{
  "architectures": [
    "GPTForCausalLM"
  ],
  "auto_map": {
    "AutoConfig": "modeling_gpt.GPTConfig",
    "AutoModelForCausalLM": "modeling_gpt.GPTForCausalLM"
  },
  "base": 10000,
  "bos_token_id": 0,
  "dtype": "float32",
  "eos_token_id": 0,
  "ffn_dropout": 0.0,
  "head_dim": 64,
  "hidden_dim": 768,
  "initial_context_len": 2048,
  "intermediate_size": 760,
  "is_decoder": true,
  "max_context_len": 2048,
  "model_type": "custom_gpt",
  "ntk_alpha": 1.0,
  "ntk_beta": 32.0,
  "num_attn_heads": 12,
  "num_experts": 4,
  "num_experts_per_tok": 2,
  "num_hidden_layers": 24,
  "num_key_value_heads": 6,
  "pad_token_id": 4,
  "route_scale": 1.0,
  "scaling_factor": 1.0,
  "tie_word_embeddings": false,
  "transformers_version": "4.57.1",
  "update_param": 0.002,
  "vocab_size": 49152
}