{
  "model_type": "gpt2",
  "architectures": [
    "GPT2LMHeadModel"
  ],
  "tokenizer_class": "GPT2Tokenizer",
  "eos_token_id": 50256,
  "pad_token_id": 50256
}