vision_model.py

import torch
from unsloth import FastVisionModel
from transformers import AutoProcessor
import config

def load_vision_model():
    """
    Loads the FastVisionModel and its adapter for inference.
    
    Returns:
        A tuple containing the loaded model and processor, or (None, None) on error.
    """
    try:
        model, processor = FastVisionModel.from_pretrained(
            model_name=config.VISION_MODEL_NAME,
            max_seq_length=config.MAX_SEQ_LENGTH,
            load_in_4bit=True,
            dtype=None,
        )
        FastVisionModel.for_inference(model)
        model.load_adapter(config.ADAPTER_PATH)
        print(f"✅ Vision model and adapter '{config.ADAPTER_PATH}' loaded successfully!")
        return model, processor
    except Exception as e:
        print(f"❌ CRITICAL ERROR during vision model loading: {e}")
        return None, None