app.py

# Get API token from environment variable
#api_token = os.getenv("HF_TOKEN").strip()

import gradio as gr
from transformers import AutoModel, AutoTokenizer
import torch

# Load the model and tokenizer
model_name = "ContactDoctor/Bio-Medical-MultiModal-Llama-3-8B-V1"
model = AutoModel.from_pretrained(model_name, trust_remote_code=True, device_map="auto", torch_dtype=torch.float16)
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)

def process_query(image, question):
    inputs = {"question": question}
    if image:
        inputs["image"] = image

    # Process the inputs and generate a response
    response = model.chat(image=inputs.get("image"), msgs=[{"role": "user", "content": question}], tokenizer=tokenizer)
    return response

iface = gr.Interface(
    fn=process_query,
    inputs=[gr.Image(label="Upload Medical Image"), gr.Textbox(label="Question")],
    outputs="text",
    title="Medical Multimodal Assistant",
    description="Upload a medical image and ask your question."
)

iface.launch()