{
  "architectures": [
    "HybridForCausalLM"
  ],
  "attention_dropout": 0.0,
  "auto_map": {
    "AutoConfig": "configuration_model.HybridModelConfig",
    "AutoModelForCausalLM": "modelling_model.HybridForCausalLM"
  },
  "bos_token_id": 151643,
  "dtype": "bfloat16",
  "eos_token_id": 248046,
  "hidden_act": "silu",
  "hidden_size": 2048,
  "initializer_range": 0.02,
  "intermediate_size": 1024,
  "kv_lora_rank": 512,
  "max_position_embeddings": 8192,
  "mhc_alpha_init": 0.0,
  "mhc_num_streams": 4,
  "mhc_readout_init": "first",
  "mhc_rmsnorm_eps": 1e-06,
  "mhc_sinkhorn_iters": 20,
  "mhc_stream_init": "paper",
  "model_type": "hybrid_model",
  "num_attention_heads": 16,
  "num_hidden_layers": 32,
  "pad_token_id": 248044,
  "q_lora_rank": 1536,
  "qk_rope_head_dim": 64,
  "rms_norm_eps": 1e-06,
  "rope_theta": 10000.0,
  "sliding_window": 4096,
  "tie_word_embeddings": true,
  "transformers_version": "5.5.0",
  "use_cache": true,
  "vocab_size": 248077
}