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": -1.5793159008026123,
      "logits/rejected": -1.3809763193130493,
      "logps/chosen": -353.9909362792969,
      "logps/pi_response": -215.0353240966797,
      "logps/ref_response": -209.04591369628906,
      "logps/rejected": -626.6854858398438,
      "loss": 0.634,
      "rewards/accuracies": 0.6215277910232544,
      "rewards/chosen": -0.19814111292362213,
      "rewards/margins": 0.29235708713531494,
      "rewards/rejected": -0.49049821496009827,
      "step": 10
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.187457503795526e-07,
      "logits/chosen": -0.5755993127822876,
      "logits/rejected": -0.19093285501003265,
      "logps/chosen": -442.72906494140625,
      "logps/pi_response": -229.82565307617188,
      "logps/ref_response": -177.60939025878906,
      "logps/rejected": -871.7946166992188,
      "loss": 0.5897,
      "rewards/accuracies": 0.762499988079071,
      "rewards/chosen": -1.2860958576202393,
      "rewards/margins": 1.7290973663330078,
      "rewards/rejected": -3.015193462371826,
      "step": 20
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8691164100062034e-07,
      "logits/chosen": -0.4923222064971924,
      "logits/rejected": 0.08401201665401459,
      "logps/chosen": -401.0098571777344,
      "logps/pi_response": -204.19842529296875,
      "logps/ref_response": -172.53121948242188,
      "logps/rejected": -765.5057373046875,
      "loss": 0.4652,
      "rewards/accuracies": 0.784375011920929,
      "rewards/chosen": -0.8636873364448547,
      "rewards/margins": 1.3079454898834229,
      "rewards/rejected": -2.171632766723633,
      "step": 30
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.4248369943086995e-07,
      "logits/chosen": 0.057084761559963226,
      "logits/rejected": 0.7233976125717163,
      "logps/chosen": -394.339111328125,
      "logps/pi_response": -229.233642578125,
      "logps/ref_response": -169.03140258789062,
      "logps/rejected": -873.4837646484375,
      "loss": 0.4326,
      "rewards/accuracies": 0.8062499761581421,
      "rewards/chosen": -1.0300148725509644,
      "rewards/margins": 1.727870225906372,
      "rewards/rejected": -2.757884979248047,
      "step": 40
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.473909705816111e-08,
      "logits/chosen": 0.40286731719970703,
      "logits/rejected": 1.1685023307800293,
      "logps/chosen": -439.542724609375,
      "logps/pi_response": -257.29583740234375,
      "logps/ref_response": -178.077880859375,
      "logps/rejected": -824.6487426757812,
      "loss": 0.438,
      "rewards/accuracies": 0.768750011920929,
      "rewards/chosen": -1.21076238155365,
      "rewards/margins": 1.5581530332565308,
      "rewards/rejected": -2.7689156532287598,
      "step": 50
    },
    {
      "epoch": 0.99,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.500854310342821,
      "train_runtime": 3404.3015,
      "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
}