chatyuan_gradio.py

import os
import time
import ftplib
import threading
from tqdm.notebook import tqdm
import zipfile
import gradio as gr
import torch
# from transformers import T5Tokenizer, T5ForConditionalGeneration
from transformers import AutoTokenizer, AutoModel


def get_model_ftp(model_path, model_name):
    ftp = ftplib.FTP('10.209.16.22')
    ftp.login('soltest', 'soltest')
    folder_path = '/ftp/3D/ai-model/ChatYuan/ClueAI/'
    ftp.cwd(folder_path)

    file_list = ftp.nlst(folder_path)
    if os.path.join(folder_path, model_name) in file_list:
        # 获取远程文件的大小
        file_size = ftp.size(model_name)

        # 创建本地文件，并用二进制写模式打开
        with open(os.path.join(model_path, model_name), 'wb') as f:
            # 下载文件并显示进度条
            with tqdm.wrapattr(f, 'write', desc="Download " + model_name, total=file_size, unit='B', unit_scale=True) as pbar:
                ftp.retrbinary('RETR ' + model_name, pbar.write)

        ftp.quit()
        unzip(model_path, model_name)



def unzip(path, file_name):
    try:
        stop_unzip = threading.Event()
        thread = threading.Thread(target=print_flush, args=(stop_unzip, "start decompression "))
        thread.start()
        zip_file = zipfile.ZipFile(os.path.join(path, file_name))
        for names in zip_file.namelist():
            zip_file.extract(names, path)
        zip_file.close()

        stop_unzip.set()
        thread.join()

    except Exception as ex:
        stop_unzip.set()
        thread.join()
        os.remove(os.path.join(path, file_name))
        raise Exception(f"\nunzip失败：" + str(ex))


def prepare_model(model_dir):
    model_path = model_dir.split('/')[0]
    model_name = model_dir.split('/')[1]
    if not os.path.exists(model_dir):
        os.makedirs("ClueAI", exist_ok=True)
        get_model_ftp(model_path, model_name + '.zip')
        os.remove(os.path.join(model_path, model_name + '.zip'))
    
            
def print_flush(stop_event, str):
    loading_strings = [str + ".", str + "..", str + "...", str + ".", str + "..", str + "..."]
    index = 0

    while not stop_event.is_set():
        loading_str = loading_strings[index]
        print(loading_str, end="\r")
        index = (index + 1) % len(loading_strings)
        time.sleep(0.5)

        # Refresh the loading string every three cycles
        if index == 0:
            print(" " * len(loading_str), end="\r")
            time.sleep(0.2)
    print(loading_strings[index], end="\r")
    print("\n" + str.split(" ")[1] + " finish.")


model_dir = 'ClueAI/ChatYuan-large-v2'
prepare_model(model_dir)

tokenizer = AutoTokenizer.from_pretrained(model_dir)
model = AutoModel.from_pretrained(model_dir, trust_remote_code=True)

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model.to(device)
# model.half()


def preprocess(text):
    base_info = ""
    text = f"{base_info}{text}"
    text = text.replace("\n", "\\n").replace("\t", "\\t")
    return text


def postprocess(text):
    return text.replace("\\n", "\n").replace("\\t", "\t").replace(
        '%20', '  ')  # .replace(" ", " ")


generate_config = {
    'do_sample': True,
    'top_p': 0.9,
    'top_k': 50,
    'temperature': 0.7,
    'num_beams': 1,
    'max_length': 1024,
    'min_length': 3,
    'no_repeat_ngram_size': 5,
    'length_penalty': 0.6,
    'return_dict_in_generate': True,
    'output_scores': True
}


def answer(
        text,
        top_p,
        temperature,
        sample=True,
):
    '''
    sample：是否抽样。生成任务，可以设置为True;
    top_p：0-1之间，生成的内容越多样
    '''
    text = preprocess(text)
    encoding = tokenizer(text=[text],
                         truncation=True,
                         padding=True,
                         max_length=1024,
                         return_tensors="pt").to(device)
    if not sample:
        out = model.generate(**encoding,
                             return_dict_in_generate=True,
                             output_scores=False,
                             max_new_tokens=1024,
                             num_beams=1,
                             length_penalty=0.6)
    else:
        out = model.generate(**encoding,
                             return_dict_in_generate=True,
                             output_scores=False,
                             max_new_tokens=1024,
                             do_sample=True,
                             top_p=top_p,
                             temperature=temperature,
                             no_repeat_ngram_size=12)
    # out=model.generate(**encoding, **generate_config)
    out_text = tokenizer.batch_decode(out["sequences"],
                                      skip_special_tokens=True)
    return postprocess(out_text[0])


def clear_session():
    return '', None


def chatyuan_bot(input, history, top_p, temperature, num):
    history = history or []
    if len(history) > num:
        history = history[-num:]

    context = "\n".join([
        f"用户：{input_text}\n小元：{answer_text}"
        for input_text, answer_text in history
    ])

    input_text = context + "\n用户：" + input + "\n小元："
    input_text = input_text.strip()
    output_text = answer(input_text, top_p, temperature)
    print("open_model".center(20, "="))
    print(f"{input_text}\n{output_text}")
    history.append((input, output_text))
    return '', history, history


def chatyuan_bot_regenerate(input, history, top_p, temperature, num):
    history = history or []

    if history:
        input = history[-1][0]
        history = history[:-1]

    if len(history) > num:
        history = history[-num:]

    context = "\n".join([
        f"用户：{input_text}\n小元：{answer_text}"
        for input_text, answer_text in history
    ])

    input_text = context + "\n用户：" + input + "\n小元："
    input_text = input_text.strip()
    output_text = answer(input_text, top_p, temperature)
    print("open_model".center(20, "="))
    print(f"{input_text}\n{output_text}")
    history.append((input, output_text))
    return '', history, history


block = gr.Blocks()

with block as demo:
    gr.Markdown("""<h1><center>元语智能——ChatYuan</center></h1>
        <font size=4>回答来自ChatYuan, 是模型生成的结果, 请谨慎辨别和参考, 不代表任何人观点 | Answer generated by ChatYuan model</font>
        <font size=4>注意：gradio对markdown代码格式展示有限</font>
    """)
    with gr.Row():
        with gr.Column(scale=3):
            chatbot = gr.Chatbot(label='ChatYuan').style(height=400)

        with gr.Column(scale=1):
            num = gr.Slider(minimum=4,
                            maximum=10,
                            label="最大的对话轮数",
                            value=5,
                            step=1)
            top_p = gr.Slider(minimum=0,
                              maximum=1,
                              label="top_p",
                              value=1,
                              step=0.1)
            temperature = gr.Slider(minimum=0,
                                    maximum=1,
                                    label="temperature",
                                    value=0.7,
                                    step=0.1)
            clear_history = gr.Button("👋 清除历史对话 | Clear History")
            send = gr.Button("🚀 发送 | Send")
            regenerate = gr.Button("🚀 重新生成本次结果 | regenerate")
    message = gr.Textbox()
    state = gr.State()
    message.submit(chatyuan_bot,
                   inputs=[message, state, top_p, temperature, num],
                   outputs=[message, chatbot, state])
    regenerate.click(chatyuan_bot_regenerate,
                     inputs=[message, state, top_p, temperature, num],
                     outputs=[message, chatbot, state])
    send.click(chatyuan_bot,
               inputs=[message, state, top_p, temperature, num],
               outputs=[message, chatbot, state])

    clear_history.click(fn=clear_session,
                        inputs=[],
                        outputs=[chatbot, state],
                        queue=False)


block = gr.Blocks()
with block as introduction:
    gr.Markdown("""<h1><center>元语智能——ChatYuan</center></h1>

<font size=4>😉ChatYuan: 元语功能型对话大模型 | General Model for Dialogue with ChatYuan
<br>
👏ChatYuan-large-v2是一个支持中英双语的功能型对话语言大模型，是继ChatYuan系列中ChatYuan-large-v1开源后的又一个开源模型。ChatYuan-large-v2使用了和 v1版本相同的技术方案，在微调数据、人类反馈强化学习、思维链等方面进行了优化。
<br>
ChatYuan large v2 is an open-source large language model for dialogue, supports both Chinese and English languages, and in ChatGPT style.
<br>
ChatYuan-large-v2是ChatYuan系列中以轻量化实现高质量效果的模型之一，用户可以在消费级显卡、 PC甚至手机上进行推理（INT4 最低只需 400M ）。
<br>
在Chatyuan-large-v1的原有功能的基础上，我们给模型进行了如下优化：
- 新增了中英双语对话能力。
- 新增了拒答能力。对于一些危险、有害的问题，学会了拒答处理。
- 新增了代码生成功能。对于基础代码生成进行了一定程度优化。
- 增强了基础能力。原有上下文问答、创意性写作能力明显提升。
- 新增了表格生成功能。使生成的表格内容和格式更适配。
- 增强了基础数学运算能力。
- 最大长度token数扩展到4096。
- 增强了模拟情景能力。.<br>
<br>
Based on the original functions of Chatyuan-large-v1, we optimized the model as follows:
-Added the ability to speak in both Chinese and English.
-Added the ability to refuse to answer. Learn to refuse to answer some dangerous and harmful questions.
-Added code generation functionality. Basic code generation has been optimized to a certain extent.
-Enhanced basic capabilities. The original contextual Q&A and creative writing skills have significantly improved.
-Added a table generation function. Make the generated table content and format more appropriate.
-Enhanced basic mathematical computing capabilities.
-The maximum number of length tokens has been expanded to 4096.
-Enhanced ability to simulate scenarios< br>
<br>
👀<a href='https://www.cluebenchmarks.com/clueai.html'>PromptCLUE-large</a>在1000亿token中文语料上预训练, 累计学习1.5万亿中文token, 并且在数百种任务上进行Prompt任务式训练. 针对理解类任务, 如分类、情感分析、抽取等, 可以自定义标签体系; 针对多种生成任务, 可以进行采样自由生成.  <br> 
<br>
 &nbsp; <a href='https://modelscope.cn/models/ClueAI/ChatYuan-large/summary' target="_blank">ModelScope</a> &nbsp; | &nbsp; <a href='https://huggingface.co/ClueAI/ChatYuan-large-v1' target="_blank">Huggingface</a> &nbsp; | &nbsp; <a href='https://www.clueai.cn' target="_blank">官网体验场</a> &nbsp; | &nbsp; <a href='https://github.com/clue-ai/clueai-python#ChatYuan%E5%8A%9F%E8%83%BD%E5%AF%B9%E8%AF%9D' target="_blank">ChatYuan-API</a> &nbsp; | &nbsp; <a href='https://github.com/clue-ai/ChatYuan' target="_blank">Github项目地址</a> &nbsp; | &nbsp; <a href='https://openi.pcl.ac.cn/ChatYuan/ChatYuan/src/branch/main/Fine_tuning_ChatYuan_large_with_pCLUE.ipynb' target="_blank">OpenI免费试用</a> &nbsp;
</font>
<center><a href="https://clustrmaps.com/site/1bts0"  title="Visit tracker"><img src="//www.clustrmaps.com/map_v2.png?d=ycVCe17noTYFDs30w7AmkFaE-TwabMBukDP1802_Lts&cl=ffffff" /></a></center>
    """)

gui = gr.TabbedInterface(
    interface_list=[introduction, demo],
    tab_names=["相关介绍 | Introduction", "开源模型 | Online Demo"])
# gui.launch(quiet=True, show_api=False, share=True)