{
  "architectures": [
    "FastPlusForCausalLM"
  ],
  "dtype": "float32",
  "hidden_size": 544,
  "initializer_range": 0.02,
  "intermediate_size": 1632,
  "kd_alpha": 0.4,
  "kd_temperature": 2.5,
  "max_position_embeddings": 512,
  "model_type": "fastplus",
  "num_attention_heads": 8,
  "num_hidden_layers": 12,
  "tie_word_embeddings": true,
  "transformers_version": "5.12.0",
  "use_cache": false,
  "vocab_size": 15968
}