checkpoint-200/trainer_state.json

{
  "best_metric": 0.5000000596046448,
  "best_model_checkpoint": "./zephyr/10-04-24-Weni-WeniGPT-Agents-Zephyr-1.0.25-KTO_Experiment with a new tokenizer configuration for chat template of zephyr-2_max_steps-1470_batch_16_2024-04-10_ppid_9/checkpoint-100",
  "epoch": 1.3201320132013201,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.13,
      "grad_norm": 57.293792724609375,
      "kl": 0.03853478282690048,
      "learning_rate": 6.222222222222222e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.7078,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 20
    },
    {
      "epoch": 0.26,
      "grad_norm": 112.50944519042969,
      "kl": 3.2648494243621826,
      "learning_rate": 0.00014666666666666666,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.6966,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 40
    },
    {
      "epoch": 0.33,
      "eval_kl": 0.0,
      "eval_logps/chosen": -413.6161193847656,
      "eval_logps/rejected": -362.2559509277344,
      "eval_loss": 0.5063381791114807,
      "eval_rewards/chosen": -13.412939071655273,
      "eval_rewards/margins": -1.0048810243606567,
      "eval_rewards/rejected": -12.408059120178223,
      "eval_runtime": 170.1826,
      "eval_samples_per_second": 2.057,
      "eval_steps_per_second": 0.517,
      "step": 50
    },
    {
      "epoch": 0.4,
      "grad_norm": 19.94582748413086,
      "kl": 0.45922356843948364,
      "learning_rate": 0.00019887719298245616,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.5743,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 60
    },
    {
      "epoch": 0.53,
      "grad_norm": 79.92957305908203,
      "kl": 0.0,
      "learning_rate": 0.0001960701754385965,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.6108,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 80
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.06103940308094025,
      "kl": 0.0,
      "learning_rate": 0.00019326315789473686,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.754,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 100
    },
    {
      "epoch": 0.66,
      "eval_kl": 0.0,
      "eval_logps/chosen": -2027.0018310546875,
      "eval_logps/rejected": -1697.82177734375,
      "eval_loss": 0.5000000596046448,
      "eval_rewards/chosen": -174.75149536132812,
      "eval_rewards/margins": -28.786863327026367,
      "eval_rewards/rejected": -145.96463012695312,
      "eval_runtime": 170.0562,
      "eval_samples_per_second": 2.058,
      "eval_steps_per_second": 0.517,
      "step": 100
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 0.0001904561403508772,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.95,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 120
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 0.00018764912280701756,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.6274,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 140
    },
    {
      "epoch": 0.99,
      "eval_kl": 0.0,
      "eval_logps/chosen": -2237.81494140625,
      "eval_logps/rejected": -1889.774169921875,
      "eval_loss": 0.5,
      "eval_rewards/chosen": -195.83285522460938,
      "eval_rewards/margins": -30.672954559326172,
      "eval_rewards/rejected": -165.15989685058594,
      "eval_runtime": 169.8795,
      "eval_samples_per_second": 2.06,
      "eval_steps_per_second": 0.518,
      "step": 150
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 0.0001848421052631579,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.6387,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 160
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 0.00018203508771929826,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.8327,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 180
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 0.00017922807017543862,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.642,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 200
    },
    {
      "epoch": 1.32,
      "eval_kl": 0.0,
      "eval_logps/chosen": -2230.916259765625,
      "eval_logps/rejected": -1884.9520263671875,
      "eval_loss": 0.5000000596046448,
      "eval_rewards/chosen": -195.14297485351562,
      "eval_rewards/margins": -30.465293884277344,
      "eval_rewards/rejected": -164.67767333984375,
      "eval_runtime": 170.1489,
      "eval_samples_per_second": 2.057,
      "eval_steps_per_second": 0.517,
      "step": 200
    }
  ],
  "logging_steps": 20,
  "max_steps": 1470,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}