import gradio as gr
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# Load the Flan-T5-XL model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("google/flan-t5-xl")
model = AutoModelForSeq2SeqLM.from_pretrained("google/flan-t5-xl")

# Define a function to generate responses
def respond(message, chat_history):
    # Format the message for the model
    prompt = f"User: {message}\nAssistant:"
    
    # Generate the response
    inputs = tokenizer(prompt, return_tensors="pt")
    outputs = model.generate(**inputs)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    
    # Update the chat history
    chat_history.append((message, response))
    
    return "", chat_history

# Create the Gradio chatbot interface
iface = gr.Chatbot()
iface.launch(handle_messages=respond)