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": 73.53016894281072,
      "learning_rate": 1.6666666666666664e-08,
      "logits/chosen": -2.377516984939575,
      "logits/rejected": -2.342954158782959,
      "logps/chosen": -267.31927490234375,
      "logps/pi_response": -149.59169006347656,
      "logps/ref_response": -149.59169006347656,
      "logps/rejected": -539.9591674804688,
      "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": 80.4074524611749,
      "learning_rate": 9.860114570402053e-08,
      "logits/chosen": -2.3693184852600098,
      "logits/rejected": -2.286440372467041,
      "logps/chosen": -298.61932373046875,
      "logps/pi_response": -132.10630798339844,
      "logps/ref_response": -131.78355407714844,
      "logps/rejected": -554.6019897460938,
      "loss": 0.6859,
      "rewards/accuracies": 0.5555555820465088,
      "rewards/chosen": -0.00936645083129406,
      "rewards/margins": 0.014261982403695583,
      "rewards/rejected": -0.023628434166312218,
      "step": 10
    },
    {
      "epoch": 0.33472803347280333,
      "grad_norm": 46.77444881870846,
      "learning_rate": 8.374915007591053e-08,
      "logits/chosen": -2.3324151039123535,
      "logits/rejected": -2.2493882179260254,
      "logps/chosen": -298.06707763671875,
      "logps/pi_response": -146.1650390625,
      "logps/ref_response": -140.46481323242188,
      "logps/rejected": -580.0590209960938,
      "loss": 0.6135,
      "rewards/accuracies": 0.721875011920929,
      "rewards/chosen": -0.18148620426654816,
      "rewards/margins": 0.20174245536327362,
      "rewards/rejected": -0.3832286298274994,
      "step": 20
    },
    {
      "epoch": 0.502092050209205,
      "grad_norm": 42.64368867415699,
      "learning_rate": 5.738232820012406e-08,
      "logits/chosen": -2.2694995403289795,
      "logits/rejected": -2.207530975341797,
      "logps/chosen": -327.8448181152344,
      "logps/pi_response": -139.88900756835938,
      "logps/ref_response": -131.197509765625,
      "logps/rejected": -638.9929809570312,
      "loss": 0.545,
      "rewards/accuracies": 0.7718750238418579,
      "rewards/chosen": -0.371069997549057,
      "rewards/margins": 0.4864231050014496,
      "rewards/rejected": -0.8574931025505066,
      "step": 30
    },
    {
      "epoch": 0.6694560669456067,
      "grad_norm": 40.23447563253812,
      "learning_rate": 2.8496739886173992e-08,
      "logits/chosen": -2.242140054702759,
      "logits/rejected": -2.160322666168213,
      "logps/chosen": -358.05889892578125,
      "logps/pi_response": -157.74525451660156,
      "logps/ref_response": -145.66329956054688,
      "logps/rejected": -707.8968505859375,
      "loss": 0.5184,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.5855440497398376,
      "rewards/margins": 0.7927876114845276,
      "rewards/rejected": -1.3783317804336548,
      "step": 40
    },
    {
      "epoch": 0.8368200836820083,
      "grad_norm": 82.30441781177622,
      "learning_rate": 6.947819411632222e-09,
      "logits/chosen": -2.2379250526428223,
      "logits/rejected": -2.158501386642456,
      "logps/chosen": -376.53204345703125,
      "logps/pi_response": -153.90878295898438,
      "logps/ref_response": -141.55398559570312,
      "logps/rejected": -747.1853637695312,
      "loss": 0.4999,
      "rewards/accuracies": 0.7437499761581421,
      "rewards/chosen": -0.6662783622741699,
      "rewards/margins": 0.9346103668212891,
      "rewards/rejected": -1.6008888483047485,
      "step": 50
    },
    {
      "epoch": 0.9874476987447699,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.5589375010991501,
      "train_runtime": 2582.4535,
      "train_samples_per_second": 5.918,
      "train_steps_per_second": 0.023
    }
  ],
  "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
}