trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9905956112852664,
  "eval_steps": 500,
  "global_step": 79,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 6.25e-08,
      "logits/chosen": -2.4858083724975586,
      "logits/rejected": -2.4783236980438232,
      "logps/chosen": -247.6826629638672,
      "logps/pi_response": -120.99482727050781,
      "logps/ref_response": -120.99482727050781,
      "logps/rejected": -278.14642333984375,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.990217055187362e-07,
      "logits/chosen": -2.825566291809082,
      "logits/rejected": -2.781306028366089,
      "logps/chosen": -260.65032958984375,
      "logps/pi_response": -135.34906005859375,
      "logps/ref_response": -134.69039916992188,
      "logps/rejected": -295.4285583496094,
      "loss": 0.6826,
      "rewards/accuracies": 0.5486111044883728,
      "rewards/chosen": -0.03667178004980087,
      "rewards/margins": 0.02899688109755516,
      "rewards/rejected": -0.06566866487264633,
      "step": 10
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.655786431300069e-07,
      "logits/chosen": -2.7526979446411133,
      "logits/rejected": -2.672511100769043,
      "logps/chosen": -268.38262939453125,
      "logps/pi_response": -126.11329650878906,
      "logps/ref_response": -123.22342681884766,
      "logps/rejected": -349.0810241699219,
      "loss": 0.6286,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": -0.3637709617614746,
      "rewards/margins": 0.3289268910884857,
      "rewards/rejected": -0.6926978230476379,
      "step": 20
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.9061232191019517e-07,
      "logits/chosen": -2.6593620777130127,
      "logits/rejected": -2.6028084754943848,
      "logps/chosen": -276.1861267089844,
      "logps/pi_response": -134.59718322753906,
      "logps/ref_response": -118.2535629272461,
      "logps/rejected": -357.50067138671875,
      "loss": 0.5811,
      "rewards/accuracies": 0.6937500238418579,
      "rewards/chosen": -0.4840686321258545,
      "rewards/margins": 0.473369300365448,
      "rewards/rejected": -0.9574378728866577,
      "step": 30
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8856223324132555e-07,
      "logits/chosen": -2.590123414993286,
      "logits/rejected": -2.516432523727417,
      "logps/chosen": -267.28839111328125,
      "logps/pi_response": -138.3324737548828,
      "logps/ref_response": -109.86214447021484,
      "logps/rejected": -358.03289794921875,
      "loss": 0.5589,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.5267149209976196,
      "rewards/margins": 0.49819517135620117,
      "rewards/rejected": -1.0249102115631104,
      "step": 40
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.7908455541642582e-07,
      "logits/chosen": -2.544290781021118,
      "logits/rejected": -2.5030927658081055,
      "logps/chosen": -292.77142333984375,
      "logps/pi_response": -157.22592163085938,
      "logps/ref_response": -112.8204116821289,
      "logps/rejected": -414.59100341796875,
      "loss": 0.5516,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": -0.6501742601394653,
      "rewards/margins": 0.6196854710578918,
      "rewards/rejected": -1.2698596715927124,
      "step": 50
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.32661172908373e-08,
      "logits/chosen": -2.5041542053222656,
      "logits/rejected": -2.484773874282837,
      "logps/chosen": -294.4966125488281,
      "logps/pi_response": -177.20556640625,
      "logps/ref_response": -120.77708435058594,
      "logps/rejected": -416.658447265625,
      "loss": 0.5251,
      "rewards/accuracies": 0.731249988079071,
      "rewards/chosen": -0.7415328025817871,
      "rewards/margins": 0.6046788096427917,
      "rewards/rejected": -1.3462116718292236,
      "step": 60
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.956279997278043e-08,
      "logits/chosen": -2.541991710662842,
      "logits/rejected": -2.4922404289245605,
      "logps/chosen": -307.61663818359375,
      "logps/pi_response": -182.88406372070312,
      "logps/ref_response": -125.622802734375,
      "logps/rejected": -445.57928466796875,
      "loss": 0.5161,
      "rewards/accuracies": 0.762499988079071,
      "rewards/chosen": -0.7298299670219421,
      "rewards/margins": 0.7636411190032959,
      "rewards/rejected": -1.4934711456298828,
      "step": 70
    },
    {
      "epoch": 0.99,
      "step": 79,
      "total_flos": 0.0,
      "train_loss": 0.5676172232326073,
      "train_runtime": 4739.1874,
      "train_samples_per_second": 4.3,
      "train_steps_per_second": 0.017
    }
  ],
  "logging_steps": 10,
  "max_steps": 79,
  "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
}