
      "logps/chosen": -437.9039001464844,
      "logps/pi_response": -225.865966796875,
      "logps/ref_response": -177.60939025878906,
      "logps/rejected": -864.3118286132812,
      "loss": 0.5899,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -1.237844467163086,
      "rewards/margins": 1.7025201320648193,
      "rewards/rejected": -2.940364360809326,
      "step": 20
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8691164100062034e-07,
      "logits/chosen": -0.6101164817810059,
      "logits/rejected": -0.05170084908604622,
      "logps/chosen": -392.63580322265625,
      "logps/pi_response": -197.5888671875,
      "logps/ref_response": -172.53121948242188,
      "logps/rejected": -751.2313842773438,
      "loss": 0.4697,
      "rewards/accuracies": 0.778124988079071,
      "rewards/chosen": -0.7799466848373413,
      "rewards/margins": 1.2489432096481323,
      "rewards/rejected": -2.0288898944854736,
      "step": 30
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.4248369943086995e-07,
      "logits/chosen": -0.06927945464849472,
      "logits/rejected": 0.5743276476860046,
      "logps/chosen": -373.8592224121094,
      "logps/pi_response": -210.5477294921875,
      "logps/ref_response": -169.03140258789062,
      "logps/rejected": -838.1275634765625,
      "loss": 0.4394,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.8252162933349609,
      "rewards/margins": 1.579106092453003,
      "rewards/rejected": -2.404322385787964,
      "step": 40
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.473909705816111e-08,
      "logits/chosen": 0.34383702278137207,
      "logits/rejected": 1.0953922271728516,
      "logps/chosen": -422.825439453125,
      "logps/pi_response": -239.9983673095703,
      "logps/ref_response": -178.077880859375,
      "logps/rejected": -802.85791015625,
      "loss": 0.4416,
      "rewards/accuracies": 0.778124988079071,
      "rewards/chosen": -1.0435901880264282,
      "rewards/margins": 1.5074162483215332,
      "rewards/rejected": -2.55100679397583,
      "step": 50
    },
    {
      "epoch": 0.99,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.5042527813022419,
      "train_runtime": 3404.6409,
      "train_samples_per_second": 4.489,
      "train_steps_per_second": 0.017
    }
  ],
  "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
}