app.py

import gradio as gr
from huggingface_hub import InferenceClient

# ---------------- CONFIG ----------------
MODEL_REPO = "HuggingFaceH4/zephyr-7b-beta"
SYSTEM_PROMPT_DEFAULT = "You are Zephyr, a helpful, concise and polite AI assistant."

MAX_NEW_TOKENS_DEFAULT = 512
TEMP_DEFAULT = 0.7
TOP_P_DEFAULT = 0.95

# Create client (calls Hugging Face Inference API, not local model)
client = InferenceClient(MODEL_REPO)

# ---------------- CHAT FUNCTION ----------------
def stream_response(message, chat_history, system_message, max_tokens, temperature, top_p):
    messages = [{"role": "system", "content": system_message}]
    for user_msg, bot_msg in chat_history:
        if user_msg:
            messages.append({"role": "user", "content": user_msg})
        if bot_msg:
            messages.append({"role": "assistant", "content": bot_msg})
    messages.append({"role": "user", "content": message})

    response = ""
    for msg in client.chat_completion(
        messages,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = msg.choices[0].delta.content or ""
        response += token
        yield "", chat_history + [(message, response)]


# ---------------- UI ----------------
with gr.Blocks(theme=gr.themes.Soft(primary_hue="violet", secondary_hue="pink")) as demo:
    gr.Markdown(
        """
        # 📱 Zephyr-7B (Hosted on Hugging Face Inference API)
        Optimized for **mobile-friendly chat** ✨  
        <span style="opacity:0.7">Powered by HuggingFaceH4/zephyr-7b-beta</span>
        """
    )

    chatbot = gr.Chatbot(
        height=500,
        bubble_full_width=False,
        show_copy_button=True,
        label="Chat"
    )

    with gr.Row():
        msg = gr.Textbox(
            label="💬 Message",
            placeholder="Type your message…",
            scale=6
        )
        send_btn = gr.Button("🚀", variant="primary", scale=1)
        clear_btn = gr.Button("🧹", scale=1)

    with gr.Accordion("⚙️ Settings", open=False):
        system_prompt = gr.Textbox(
            label="System Prompt",
            value=SYSTEM_PROMPT_DEFAULT,
            lines=3
        )
        temperature = gr.Slider(0.1, 1.5, value=TEMP_DEFAULT, step=0.1, label="Temperature")
        top_p = gr.Slider(0.1, 1.0, value=TOP_P_DEFAULT, step=0.05, label="Top-p")
        max_tokens = gr.Slider(32, 2048, value=MAX_NEW_TOKENS_DEFAULT, step=16, label="Max new tokens")

    # Events (streaming response)
    send_btn.click(
        stream_response,
        [msg, chatbot, system_prompt, max_tokens, temperature, top_p],
        [msg, chatbot]
    )
    msg.submit(
        stream_response,
        [msg, chatbot, system_prompt, max_tokens, temperature, top_p],
        [msg, chatbot]
    )
    clear_btn.click(lambda: None, None, chatbot, queue=False)

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