app.py

import gradio as gr
from huggingface_hub import InferenceClient
from transformers import pipeline

# Initialize the InferenceClient with the Zephyr model
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")

# Use a pipeline for the ParlBERT model for fill-mask
mask_pipe = pipeline("fill-mask", model="InfAI/parlbert-german-law")

# Define the function for chat completion
def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    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})

    response = ""

    for message in client.chat_completion(
        messages,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = message.choices[0].delta.content

        response += token
        yield response

# Function to handle mask filling
def fill_mask_function(text):
    return mask_pipe(text)

# Create the Gradio interface
demo = gr.Interface(
    fn=respond,
    inputs=[
        gr.Textbox(label="Enter your message"),
        gr.State(),  # History
        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)"),
    ],
    outputs=[
        gr.Textbox(label="Response"),
    ],
    title="Zephyr and ParlBERT Chatbot",
    description="This chatbot uses Zephyr-7B model and ParlBERT (German Law) for conversation and masked word predictions."
)

# Create a separate Gradio interface for the fill-mask model
mask_demo = gr.Interface(
    fn=fill_mask_function,
    inputs=gr.Textbox(label="Input text with [MASK] token"),
    outputs=gr.JSON(label="Model output"),
    live=True,
    title="InfAI ParlBERT - German Law",
    description="Enter a sentence with a [MASK] token to get predictions from the InfAI ParlBERT model trained on German law text."
)

# Launch both interfaces
if __name__ == "__main__":
    demo.launch()
    mask_demo.launch()