config.json

{
    "architectures": [
        "LlavaForConditionalGeneration"
    ],
    "ignore_index": -100,
    "image_token_index": 32038,
    "model_type": "llava",
    "pad_token_id": 32039,
    "projector_hidden_act": "gelu",
    "quantization": {
        "group_size": 64,
        "bits": 4
    },
    "text_config": {
        "architectures": [
            "LlamaForCausalLM"
        ],
        "eos_token_id": 32000,
        "hidden_size": 3072,
        "intermediate_size": 8192,
        "max_position_embeddings": 4096,
        "model_type": "llama",
        "original_max_position_embeddings": 4096,
        "pad_token_id": 32000,
        "rms_norm_eps": 1e-05,
        "sliding_window": 2048,
        "torch_dtype": "float16",
        "vocab_size": 32064
    },
    "torch_dtype": "float16",
    "transformers_version": "4.40.1",
    "vision_config": {
        "architectures": [
            "CLIPVisionModel"
        ],
        "dropout": 0.0,
        "hidden_size": 1024,
        "image_size": 336,
        "intermediate_size": 4096,
        "model_type": "clip_vision_model",
        "num_attention_heads": 16,
        "num_hidden_layers": 24,
        "patch_size": 14,
        "projection_dim": 768,
        "torch_dtype": "float32"
    },
    "vision_feature_layer": -2,
    "vision_feature_select_strategy": "default"
}