logging/checkpoint_perplexities.json

[
  {
    "checkpoint": "chck_10M",
    "epoch": 1,
    "global_step": 1600,
    "total_tokens": 13132619,
    "train_loss": 5.961990402844552,
    "val_loss": 7.128291858616372,
    "val_perplexity": 1246.7455293995722
  },
  {
    "checkpoint": "chck_20M",
    "epoch": 2,
    "global_step": 3200,
    "total_tokens": 26265238,
    "train_loss": 5.200628490074942,
    "val_loss": 6.857324192515093,
    "val_perplexity": 950.8194507421762
  },
  {
    "checkpoint": "chck_30M",
    "epoch": 3,
    "global_step": 4800,
    "total_tokens": 39397857,
    "train_loss": 5.0132061014271985,
    "val_loss": 6.694719708098798,
    "val_perplexity": 808.127390900081
  },
  {
    "checkpoint": "chck_40M",
    "epoch": 4,
    "global_step": 6400,
    "total_tokens": 52530476,
    "train_loss": 4.8878661164711295,
    "val_loss": 6.574928881067326,
    "val_perplexity": 716.8946377575272
  },
  {
    "checkpoint": "chck_50M",
    "epoch": 5,
    "global_step": 8000,
    "total_tokens": 65663095,
    "train_loss": 4.796031263309533,
    "val_loss": 6.49725505410517,
    "val_perplexity": 663.3183587886707
  },
  {
    "checkpoint": "chck_60M",
    "epoch": 6,
    "global_step": 9600,
    "total_tokens": 78795714,
    "train_loss": 4.725334499922877,
    "val_loss": 6.43682826052811,
    "val_perplexity": 624.4231481152083
  },
  {
    "checkpoint": "chck_70M",
    "epoch": 7,
    "global_step": 11200,
    "total_tokens": 91928333,
    "train_loss": 4.673132990233252,
    "val_loss": 6.401044900532549,
    "val_perplexity": 602.4742347394899
  },
  {
    "checkpoint": "chck_80M",
    "epoch": 8,
    "global_step": 12800,
    "total_tokens": 105060952,
    "train_loss": 4.637369798139151,
    "val_loss": 6.377206114588174,
    "val_perplexity": 588.2818175268471
  },
  {
    "checkpoint": "chck_90M",
    "epoch": 9,
    "global_step": 14400,
    "total_tokens": 118193571,
    "train_loss": 4.616350795298911,
    "val_loss": 6.368870010163261,
    "val_perplexity": 583.3982222251805
  },
  {
    "checkpoint": "chck_100M",
    "epoch": 10,
    "global_step": 16000,
    "total_tokens": 131326190,
    "train_loss": 4.607366539570785,
    "val_loss": 6.369217479096026,
    "val_perplexity": 583.6009702051064
  }
]