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": 0.016736401673640166,
      "grad_norm": 11.912498683051679,
      "learning_rate": 8.333333333333333e-08,
      "logits/chosen": -2.8434743881225586,
      "logits/rejected": -2.769583225250244,
      "logps/chosen": -178.4530487060547,
      "logps/rejected": -233.52891540527344,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.16736401673640167,
      "grad_norm": 11.520133874238494,
      "learning_rate": 4.930057285201027e-07,
      "logits/chosen": -2.754981756210327,
      "logits/rejected": -2.7299110889434814,
      "logps/chosen": -225.8627471923828,
      "logps/rejected": -265.90814208984375,
      "loss": 0.6849,
      "rewards/accuracies": 0.5902777910232544,
      "rewards/chosen": -0.03168363496661186,
      "rewards/margins": 0.02158752828836441,
      "rewards/rejected": -0.053271159529685974,
      "step": 10
    },
    {
      "epoch": 0.33472803347280333,
      "grad_norm": 25.56430149750331,
      "learning_rate": 4.187457503795526e-07,
      "logits/chosen": -2.799773693084717,
      "logits/rejected": -2.7496700286865234,
      "logps/chosen": -259.5813293457031,
      "logps/rejected": -312.0262451171875,
      "loss": 0.6336,
      "rewards/accuracies": 0.643750011920929,
      "rewards/chosen": -0.2481800615787506,
      "rewards/margins": 0.2926359176635742,
      "rewards/rejected": -0.5408159494400024,
      "step": 20
    },
    {
      "epoch": 0.502092050209205,
      "grad_norm": 15.566903607908545,
      "learning_rate": 2.8691164100062034e-07,
      "logits/chosen": -2.7819409370422363,
      "logits/rejected": -2.750037431716919,
      "logps/chosen": -289.746826171875,
      "logps/rejected": -358.7594299316406,
      "loss": 0.5846,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": -0.41477519273757935,
      "rewards/margins": 0.5125462412834167,
      "rewards/rejected": -0.9273213148117065,
      "step": 30
    },
    {
      "epoch": 0.6694560669456067,
      "grad_norm": 18.95635048628815,
      "learning_rate": 1.4248369943086995e-07,
      "logits/chosen": -2.6801064014434814,
      "logits/rejected": -2.6403772830963135,
      "logps/chosen": -270.6723937988281,
      "logps/rejected": -362.1679992675781,
      "loss": 0.5577,
      "rewards/accuracies": 0.706250011920929,
      "rewards/chosen": -0.5635426640510559,
      "rewards/margins": 0.5637884736061096,
      "rewards/rejected": -1.127331018447876,
      "step": 40
    },
    {
      "epoch": 0.8368200836820083,
      "grad_norm": 20.509748331156263,
      "learning_rate": 3.473909705816111e-08,
      "logits/chosen": -2.7062981128692627,
      "logits/rejected": -2.653395175933838,
      "logps/chosen": -261.1449279785156,
      "logps/rejected": -334.8601379394531,
      "loss": 0.563,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": -0.5345139503479004,
      "rewards/margins": 0.5169811844825745,
      "rewards/rejected": -1.0514951944351196,
      "step": 50
    },
    {
      "epoch": 0.9874476987447699,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.5989973585484392,
      "train_runtime": 1077.191,
      "train_samples_per_second": 14.188,
      "train_steps_per_second": 0.055
    }
  ],
  "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": 4,
  "trial_name": null,
  "trial_params": null
}