app.py

#!/usr/bin/env python3
import os
import json
import requests
import gradio as gr

ENDPOINT = os.environ.get("VLLM_ENDPOINT") or os.getenv("VLLM_ENDPOINT")
MODEL = os.environ.get("VLLM_MODEL") or os.getenv("VLLM_MODEL")

print(f"Debug - Found ENDPOINT: {ENDPOINT is not None}")
print(f"Debug - Found MODEL: {MODEL is not None}")
print(f"Debug - All env vars: {list(os.environ.keys())}")

if not ENDPOINT or not MODEL:
    raise ValueError("VLLM_ENDPOINT and VLLM_MODEL environment variables must be set")


def respond(
    message,
    history: list[dict[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    """
    Send messages to vLLM endpoint and stream the response.
    """
    messages = [{"role": "system", "content": system_message}]
    messages.extend(history)
    messages.append({"role": "user", "content": message})

    payload = {
        "model": MODEL,
        "messages": messages,
        "max_tokens": max_tokens,
        "temperature": temperature,
        "top_p": top_p,
        "stream": True
    }

    try:
        response = requests.post(
            ENDPOINT,
            headers={"Content-Type": "application/json"},
            data=json.dumps(payload),
            stream=True
        )
        response.raise_for_status()

        accumulated_response = ""
        
        for line in response.iter_lines():
            if line:
                line = line.decode('utf-8')
                if line.startswith('data: '):
                    line = line[6:]  # Remove 'data: ' prefix
                    
                if line.strip() == '[DONE]':
                    break
                    
                try:
                    chunk = json.loads(line)
                    if 'choices' in chunk and len(chunk['choices']) > 0:
                        delta = chunk['choices'][0].get('delta', {})
                        content = delta.get('content', '')
                        if content:
                            accumulated_response += content
                            yield accumulated_response
                except json.JSONDecodeError:
                    continue
                    
    except Exception as e:
        yield f"Error: {str(e)}"


"""
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""
chatbot = gr.ChatInterface(
    respond,
    type="messages",
    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)",
        ),
    ],
)

with gr.Blocks(title="vLLM Chatbot") as demo:
    gr.Markdown("# 💬 Chat Interface")
    gr.Markdown("""
    Configure the endpoint via environment variables:
    - `VLLM_ENDPOINT`: vLLM server URL
    - `VLLM_MODEL`: Model name
    """)
    chatbot.render()


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