{
  "batch_size": 128,
  "sequence_length": 64,
  "num_epochs": 1,
  "learning_rate": 0.001,
  "seed": 42,
  "n_iterations": 10,
  "n_freq_train": 1,
  "dropout_rate": 0.2,
  "num_layers": 6,
  "embed_size": 256,
  "num_heads": 8,
  "head_size": 32,
  "use_wandb": false,
  "wandb_project": "nanollm",
  "wandb_entity": null,
  "push_to_hub": true,
  "repo_id": "SauravMaheshkar/nanollm",
  "vocab_size": 65
}