{
  "architectures": ["MultiModalityCausalLM"],
  "model_type": "multi_modality",
  "torch_dtype": "bfloat16",
  
  "llama_config": {
    "architectures": ["LlamaForCausalLM"],
    "hidden_size": 2048,
    "intermediate_size": 5632,
    "max_position_embeddings": 16384,
    "num_attention_heads": 16,
    "num_hidden_layers": 24,
    "num_key_value_heads": 16,
    "rms_norm_eps": 1e-6,
    "vocab_size": 102400,
    "torch_dtype": "bfloat16"
  },

  "clip_vision_config": {
    "architectures": ["CLIPVisionModel"],
    "hidden_size": 1024,
    "image_size": 384,
    "patch_size": 16,
    "projection_dim": 2048,
    "num_hidden_layers": 24,
    "num_attention_heads": 16,
    "intermediate_size": 4096
  },

  "alignment_config": {
    "projector_type": "mlp_gelu",
    "depth": 2,
    "input_dim": 1024,
    "n_embed": 2048
  },

  "generation_config": {
    "image_token_size": 16384,
    "vq_codebook_size": 8,
    "image_token_embed": 2048
  }
}