app.py

from transformers import pipeline
import gradio as gr

# Initialize the pipeline with your specific model and specify the max_length
chatbot = pipeline(model="morph-labs/morph-prover-v0-7b", 
                   max_length=512,  # Increase this as needed
                   )

def vanilla_chatbot(message):
    try:
        # Format the message with the instructional tokens
        formatted_message = f"[INST] <<SYS>>\nYou are a helpful assistant.\n<</SYS>>\n\n{message}[/INST]"
        
        # Generate the model's response with a specified max_new_tokens
        output = chatbot(formatted_message, max_new_tokens=500)  # Adjust max_new_tokens as needed
        
        # Extract the model's response from the output
        response = output[0] if output else "No response generated."
        
        # Return the response to be displayed in the Gradio interface
        return response
    except Exception as e:
        # Log the exception message
        print(f"An unexpected error occurred: {e}")
        # Return a generic error message
        return "An unexpected error occurred."

# Create a Gradio interface
demo_chatbot = gr.Interface(
    fn=vanilla_chatbot,
    inputs="text",
    outputs="text",
    title="Vanilla Chatbot",
    description="Enter text to start chatting."
)

# Launch the Gradio app
demo_chatbot.launch()