app.py

"""
来自 https://github.com/OpenLMLab/MOSS/blob/main/moss_web_demo_gradio.py


# 难点



# 单卡报错
python moss_web_demo_gradio.py --model_name fnlp/moss-moon-003-sft --gpu 0,1,2,3

# TODO
- 第一句：
- 代码和表格的预览
- 可编辑chatbot：https://github.com/gradio-app/gradio/issues/4444
- 一个button，

## Reference

-
"""


import gradio as gr
# from models.hf_qwen2 import bot
from models.cpp_qwen2 import bot


#
# def postprocess(self, y):
#     if y is None:
#         return []
#     for i, (message, response) in enumerate(y):
#         y[i] = (
#             None if message is None else mdtex2html.convert((message)),
#             None if response is None else mdtex2html.convert(response),
#         )
#     return y
#
#
# gr.Chatbot.postprocess = postprocess


def parse_text(text):
    """copy from https://github.com/GaiZhenbiao/ChuanhuChatGPT/"""
    lines = text.split("\n")
    lines = [line for line in lines if line != ""]
    count = 0
    for i, line in enumerate(lines):
        if "```" in line:
            count += 1
            items = line.split('`')
            if count % 2 == 1:
                lines[i] = f'<pre><code class="language-{items[-1]}">'
            else:
                lines[i] = f'<br></code></pre>'
        else:
            if i > 0:
                if count % 2 == 1:
                    line = line.replace("`", "\`")
                    line = line.replace("<", "&lt;")
                    line = line.replace(">", "&gt;")
                    line = line.replace(" ", "&nbsp;")
                    line = line.replace("*", "&ast;")
                    line = line.replace("_", "&lowbar;")
                    line = line.replace("-", "&#45;")
                    line = line.replace(".", "&#46;")
                    line = line.replace("!", "&#33;")
                    line = line.replace("(", "&#40;")
                    line = line.replace(")", "&#41;")
                    line = line.replace("$", "&#36;")
                lines[i] = "<br>" + line
    text = "".join(lines)
    return text


def generate_query(chatbot, history):
    if history and history[-1]["role"] == "user":  # 该生成response了
        gr.Warning('You should generate assistant-response.')
        yield None, chatbot, history
    else:
        chatbot.append(None)
        streamer = bot.generate_query(history, stream=True)
        for query in streamer:
            chatbot[-1] = (query, None)
            yield query, chatbot, history

        history.append({"role": "user", "content": query})
        yield query, chatbot, history


def generate_response(query, chatbot, history):
    """
    自动模式下：query is None
    人工模式下：query 是用户输入
    :param query:
    :param chatbot:
    :param history:
    :return:
    """
    if query and history[-1]["role"] != "user":
        history.append({"role": "user", "content": query})
    query = history[-1]["content"]

    if history[-1]["role"] != "user":
        gr.Warning('You should generate or type user-input first.')
        yield chatbot, history
    else:
        streamer = bot.generate_response(history, stream=True)
        for response in streamer:
            chatbot[-1] = (query, response)
            yield chatbot, history

        history.append({"role": "assistant", "content": response})
        print(f"chatbot is {chatbot}")
        print(f"history is {history}")
        yield chatbot, history


def generate():
    """

    :return:
    """
    pass


def regenerate():
    """
    删除上一轮，重新生成。
    :return:
    """
    pass


def reset_user_input():
    return gr.update(value='')


def reset_state(system):
    return [], [{"role": "system", "content": system}]


system_list = [
    "You are a helpful assistant.",
    "你是一个导游。",
    "你是一个英语老师。",
    "你是一个程序员。",
    "你是一个心理咨询师。",
]

"""
TODO: 使用说明
"""
with gr.Blocks() as demo:
    # Knowledge Distillation through Self Chatting
    gr.HTML("""<h1 align="center">Distilling the Knowledge through Self Chatting</h1>""")
    system = gr.Dropdown(
        choices=system_list,
        value=system_list[0],
        allow_custom_value=True,
        interactive=True,
        label="System message"
    )
    chatbot = gr.Chatbot(avatar_images=("assets/man.png", "assets/bot.png"))
    with gr.Row():
        with gr.Column(scale=4):
            user_input = gr.Textbox(show_label=False, placeholder="Input...", lines=10)
            with gr.Row():
                generate_query_btn = gr.Button("生成问题")
                regen_btn = gr.Button("🤔️ Regenerate (重试)")
                submit_btn = gr.Button("生成回复", variant="primary")
                stop_btn = gr.Button("停止生成", variant="primary")
                empty_btn = gr.Button("🧹 Clear History (清除历史)")
        with gr.Column(scale=1):
            # generate_query_btn = gr.Button("Generate First Query")

            clear_btn = gr.Button("重置")
            gr.Dropdown(
                ["moss", "chatglm-2", "chatpdf"],
                value="moss",
                label="问题生成器",
                # info="Will add more animals later!"
            ),
            gr.Dropdown(
                ["moss", "chatglm-2", "gpt3.5-turbo"],
                value="gpt3.5-turbo",
                label="回复生成器",
                # info="Will add more animals later!"
            ),

    history = gr.State([{"role": "system", "content": system_list[0]}])

    system.change(reset_state, inputs=[system], outputs=[chatbot, history], show_progress="full")

    submit_btn.click(generate_response, [user_input, chatbot, history], [chatbot, history],
                     show_progress="full")
    # submit_btn.click(reset_user_input, [], [user_input])

    clear_btn.click(reset_state, inputs=[system], outputs=[chatbot, history], show_progress="full")

    generate_query_btn.click(generate_query, [chatbot, history], outputs=[user_input, chatbot, history],
                             show_progress="full")

    # generate_query_btn.

    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",
              info="Larger temperature increase the randomness"),
    gr.Slider(
        minimum=0.1,
        maximum=1.0,
        value=0.95,
        step=0.05,
        label="Top-p (nucleus sampling)",
    ),

demo.queue().launch(share=False, server_name="0.0.0.0")
# demo.queue().launch(share=True)