multimodal_queries.py

import torch
import gradio as gr
from PIL import Image
from transformers import AutoProcessor, AutoModel

# Load the model and processor
model_id = "OpenGVLab/InternVL2_5-78B"
device = "cuda" if torch.cuda.is_available() else "cpu"

# Initialize the model and processor
model = AutoModel.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
    use_flash_attn=True,
    trust_remote_code=True
).eval().to(device)

processor = AutoProcessor.from_pretrained(model_id)

def generate_model_response(image_file, user_query):
    """
    Processes the uploaded image and user query to generate a response from the model.
    
    Parameters:
    - image_file: The uploaded image file.
    - user_query: The user's question about the image.

    Returns:
    - str: The generated response from the model.
    """
    try:
        # Load and prepare the image
        raw_image = Image.open(image_file).convert("RGB")
        
        # Prepare inputs for the model using the processor
        inputs = processor(images=raw_image, text=user_query, return_tensors="pt").to(device)
        
        # Generate response from the model
        outputs = model.generate(**inputs)
        
        # Decode and return the response
        response_text = processor.decode(outputs[0], skip_special_tokens=True)
        
        return response_text
    
    except Exception as e:
        print(f"Error in generating response: {e}")
        return f"An error occurred: {str(e)}"

# Gradio Interface
iface = gr.Interface(
    fn=generate_model_response,
    inputs=[
        gr.Image(type="file", label="Upload Image"),
        gr.Textbox(label="Enter your question", placeholder="What do you want to know about this image?")
    ],
    outputs="text",
)

iface.launch(share=True)