checkpoint-200/trainer_state.json

{
  "best_metric": 0.021165071055293083,
  "best_model_checkpoint": "./Zephyr/28-03-24-Weni-WeniGPT-QA-Zephyr-7B-5.0.0-KTO_WeniGPT Experiment using KTO trainer with no collator, Mixstral model and random system prompt.-2_max_steps-786_batch_32_2024-03-28_ppid_9/checkpoint-200",
  "epoch": 1.5209125475285172,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.15,
      "grad_norm": 3.2390341758728027,
      "kl": 1.4365953207015991,
      "learning_rate": 0.0001666666666666667,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 1.2323,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 20
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6226775050163269,
      "kl": 0.1346512734889984,
      "learning_rate": 0.00019580052493438322,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.1437,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 40
    },
    {
      "epoch": 0.38,
      "eval_kl": 0.0960722267627716,
      "eval_logps/chosen": -158.278564453125,
      "eval_logps/rejected": -400.1436767578125,
      "eval_loss": 0.028233768418431282,
      "eval_rewards/chosen": 5.2842488288879395,
      "eval_rewards/margins": 25.394336700439453,
      "eval_rewards/rejected": -20.11008644104004,
      "eval_runtime": 214.1148,
      "eval_samples_per_second": 2.232,
      "eval_steps_per_second": 0.56,
      "step": 50
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9644193649291992,
      "kl": 0.026479745283722878,
      "learning_rate": 0.0001905511811023622,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.1405,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 60
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.28626948595046997,
      "kl": 0.018679404631257057,
      "learning_rate": 0.00018530183727034121,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.1041,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 80
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.0979766920208931,
      "kl": 0.0,
      "learning_rate": 0.00018005249343832022,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0615,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 100
    },
    {
      "epoch": 0.76,
      "eval_kl": 0.5886130928993225,
      "eval_logps/chosen": -153.61863708496094,
      "eval_logps/rejected": -383.54296875,
      "eval_loss": 0.022206587716937065,
      "eval_rewards/chosen": 5.750241756439209,
      "eval_rewards/margins": 24.20025062561035,
      "eval_rewards/rejected": -18.450010299682617,
      "eval_runtime": 213.9917,
      "eval_samples_per_second": 2.234,
      "eval_steps_per_second": 0.561,
      "step": 100
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3924414813518524,
      "kl": 0.0,
      "learning_rate": 0.00017480314960629923,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0396,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 120
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.0917697548866272,
      "kl": 0.2697001099586487,
      "learning_rate": 0.00016955380577427821,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0346,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 140
    },
    {
      "epoch": 1.14,
      "eval_kl": 0.3035987615585327,
      "eval_logps/chosen": -162.28248596191406,
      "eval_logps/rejected": -612.7335205078125,
      "eval_loss": 0.03983871266245842,
      "eval_rewards/chosen": 4.883855819702148,
      "eval_rewards/margins": 46.25292205810547,
      "eval_rewards/rejected": -41.36906814575195,
      "eval_runtime": 214.0091,
      "eval_samples_per_second": 2.234,
      "eval_steps_per_second": 0.561,
      "step": 150
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.16112379729747772,
      "kl": 0.2598298192024231,
      "learning_rate": 0.00016430446194225722,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0437,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 160
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.04856234788894653,
      "kl": 0.09709737449884415,
      "learning_rate": 0.00015958005249343832,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0288,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 180
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.5220792889595032,
      "kl": 0.055810533463954926,
      "learning_rate": 0.00015433070866141733,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0563,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 200
    },
    {
      "epoch": 1.52,
      "eval_kl": 0.15839433670043945,
      "eval_logps/chosen": -149.37533569335938,
      "eval_logps/rejected": -463.8907165527344,
      "eval_loss": 0.021165071055293083,
      "eval_rewards/chosen": 6.174570083618164,
      "eval_rewards/margins": 32.65935134887695,
      "eval_rewards/rejected": -26.48478126525879,
      "eval_runtime": 214.0155,
      "eval_samples_per_second": 2.233,
      "eval_steps_per_second": 0.561,
      "step": 200
    }
  ],
  "logging_steps": 20,
  "max_steps": 786,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}