{ 
  "architectures": [
    "GPTForCausalLM"
  ],
  "vocab_size": 50304,
  "hidden_size": 1024, 
  "num_hidden_layers": 24,
  "num_attention_heads": 16, 
  "intermediate_size": 4096,
  "hidden_act": "gelu", 
  "hidden_dropout_prob": 0.1, 
  "attention_probs_dropout_prob": 0.1, 
  "max_position_embeddings": 1024, 
  "type_vocab_size": 1,
  "initializer_range": 0.02, 
  "pad_token_id": 0, 
  "eos_token_id": 50256, 
  "bos_token_id": 50256, 
  "eol_token_id": 3 
}