{
  "model_type": "transformer_lm",
  "architectures": [
    "TransformerLM"
  ],
  "vocab_size": 10000,
  "context_length": 256,
  "num_layers": 4,
  "d_model": 512,
  "num_heads": 16,
  "d_ff": 1344,
  "theta": 10000.0
}