config.json

{
  "architectures": [
    "LegatoModel"
  ],
  "encoder_pretrained_model_name_or_path": "./vision_encoder",
  "image_token_index": 4096,
  "model_type": "legato",
  "pad_token_id": 4,
  "text_config": {
    "bos_token_id": 1,
    "cross_attention_layers": [
      3,
      7,
      11,
      15
    ],
    "dropout": 0,
    "eos_token_id": [
      2
    ],
    "hidden_act": "silu",
    "hidden_size": 768,
    "initializer_range": 0.02,
    "intermediate_size": 1526,
    "max_position_embeddings": 131072,
    "model_type": "mllama_text_model",
    "num_attention_heads": 12,
    "num_hidden_layers": 18,
    "num_key_value_heads": 6,
    "pad_token_id": 4,
    "rms_norm_eps": 1e-05,
    "rope_scaling": {
      "factor": 8.0,
      "high_freq_factor": 4.0,
      "low_freq_factor": 1.0,
      "original_max_position_embeddings": 8192,
      "rope_type": "llama3"
    },
    "rope_theta": 500000,
    "torch_dtype": "float32",
    "use_cache": true,
    "vocab_size": 4097
  },
  "torch_dtype": "float32",
  "transformers_version": "4.50.0",
  "vision_config": {
    "_attn_implementation_autoset": true,
    "_name_or_path": "meta-llama/Llama-3.2-11B-Vision",
    "attention_heads": 16,
    "hidden_act": "gelu",
    "hidden_size": 1280,
    "image_size": 448,
    "initializer_range": 0.02,
    "intermediate_layers_indices": [
      3,
      7,
      15,
      23,
      30
    ],
    "intermediate_size": 5120,
    "max_num_tiles": 4,
    "model_type": "mllama_vision_model",
    "norm_eps": 1e-05,
    "num_channels": 3,
    "num_global_layers": 8,
    "num_hidden_layers": 32,
    "patch_size": 14,
    "supported_aspect_ratios": [
      [
        1,
        1
      ],
      [
        1,
        2
      ],
      [
        1,
        3
      ],
      [
        1,
        4
      ],
      [
        2,
        1
      ],
      [
        2,
        2
      ],
      [
        3,
        1
      ],
      [
        4,
        1
      ]
    ],
    "torch_dtype": "float32",
    "vision_output_dim": 7680
  }
}