checkpoint-100/trainer_state.json

{
  "best_metric": 0.3754348158836365,
  "best_model_checkpoint": "./zephyr/05-04-24-Weni-WeniGPT-Agents-Zephyr-1.0.1-KTO_testing kto dataset during training-3_max_steps-145_batch_16_2024-04-05_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": 7.029934406280518,
      "kl": 2.2542672157287598,
      "learning_rate": 0.00018,
      "logps/chosen": -356.6376037597656,
      "logps/rejected": -393.9584655761719,
      "loss": 0.4417,
      "rewards/chosen": -3.5771310329437256,
      "rewards/margins": 2.5301473140716553,
      "rewards/rejected": -6.391368865966797,
      "step": 20
    },
    {
      "epoch": 0.27,
      "grad_norm": 19.954599380493164,
      "kl": 6.632220268249512,
      "learning_rate": 0.0001542857142857143,
      "logps/chosen": -319.3048400878906,
      "logps/rejected": -348.5643615722656,
      "loss": 0.4315,
      "rewards/chosen": -1.7924585342407227,
      "rewards/margins": 1.6441251039505005,
      "rewards/rejected": -3.656928539276123,
      "step": 40
    },
    {
      "epoch": 0.34,
      "eval_kl": 32.934268951416016,
      "eval_logps/chosen": -264.7897033691406,
      "eval_logps/rejected": -267.6116027832031,
      "eval_loss": 0.4092177450656891,
      "eval_rewards/chosen": 5.2044806480407715,
      "eval_rewards/margins": 2.720277786254883,
      "eval_rewards/rejected": 2.5654892921447754,
      "eval_runtime": 139.3761,
      "eval_samples_per_second": 2.152,
      "eval_steps_per_second": 0.538,
      "step": 50
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.9014763832092285,
      "kl": 22.45343017578125,
      "learning_rate": 0.00012571428571428572,
      "logps/chosen": -290.8168029785156,
      "logps/rejected": -297.7256774902344,
      "loss": 0.4154,
      "rewards/chosen": 3.169417381286621,
      "rewards/margins": 1.9798123836517334,
      "rewards/rejected": 1.3037060499191284,
      "step": 60
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.188873291015625,
      "kl": 15.003771781921387,
      "learning_rate": 9.714285714285715e-05,
      "logps/chosen": -279.8699951171875,
      "logps/rejected": -332.38848876953125,
      "loss": 0.3492,
      "rewards/chosen": 1.9723680019378662,
      "rewards/margins": 4.276235103607178,
      "rewards/rejected": -2.454475164413452,
      "step": 80
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.462972640991211,
      "kl": 4.606294631958008,
      "learning_rate": 6.857142857142858e-05,
      "logps/chosen": -278.8154602050781,
      "logps/rejected": -356.94451904296875,
      "loss": 0.3678,
      "rewards/chosen": 1.00811767578125,
      "rewards/margins": 5.261899471282959,
      "rewards/rejected": -4.226025581359863,
      "step": 100
    },
    {
      "epoch": 0.68,
      "eval_kl": 1.3235526084899902,
      "eval_logps/chosen": -330.0657653808594,
      "eval_logps/rejected": -357.2244873046875,
      "eval_loss": 0.3754348158836365,
      "eval_rewards/chosen": -1.3231240510940552,
      "eval_rewards/margins": 5.1890459060668945,
      "eval_rewards/rejected": -6.395801067352295,
      "eval_runtime": 139.0506,
      "eval_samples_per_second": 2.157,
      "eval_steps_per_second": 0.539,
      "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
}