checkpoint-1200/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4997917534360683,
  "eval_steps": 240,
  "global_step": 1200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00041649312786339027,
      "grad_norm": 7.0,
      "learning_rate": 2e-06,
      "loss": 0.7314,
      "step": 1
    },
    {
      "epoch": 0.04164931278633902,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0002,
      "loss": 0.377,
      "step": 100
    },
    {
      "epoch": 0.08329862557267805,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0004,
      "loss": 0.2401,
      "step": 200
    },
    {
      "epoch": 0.09995835068721366,
      "eval_peoplespeech-clean-transcription_loss": 3.7064812183380127,
      "eval_peoplespeech-clean-transcription_model_preparation_time": 0.0062,
      "eval_peoplespeech-clean-transcription_runtime": 14.4755,
      "eval_peoplespeech-clean-transcription_samples_per_second": 4.421,
      "eval_peoplespeech-clean-transcription_steps_per_second": 0.069,
      "step": 240
    },
    {
      "epoch": 0.12494793835901707,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0006,
      "loss": 0.209,
      "step": 300
    },
    {
      "epoch": 0.1665972511453561,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0008,
      "loss": 0.1586,
      "step": 400
    },
    {
      "epoch": 0.19991670137442732,
      "eval_peoplespeech-clean-transcription_loss": 2.0534801483154297,
      "eval_peoplespeech-clean-transcription_model_preparation_time": 0.0062,
      "eval_peoplespeech-clean-transcription_runtime": 13.9793,
      "eval_peoplespeech-clean-transcription_samples_per_second": 4.578,
      "eval_peoplespeech-clean-transcription_steps_per_second": 0.072,
      "step": 480
    },
    {
      "epoch": 0.20824656393169513,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.001,
      "loss": 0.1178,
      "step": 500
    },
    {
      "epoch": 0.24989587671803415,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0012,
      "loss": 0.1031,
      "step": 600
    },
    {
      "epoch": 0.2915451895043732,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0014,
      "loss": 0.0942,
      "step": 700
    },
    {
      "epoch": 0.299875052061641,
      "eval_peoplespeech-clean-transcription_loss": 1.931348443031311,
      "eval_peoplespeech-clean-transcription_model_preparation_time": 0.0062,
      "eval_peoplespeech-clean-transcription_runtime": 14.1549,
      "eval_peoplespeech-clean-transcription_samples_per_second": 4.521,
      "eval_peoplespeech-clean-transcription_steps_per_second": 0.071,
      "step": 720
    },
    {
      "epoch": 0.3331945022907122,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0016,
      "loss": 0.0885,
      "step": 800
    },
    {
      "epoch": 0.3748438150770512,
      "grad_norm": 0.035888671875,
      "learning_rate": 0.0018000000000000002,
      "loss": 0.0853,
      "step": 900
    },
    {
      "epoch": 0.39983340274885465,
      "eval_peoplespeech-clean-transcription_loss": 1.8942928314208984,
      "eval_peoplespeech-clean-transcription_model_preparation_time": 0.0062,
      "eval_peoplespeech-clean-transcription_runtime": 13.7985,
      "eval_peoplespeech-clean-transcription_samples_per_second": 4.638,
      "eval_peoplespeech-clean-transcription_steps_per_second": 0.072,
      "step": 960
    },
    {
      "epoch": 0.41649312786339027,
      "grad_norm": 0.033203125,
      "learning_rate": 0.002,
      "loss": 0.0817,
      "step": 1000
    },
    {
      "epoch": 0.45814244064972925,
      "grad_norm": 0.03173828125,
      "learning_rate": 0.001996926043706003,
      "loss": 0.0798,
      "step": 1100
    },
    {
      "epoch": 0.4997917534360683,
      "grad_norm": 0.029296875,
      "learning_rate": 0.0019877251730624503,
      "loss": 0.0777,
      "step": 1200
    },
    {
      "epoch": 0.4997917534360683,
      "eval_peoplespeech-clean-transcription_loss": 1.863105297088623,
      "eval_peoplespeech-clean-transcription_model_preparation_time": 0.0062,
      "eval_peoplespeech-clean-transcription_runtime": 13.1079,
      "eval_peoplespeech-clean-transcription_samples_per_second": 4.883,
      "eval_peoplespeech-clean-transcription_steps_per_second": 0.076,
      "step": 1200
    }
  ],
  "logging_steps": 100,
  "max_steps": 4800,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 1200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.7954024695099556e+18,
  "train_batch_size": 576,
  "trial_name": null,
  "trial_params": null
}