checkpoint-100/trainer_state.json

{
  "best_metric": 0.47333332896232605,
  "best_model_checkpoint": "./zephyr/06-04-24-Weni-WeniGPT-Agents-Zephyr-1.0.1-KTO_testing kto dataset during training-3_max_steps-145_batch_16_2024-04-06_ppid_9/checkpoint-100",
  "epoch": 0.684931506849315,
  "eval_steps": 50,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.14,
      "grad_norm": 34.550357818603516,
      "kl": 40.30364227294922,
      "learning_rate": 0.00018571428571428572,
      "logps/chosen": -341.70428466796875,
      "logps/rejected": -376.15447998046875,
      "loss": 0.4341,
      "rewards/chosen": 4.142109394073486,
      "rewards/margins": 5.559747695922852,
      "rewards/rejected": -1.2980353832244873,
      "step": 20
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.0,
      "kl": 9.54911994934082,
      "learning_rate": 0.00015857142857142857,
      "logps/chosen": -2304.16552734375,
      "logps/rejected": -2427.63427734375,
      "loss": 0.5152,
      "rewards/chosen": -194.421875,
      "rewards/margins": 7.0421857833862305,
      "rewards/rejected": -205.4108428955078,
      "step": 40
    },
    {
      "epoch": 0.34,
      "eval_kl": 0.0,
      "eval_logps/chosen": -3220.12841796875,
      "eval_logps/rejected": -2905.742919921875,
      "eval_loss": 0.47333332896232605,
      "eval_rewards/chosen": -289.121337890625,
      "eval_rewards/margins": -35.262752532958984,
      "eval_rewards/rejected": -256.6285705566406,
      "eval_runtime": 138.045,
      "eval_samples_per_second": 2.173,
      "eval_steps_per_second": 0.543,
      "step": 50
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 0.00013000000000000002,
      "logps/chosen": -3215.18505859375,
      "logps/rejected": -3101.622802734375,
      "loss": 0.4344,
      "rewards/chosen": -290.7165222167969,
      "rewards/margins": -12.484609603881836,
      "rewards/rejected": -276.9840393066406,
      "step": 60
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 0.00010142857142857143,
      "logps/chosen": -3785.22607421875,
      "logps/rejected": -3411.793701171875,
      "loss": 0.4781,
      "rewards/chosen": -344.2449951171875,
      "rewards/margins": -43.561405181884766,
      "rewards/rejected": -306.4319152832031,
      "step": 80
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 7.285714285714286e-05,
      "logps/chosen": -3814.2275390625,
      "logps/rejected": -3579.71630859375,
      "loss": 0.4344,
      "rewards/chosen": -343.0238952636719,
      "rewards/margins": -25.151195526123047,
      "rewards/rejected": -323.59967041015625,
      "step": 100
    },
    {
      "epoch": 0.68,
      "eval_kl": 0.0,
      "eval_logps/chosen": -3431.77734375,
      "eval_logps/rejected": -3077.673583984375,
      "eval_loss": 0.47333332896232605,
      "eval_rewards/chosen": -310.2862854003906,
      "eval_rewards/margins": -38.59312438964844,
      "eval_rewards/rejected": -273.8216552734375,
      "eval_runtime": 138.0016,
      "eval_samples_per_second": 2.174,
      "eval_steps_per_second": 0.543,
      "step": 100
    }
  ],
  "logging_steps": 20,
  "max_steps": 145,
  "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
}