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": -2.744870185852051,
      "logits/rejected": -2.706969738006592,
      "logps/chosen": -220.617919921875,
      "logps/pi_response": -125.63188934326172,
      "logps/ref_response": -125.32001495361328,
      "logps/rejected": -271.0539245605469,
      "loss": 0.685,
      "rewards/accuracies": 0.5763888955116272,
      "rewards/chosen": -0.025511225685477257,
      "rewards/margins": 0.025097062811255455,
      "rewards/rejected": -0.05060828477144241,
      "step": 10
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.187457503795526e-07,
      "logits/chosen": -2.7647249698638916,
      "logits/rejected": -2.7212321758270264,
      "logps/chosen": -259.55303955078125,
      "logps/pi_response": -120.53865051269531,
      "logps/ref_response": -122.2673110961914,
      "logps/rejected": -314.45928955078125,
      "loss": 0.6284,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.23639819025993347,
      "rewards/margins": 0.2823827862739563,
      "rewards/rejected": -0.5187809467315674,
      "step": 20
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8691164100062034e-07,
      "logits/chosen": -2.743373394012451,
      "logits/rejected": -2.7053332328796387,
      "logps/chosen": -287.496337890625,
      "logps/pi_response": -147.43978881835938,
      "logps/ref_response": -124.40946197509766,
      "logps/rejected": -358.22222900390625,
      "loss": 0.5878,
      "rewards/accuracies": 0.706250011920929,
      "rewards/chosen": -0.5064767599105835,
      "rewards/margins": 0.4633422791957855,
      "rewards/rejected": -0.9698190689086914,
      "step": 30
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.4248369943086995e-07,
      "logits/chosen": -2.6640450954437256,
      "logits/rejected": -2.628188371658325,
      "logps/chosen": -285.31231689453125,
      "logps/pi_response": -154.46168518066406,
      "logps/ref_response": -118.28387451171875,
      "logps/rejected": -360.1282958984375,
      "loss": 0.565,
      "rewards/accuracies": 0.659375011920929,
      "rewards/chosen": -0.5717625021934509,
      "rewards/margins": 0.48595350980758667,
      "rewards/rejected": -1.0577160120010376,
      "step": 40
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.473909705816111e-08,
      "logits/chosen": -2.641026020050049,
      "logits/rejected": -2.589219808578491,
      "logps/chosen": -289.4375,
      "logps/pi_response": -166.83139038085938,
      "logps/ref_response": -121.82298278808594,
      "logps/rejected": -362.04998779296875,
      "loss": 0.5603,
      "rewards/accuracies": 0.715624988079071,
      "rewards/chosen": -0.6004408597946167,
      "rewards/margins": 0.5109294652938843,
      "rewards/rejected": -1.111370325088501,
      "step": 50
    },
    {
      "epoch": 0.99,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.5961699728238381,
      "train_runtime": 3580.1113,
      "train_samples_per_second": 4.269,
      "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
}