app.py

import os
from openai import AsyncOpenAI
import gradio as gr

default_model = "llama3:8b-instruct-q4_K_M"
models = ["llama3:8b-instruct-q4_K_M", "codestral:22b-v0.1-q4_K_M"]
description = "Learn more at https://replicantzk.com."
base_url = os.getenv("OPENAI_BASE_URL") or "https://platform.replicantzk.com"
api_key = os.getenv("OPENAI_API_KEY")


async def predict(message, history, model, temperature, stream, base_url, api_key):
    client = AsyncOpenAI(base_url=base_url, api_key=api_key)

    history_openai_format = []
    for human, assistant in history:
        history_openai_format.append({"role": "user", "content": human})
        history_openai_format.append({"role": "assistant", "content": assistant})

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

    try:
        response = await client.chat.completions.create(
            model=model,
            messages=history_openai_format,
            temperature=temperature,
            stream=stream,
        )

        if stream:
            partial_message = ""
            async for chunk in response:
                if chunk.choices[0].delta.content is not None:
                    partial_message += chunk.choices[0].delta.content
                    yield partial_message
        else:
            yield response.choices[0].message.content

    except Exception as e:
        raise gr.Error(str(e))


model = gr.Dropdown(label="Model", choices=models, value=default_model)
temperature = gr.Slider(0, 1, value=0, label="Temperature")
stream = gr.Checkbox(value=True, label="Stream")
base_url = gr.Textbox(label="OpenAI-compatible base URL", value=base_url)
api_key = gr.Textbox(label="OpenAI-compatible API key", type="password", value=api_key)
demo = gr.ChatInterface(
    fn=predict,
    additional_inputs=[model, temperature, stream, base_url, api_key],
    description=description,
)

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