checkpoint-300/trainer_state.json

{
  "best_metric": 1.147840142250061,
  "best_model_checkpoint": "./mistral/10-04-24-Weni-WeniGPT-Agents-Mistral-1.0.0-SFT_Experiment with SFT and a new tokenizer configuration for chat template of mistral-2_max_steps-312_batch_4_2024-04-10_ppid_9/checkpoint-100",
  "epoch": 2.8708133971291865,
  "eval_steps": 50,
  "global_step": 300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.19,
      "grad_norm": 0.6767070889472961,
      "learning_rate": 0.0001993503206718859,
      "loss": 1.421,
      "step": 20
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7594590783119202,
      "learning_rate": 0.00019487887022684336,
      "loss": 1.1094,
      "step": 40
    },
    {
      "epoch": 0.48,
      "eval_loss": 1.1684353351593018,
      "eval_runtime": 15.9813,
      "eval_samples_per_second": 2.878,
      "eval_steps_per_second": 1.439,
      "step": 50
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8954619765281677,
      "learning_rate": 0.00018634217048966637,
      "loss": 1.0392,
      "step": 60
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6521860361099243,
      "learning_rate": 0.0001741059911251997,
      "loss": 0.9187,
      "step": 80
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.869737446308136,
      "learning_rate": 0.0001586946124013354,
      "loss": 0.879,
      "step": 100
    },
    {
      "epoch": 0.96,
      "eval_loss": 1.147840142250061,
      "eval_runtime": 15.9788,
      "eval_samples_per_second": 2.879,
      "eval_steps_per_second": 1.439,
      "step": 100
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7937932014465332,
      "learning_rate": 0.00014076836149416887,
      "loss": 0.6586,
      "step": 120
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0015738010406494,
      "learning_rate": 0.00012109531962807332,
      "loss": 0.6885,
      "step": 140
    },
    {
      "epoch": 1.44,
      "eval_loss": 1.1584892272949219,
      "eval_runtime": 15.9777,
      "eval_samples_per_second": 2.879,
      "eval_steps_per_second": 1.44,
      "step": 150
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8210965991020203,
      "learning_rate": 0.00010051841230721065,
      "loss": 0.6127,
      "step": 160
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9020585417747498,
      "learning_rate": 7.991929271442817e-05,
      "loss": 0.4806,
      "step": 180
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1187869310379028,
      "learning_rate": 6.018056575578075e-05,
      "loss": 0.5073,
      "step": 200
    },
    {
      "epoch": 1.91,
      "eval_loss": 1.1682521104812622,
      "eval_runtime": 15.98,
      "eval_samples_per_second": 2.879,
      "eval_steps_per_second": 1.439,
      "step": 200
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.101958990097046,
      "learning_rate": 4.2147971326825966e-05,
      "loss": 0.4016,
      "step": 220
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.0914422273635864,
      "learning_rate": 2.659414712405398e-05,
      "loss": 0.3815,
      "step": 240
    },
    {
      "epoch": 2.39,
      "eval_loss": 1.2108865976333618,
      "eval_runtime": 15.9816,
      "eval_samples_per_second": 2.878,
      "eval_steps_per_second": 1.439,
      "step": 250
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.116877555847168,
      "learning_rate": 1.4185523646469822e-05,
      "loss": 0.3674,
      "step": 260
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0452585220336914,
      "learning_rate": 5.453769828241872e-06,
      "loss": 0.3678,
      "step": 280
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9877675771713257,
      "learning_rate": 7.730127636723539e-07,
      "loss": 0.2894,
      "step": 300
    },
    {
      "epoch": 2.87,
      "eval_loss": 1.207001805305481,
      "eval_runtime": 15.9786,
      "eval_samples_per_second": 2.879,
      "eval_steps_per_second": 1.439,
      "step": 300
    }
  ],
  "logging_steps": 20,
  "max_steps": 312,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 1.5078115136355533e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}