{
  "architectures": [
    "DecoderOnlyTransformer"
  ],
  "d_model": 128,
  "model_type": "decoder-only",
  "nhead": 2,
  "num_layers": 2,
  "torch_dtype": "float32",
  "transformers_version": "4.51.3",
  "vocab_size": 50257
}