开源模型应用落地-qwen模型小试-入门篇（四）

一、前言

    通过前面三篇的学习，我相信您已经学会了如何使用qwen模型的基本方法。现在，我们将利用gradio工具，为qwen模型加上一个漂亮的界面，以实现人与机器之间的交互操作。

---

二、术语

2.1. gradio

    Gradio是一个Python库，它可以帮助我们轻松地为机器学习模型创建用户界面。它提供了一个简单而强大的方式，让我们能够通过图形界面与模型进行交互。使用Gradio，我们可以快速地将模型部署为一个Web应用程序或者本地应用程序，而无需编写复杂的前端代码。它还支持文本输入、图像输入和实时摄像头输入等多种输入方式，让用户能够方便地与模型进行对话或者测试。

---

三、前置条件

3.1.windows操作系统

3.2.下载Qwen-1_8B-Chat模型

    从huggingface下载：https://huggingface.co/Qwen/Qwen-1_8B-Chat/tree/main

    从魔搭下载：https://modelscope.cn/models/qwen/Qwen-1_8B-Chat/files

---

四、技术实现

# -*-  coding = utf-8 -*-
import os
import logging
import warnings

import gradio as gr
import torch
import sys
import mdtex2html
from transformers import GenerationConfig, AutoTokenizer, AutoModelForCausalLM

warnings.filterwarnings("ignore")

logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s [%(levelname)s]: %(message)s',  # 指定日志输出格式
    datefmt='%Y-%m-%d %H:%M:%S'  # 指定日期时间格式
)

root_path = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
sys.path.append(root_path)
logging.info(f'root_path: {root_path}')

model_name = "qwen-1_8b-chat"

if sys.platform == "linux":
    model_path = f"/data/model/{model_name}"
else:
    model_path =  f"E:\\model\\{model_name}"

system = '你是一个乐于助人的人工智能助手，愿意解决人类提出的各种问题'

def loadTokenizer(modelPath):
    tokenizer = AutoTokenizer.from_pretrained(modelPath, trust_remote_code=True)
    return tokenizer

def loadModel(modelPath,config):
    if torch.cuda.is_available():
        model = AutoModelForCausalLM.from_pretrained(modelPath, device_map="auto", trust_remote_code=True,fp16=True).eval()
    else:
        model = AutoModelForCausalLM.from_pretrained(modelPath, device_map="cpu", trust_remote_code=True).eval()
    model.generation_config = config
    return model


def predict(_query, _chatbot, _task_history):
    global model,tokenizer

    _chatbot.append((_parse_text(_query), ""))
    result = ""

    for response in model.chat_stream(tokenizer=tokenizer,query=_query, history=_task_history, system=system):
        _chatbot[-1] = (_parse_text(_query), _parse_text(response))

        yield _chatbot
        result = _parse_text(response)

    # print(f"History: {_task_history}")
    _task_history.append((_query, result))
    # print(f"Qwen-Chat: {_parse_text(result)}")


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):
    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'
'
            else:
                lines[i] = f"
"
        else:
            if i > 0:
                if count % 2 == 1:
                    line = line.replace("`", r"\`")
                    line = line.replace("<", "<")
                    line = line.replace(">", ">")
                    line = line.replace(" ", " ")
                    line = line.replace("*", "*")
                    line = line.replace("_", "_")
                    line = line.replace("-", "-")
                    line = line.replace(".", ".")
                    line = line.replace("!", "!")
                    line = line.replace("(", "(")
                    line = line.replace(")", ")")
                    line = line.replace("$", "$")
                lines[i] = "
" + line
    text = "".join(lines)
    return text


def _gc():
    import gc
    gc.collect()
    if torch.cuda.is_available():
        torch.cuda.empty_cache()


def regenerate(_chatbot, _task_history):
    if not _task_history:
        yield _chatbot
        return
    item = _task_history.pop(-1)
    _chatbot.pop(-1)
    yield from predict(_query=item[0], _chatbot=_chatbot, _task_history=_task_history)


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


def reset_state(_chatbot, _task_history):
    _task_history.clear()
    _chatbot.clear()
    _gc()
    return _chatbot


with gr.Blocks(css=".app.svelte-1kyws56.svelte-1kyws56 {width: 50%}") as interface:
    # 设置tab选项卡
    with gr.Tab("开放式输入"):
        chatbot = gr.Chatbot()
        query = gr.Textbox(lines=2, label='Input')
        task_history = gr.State([])

        with gr.Row():
            empty_btn = gr.Button("清除历史")
            submit_btn = gr.Button("发送")
            regen_btn = gr.Button("重试")

    # 聊天
    submit_btn.click(predict, [query, chatbot, task_history], [chatbot], show_progress=True)
    submit_btn.click(reset_user_input, [], [query])
    empty_btn.click(reset_state, [chatbot, task_history], outputs=[chatbot], show_progress=True)
    regen_btn.click(regenerate, [chatbot, task_history], [chatbot], show_progress=True)



if __name__ == "__main__":
    config = GenerationConfig.from_pretrained(model_path, trust_remote_code=True, top_p=0.9,
                                              temperature=0.45, repetition_penalty=1.1, do_sample=True,
                                              max_new_tokens=8192)
    tokenizer = loadTokenizer(model_path)
    model = loadModel(model_path,config)

    app, local_url, share_url = interface.queue(concurrency_count=5).launch(debug=True, share=False, show_api=False,
                                                                            inbrowser=False, server_name='0.0.0.0',
                                                                            auth=("zhangsan", '123456'), width='70%')
   

启动：

---

五、测试

5.1.访问

    浏览器访问：http://127.0.0.1:7860

5.2.交互

5.3.清除历史

5.4.重试

---

六、附带说明

6.1. 登录账号

    在代码指定登录信息：auth=("zhangsan", '123456')

    若需要自定义校验规则可以指定校验方法：例如

app, local_url, share_url = interface.queue().launch(debug=True,share=False,show_api=False,inbrowser=False, server_name='0.0.0.0',auth=same_auth, auth_message="username and password must be the same")


# 账户和密码相同就可以通过
def same_auth(username, password):
    return username == password