app.py

import spaces
import gradio as gr
from transformers import pipeline
import torch
import os

hf_token = os.environ["HF_TOKEN"]

# Load the Gemma 3 pipeline
pipe = pipeline(
    "image-text-to-text",
    model="google/gemma-3-4b-it",
    device="cuda",
    torch_dtype=torch.bfloat16,
    use_auth_token=hf_token
)

@spaces.GPU
def get_response(message, chat_history, image):
    # Check if image is provided
    if image is None:
        chat_history.append((message, "Please upload an image (required)"))
        return "", chat_history
    
    messages = [
        {
            "role": "system",
            "content": [{"type": "text", "text": "You are a helpful assistant."}]
        }
    ]
    
    user_content = [{"type": "image", "image": image}]
    
    # Add text message if provided
    if message:
        user_content.append({"type": "text", "text": message})
        
    messages.append({"role": "user", "content": user_content})
    
    # Call the pipeline
    output = pipe(text=messages, max_new_tokens=200)
    
    try:
        response = output[0]["generated_text"][-1]["content"]
        chat_history.append((message, response))
    except (KeyError, IndexError, TypeError) as e:
        error_message = f"Error processing the response: {str(e)}"
        chat_history.append((message, error_message))
    
    return "", chat_history

with gr.Blocks() as demo:
    gr.Markdown("# Gemma 3 Image Chat")
    gr.Markdown("Chat with Gemma 3 about images. Image upload is required for each message.")
    
    chatbot = gr.Chatbot()
    
    with gr.Row():
        msg = gr.Textbox(
            show_label=False,
            placeholder="Type your message here about the image...",
            scale=4
        )
        img = gr.Image(
            type="pil", 
            label="Upload image (required)", 
            scale=1
        )
    
    submit_btn = gr.Button("Send")
    
    # Clear button to reset the interface
    clear_btn = gr.Button("Clear")
    
    def clear_interface():
        return "", [], None
    
    submit_btn.click(
        get_response,
        inputs=[msg, chatbot, img],
        outputs=[msg, chatbot]
    )
    
    msg.submit(
        get_response,
        inputs=[msg, chatbot, img],
        outputs=[msg, chatbot]
    )
    
    clear_btn.click(
        clear_interface,
        inputs=None,
        outputs=[msg, chatbot, img]
    )

if __name__ == "__main__":
    demo.launch()