app.py

import gradio as gr
from huggingface_hub import InferenceClient

# Load the ClinicalBERT model
MODEL_NAME = "emilyalsentzer/Bio_ClinicalBERT"
client = InferenceClient(MODEL_NAME)

def respond(
    message,
    history,
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    if history is None:
        history = []

    messages = [{"role": "system", "content": system_message}]

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": message})

    # Ensure input includes [MASK] for ClinicalBERT
    if "[MASK]" not in message:
        message += " [MASK]"

    try:
        response = client.fill_mask(message)
        prediction = response[0]["sequence"]
        confidence = response[0]["score"]

        # Append to history
        history.append((message, prediction))

        return prediction, history  # Must return history explicitly as an output
    
    except Exception as e:
        return {"error": str(e)}, history

# Create Gradio interface
demo = gr.Interface(
    fn=respond,
    inputs=[
        gr.Textbox(label="User Input"),
        gr.State(),  # Explicit state for history tracking
        gr.Textbox(value="You are a friendly medical assistant.", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),
    ],
    outputs=[
        gr.Textbox(label="Model Response"),
        gr.State(),  # Explicit state output
    ],
)

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