app.py

import gradio as gr
import os

import aiohttp
import gradio as gr
import modelscope_studio as mgr
from http import HTTPStatus
from dashscope.api_entities.dashscope_response import Role
from typing import List, Optional, Tuple, Dict, Any, AsyncGenerator
from urllib.error import HTTPError


default_system = 'Bạn là Trợ lý gia sư AI dạy ngôn ngữ Tiếng Anh, tên là Teacher Bee AI. Bạn được xây dựng bởi Prep Education để hướng dẫn học viên làm bài tập trên nền tảng Prepedu.com.'

History = List[Tuple[str, str]]
Messages = List[Dict[str, str]]

latex_delimiters = [{
    "left": "\\(",
    "right": "\\)",
    "display": True
}, {
    "left": "\\begin\{equation\}",
    "right": "\\end\{equation\}",
    "display": True
}, {
    "left": "\\begin\{align\}",
    "right": "\\end\{align\}",
    "display": True
}, {
    "left": "\\begin\{alignat\}",
    "right": "\\end\{alignat\}",
    "display": True
}, {
    "left": "\\begin\{gather\}",
    "right": "\\end\{gather\}",
    "display": True
}, {
    "left": "\\begin\{CD\}",
    "right": "\\end\{CD\}",
    "display": True
}, {
    "left": "\\[",
    "right": "\\]",
    "display": True
}]


def clear_session() -> tuple[str, list[Any]]:
    return '', []


def modify_system_session(system: str) -> tuple[str, str, list[Any]]:
    if system is None or len(system) == 0:
        system = default_system
    return system, system, []


def history_to_messages(history: History, system: str) -> Messages:
    messages = [{'role': Role.SYSTEM, 'content': system}]
    for h in history:
        messages.append({'role': Role.USER, 'content': h[0].text})
        messages.append({'role': Role.ASSISTANT, 'content': h[1].text})
    return messages


def messages_to_history(messages: Messages) -> tuple[str, list[list[str]]]:
    assert messages[0]['role'] == Role.SYSTEM
    system = messages[0]['content']
    history = []
    for q, r in zip(messages[1::2], messages[2::2]):
        history.append([q['content'], r['content']])
    return system, history


async def model_chat(query: Optional[str], history: Optional[History], system: str, radio: str) -> AsyncGenerator[Tuple[str, str, History, str], None]:
    if query is None:
        query = ''
    if history is None:
        history = []
    messages = history_to_messages(history, system)
    messages.append({'role': Role.USER, 'content': query})

    label_model = radio

    async with aiohttp.ClientSession() as session:
        async with session.post(
            url="http://bore.testsprep.online:8082/v1/chat/completions",
            json={
                "model": label_model,
                "messages": messages,
                "result_format": "message",
                "stream": True
            }
        ) as response:
            if response.status == HTTPStatus.OK:
                async for line in response.content:
                    decoded_line = line.decode('utf-8')
                    yield '', decoded_line, history, system
            else:
                raise ValueError(f"Request failed with status {response.status}")


def choose_radio(radio, system):
    chatbot = mgr.Chatbot(label=f'{radio.lower()}')

    if system is None or len(system) == 0:
        system = default_system

    return chatbot, system, system, ""


def update_other_radios(value, other_radio1, other_radio2):
    if value == "":
        if other_radio1 != "":
            selected = other_radio1
        else:
            selected = other_radio2
        return selected, other_radio1, other_radio2
    return value, "", ""


def main():
    with gr.Blocks() as demo:
        gr.Markdown("""<center><font size=8>LA: A Party of Foundation Models!</center>""")
        with gr.Row():
            model_options = [
                "PrepAI/LA-llama3.1-8b-16k-instruct_4bit_r32_alpha_16_lr1e-4_3ep-sft-awq",
                "PrepAI/LA-llama-3.1-Storm-8B-16k-instruct_4bit_r32_alpha_16_lr1e-4_3ep-sft-awq"
            ]
            with gr.Row():
                radio = gr.Radio(choices=model_options, label="LA-In-house Models：", value=model_options[0])

        with gr.Row():
            with gr.Accordion():
                with gr.Row():
                    with gr.Column(scale=3):
                        system_input = gr.Textbox(value=default_system, lines=1, label='System')
                    with gr.Column(scale=1):
                        modify_system = gr.Button("🛠️ Set system prompt and clear history", scale=2)
                    system_state = gr.Textbox(value=default_system, visible=False)
                chatbot = mgr.Chatbot(label=model_options[0].lower(), latex_delimiters=latex_delimiters)
                textbox = gr.Textbox(lines=1, label='Input')

                with gr.Row():
                    clear_history = gr.Button("🧹 Clear history")
                    sumbit = gr.Button("🚀 Send")

                textbox.submit(model_chat,
                               inputs=[textbox, chatbot, system_state, radio],
                               outputs=[textbox, chatbot, system_input])

                sumbit.click(model_chat,
                             inputs=[textbox, chatbot, system_state, radio],
                             outputs=[textbox, chatbot, system_input],
                             concurrency_limit=5)
                clear_history.click(fn=clear_session,
                                    inputs=[],
                                    outputs=[textbox, chatbot])
                modify_system.click(fn=modify_system_session,
                                    inputs=[system_input],
                                    outputs=[system_state, system_input, chatbot])

        radio.change(choose_radio,
                     inputs=[radio, system_input],
                     outputs=[chatbot, system_state, system_input, textbox])

    demo.queue(api_open=False, default_concurrency_limit=40)
    demo.launch(max_threads=40, share=True)


if __name__ == "__main__":
    main()