config.json

{
  "vocab_size": 8000,
  "d_model": 256,
  "n_layers": 16,
  "n_heads": 16,
  "max_seq_len": 4096,
  "dropout": 0.1,
  "model_type": "scratch_transformer",
  "architectures": [
    "ScratchTransformer"
  ]
}