app.py

import gradio as gr
import os
import cohere
from huggingface_hub import InferenceClient

# Retrieve API keys from environment variables
HF_API_KEY = os.getenv("HF_API_KEY")  
COHERE_API_KEY = os.getenv("COHERE_API_KEY")  

# Initialize clients
hf_model = "meta-llama/Llama-3.2-3B-Instruct"  # Change to preferred HF model
hf_client = InferenceClient(model=hf_model, token=HF_API_KEY)

cohere_client = cohere.Client(COHERE_API_KEY)


def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
    use_cohere,  # Checkbox input
):
    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})

    # 🔹 **Switch API based on checkbox**
    if use_cohere:
        response = cohere_client.chat(
            model="command-r-plus",  
            message=message,
            chat_history=[{"user_name": "User", "text": h[0]} for h in history if h[0]] +
                         [{"user_name": "Assistant", "text": h[1]} for h in history if h[1]],
            temperature=temperature,
            max_tokens=max_tokens,
            p=top_p,
        )
        return response.text  # Cohere returns full response

    else:
        response = ""
        for message in hf_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  # Hugging Face supports streaming


# 🔥 **Gradio UI with Checkbox**
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)",
        ),
        gr.Checkbox(label="Use Cohere API", value=False),  # Checkbox to toggle API
    ],
)

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