app.py

import gradio as gr
from huggingface_hub import InferenceClient
from transformers import pipeline
from typing import List, Tuple  # Importing for type annotations

# Initialize the BERT model pipeline for the "fill-mask" task
pipe = pipeline("fill-mask", model="bert-base-uncased")

# Function to handle the response generation using BERT
def respond(
    message: str,
    history: List[Tuple[str, str]],  # Using List and Tuple for type annotation
    system_message: str,
    max_tokens: int,
    temperature: float,
    top_p: float,
):
    messages = [{"role": "system", "content": system_message}]
    
    # Append history to the messages list
    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    # Append the current user message
    messages.append({"role": "user", "content": message})

    response = ""

    # Using the BERT pipeline to fill the mask (this is different from the GPT-style completion)
    # BERT doesn't generate text the same way, so we are simulating a response for this demo
    result = pipe(f"Hello, how are you today? {message} [MASK]")

    # Collecting the filled-in mask (likely output from BERT's "fill-mask" task)
    response = result[0]['sequence']

    yield response


# Setting up Gradio Interface
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a friendly Chatbot.", 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 (nucleus sampling)",
        ),
    ],
)

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