app.py

import os
import gradio as gr  
from openai import OpenAI  

# Load system prompt from environment variable
SYSTEM_PROMPT = os.getenv("SYSTEM_PROMPT")

# Initialize client
client = OpenAI(  
    base_url="http://soggy-sage-goat-8000.1.cricket.hyperbolic.xyz:30000/v1/",  
    api_key="hyperbolic"  
)  

def predict(message, history):  
    # If history is empty, insert the system prompt
    if not any(msg["role"] == "system" for msg in history):
        history.insert(0, {"role": "system", "content": SYSTEM_PROMPT})

    history.append({"role": "user", "content": message})  

    stream = client.chat.completions.create(  
        messages=history,  
        model="meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",  
        stream=True  
    )  

    chunks = []  
    for chunk in stream:  
        chunks.append(chunk.choices[0].delta.content or "")  
        yield "".join(chunks)  

demo = gr.ChatInterface(
    fn=predict,
    type="messages",
    chatbot=gr.Chatbot(
        type="messages", 
        label="💙ACC Emulect+💙", 
        show_copy_button=True, 
        avatar_images=(
            "https://huggingface.co/spaces/TejAndrewsACC/Z3ta_Z/resolve/main/Screenshot_20250201-131420.png", 
            "https://huggingface.co/spaces/TejAndrewsACC/Z3ta_Z/resolve/main/Screenshot_20250201-131420.png"
        ), 
        placeholder="💙Hi, I'm ACC Emulect+💙", 
    ),
    theme="TejAndrewsACC/Emulect",
)

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