阿里通义千问Qwen1.5使用教程_qwen怎么用

1.下载文件

在Hugging Face下载各个版本LLM的文件，然后存入qwen文件夹，每个文件夹下各子文件夹即代表各个版本的llm，模型文件等存储在对应版本的子文件夹下

XXX-GPTQ-Int4的意思是对模型进行了精度为Int4的量化，虽然实测降低了一点生成速度但是显著降低了占用的显存，适合显卡资源不足的研究者（毕竟Qwen1.5这几个小模型就是主打一个显存低啊哈）

1.安装环境

博主python版本为3.8，cuda(cudatoolkit)版本为11.8（版本太低运行时会报错）

pip install transformers==4.37.0
1

2.本地终端运行

from transformers import AutoModelForCausalLM, AutoTokenizer
device = "cuda" # the device to load the model onto

model = AutoModelForCausalLM.from_pretrained(
    "Qwen/Qwen1.5-14B-Chat-AWQ",
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen1.5-14B-Chat-AWQ")

prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(device)

generated_ids = model.generate(
    model_inputs.input_ids,
    max_new_tokens=512
)
generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31

3.web网页端运行

pip install gradio==4.19.2
1

web网页端我使用了比较热门的gradio框架，即官方推的聊天机器人框架:

import gradio as gr
import time

def echo(message, history, system_prompt, tokens):
    response = f"System prompt: {system_prompt}\n Message: {message}."
    for i in range(min(len(response), int(tokens))):
        time.sleep(0.05)
        yield response[: i+1]

demo = gr.ChatInterface(echo, 
                        additional_inputs=[
                            gr.Textbox("You are helpful AI.", label="System Prompt"), 
                            gr.Slider(10, 100)
                        ]
                       )

if __name__ == "__main__":
    demo.queue().launch()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

公网：在demo.queue().launch()中加上share=True运行时即会生成一个随机公网地址（注意别开VPN）

4.添加DIY文件上传功能

pip install PyPDF2, docx
1

import warnings, sys, os, time
import transformers, PyPDF2
import gradio as gr
from docx import Document
warnings.filterwarnings('ignore')
transformers.logging.set_verbosity_error()

def txt_to_str(file_path):
    """
    txt, html, json, py, md, java, c, cpp, h, js, css, xml, yml, yaml, sql, sh, bat
    """
    try:
        with open(file_path, 'r', encoding='utf-8') as file:
            return file.read()
    except Exception as e:
        return f"读取文件时发生错误: {e}"

def pdf_to_str(file_path):
    """
    pdf
    """
    try:
        with open(file_path, 'rb') as file:
            pdf_reader = PyPDF2.PdfReader(file)
            
            # # 检查PDF是否有页
            # if not len(pdf_reader.pages):
            #     return ""

            # 读取所有页面的内容并合并为一个字符串
            content = ""
            for page_num in range(len(pdf_reader.pages)):
                page_obj = pdf_reader.pages[page_num]
                content += page_obj.extract_text()

            # 返回提取的文本
            return content

    except Exception as e:
        return f"读取文件时发生错误: {e}"

def docx_to_str(file_path):
    """
    docx
    """
    try:
        # 打开docx文件
        doc = Document(file_path)

        # 将所有段落内容合并为一个字符串
        content = '\n'.join([paragraph.text for paragraph in doc.paragraphs])

        # 返回提取的文本
        return content

    except Exception as e:
        return f"读取docx文件时发生错误: {e}"

def file_to_str(file_path):
    
    file_types = ['.txt', '.html', '.json', '.py', '.md', '.java', '.c', '.cpp', '.h', '.js', '.css', '.xml', '.yml', '.yaml', '.sql', '.sh', '.bat'] + ['.pdf'] + ['.docx']
    if not file_path.endswith(tuple(file_types)):
        raise TypeError("文件类型不支持")
    if file_path.endswith('.pdf'):
        return pdf_to_str(file_path)
    elif file_path.endswith('.docx'):
        return docx_to_str(file_path)
    else:
        return txt_to_str(file_path)

def list_to_str(histories):
    histories_str = ""
    for i in range(len(histories)):
        user_prompt, response = histories[i]
        histories_str += f"I: {user_prompt}\nYou: {response}\n\n"
    return histories_str

def qwen(user_prompt="", histories=list(), system_prompt="", max_length=512, model_choice="Qwen1.5-0.5B-Chat"):
    if(histories != list()):
        user_prompt = user_prompt + "\n\n这是我们俩之前的聊天记录：{}".format(list_to_str(histories))
    device = "cuda" # the device to load the model onto

    model = transformers.AutoModelForCausalLM.from_pretrained(
        model_choice,
        torch_dtype="auto",
        device_map="auto"
    )
    tokenizer = transformers.AutoTokenizer.from_pretrained(model_choice)
    messages = [
        {"role": "system", "content": system_prompt},
        {"role": "user", "content": user_prompt}
    ]
    text = tokenizer.apply_chat_template(
        messages,
        tokenize=False,
        add_generation_prompt=True
    )
    model_inputs = tokenizer([text], return_tensors="pt").to(device)

    generated_ids = model.generate(
        model_inputs.input_ids,
        max_new_tokens=max_length
    )
    generated_ids = [
        output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
    ]

    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]

    with open("./history/"+find_max("./history"), "a", encoding='utf-8') as f:
        f.write("User: " + user_prompt + "\nResponse: " + response + "\n\n")

    return response

def echo(user_prompt, histories, system_prompt, max_length, model_choice, uploaded_file_path=None):
    if uploaded_file_path:
        file_content = file_to_str(uploaded_file_path)
        # print(f"已成功上传并读取了文件的部分内容：\n{file_content[:70]}...")
        user_prompt += "\n\n这是文件内容:\n" + file_content

    response = qwen(user_prompt, histories, system_prompt, max_length, model_choice)

    for i in range(len(response)):
        time.sleep(0.05)
        yield response[: i+1]

def find_max(folder_path):
    # 使用os.listdir()列出文件夹中的所有条目
    all_entries = os.listdir(folder_path)
    # 创建一个空列表来存放文件名
    file_names = []
    # 遍历列表，检查每个条目是否为文件而非子目录，并将文件名添加到列表中
    for entry in all_entries:
        full_path = os.path.join(folder_path, entry)
        if os.path.isfile(full_path):
            file_names.append(entry)

    return max(file_names)

def main():
    current_time = time.strftime("%Y-%m-%d_%H-%M-%S", time.localtime())
    with open("./history/history_{}.txt".format(current_time), "w", encoding='utf-8') as f:
        f.write("")
    demo = gr.ChatInterface(echo, 
                            additional_inputs=[
                                gr.Textbox("You are a helpful assistant.", label="System Prompt"), 
                                gr.Slider(128, 2048, label="The max length of tokens", interactive=True),
                                gr.Radio([f"程俊凯的辣鸡大模型Qwen1.5-{i}B-Chat" for i in [0.5, 1.8, 4, 4]], label="Model Choice"),
                                gr.UploadButton(label="Upload a file", type="filepath", file_count="single", file_types=["file"], interactive=True, size="sm")
                            ]
                        )
    
    # demo.queue().launch()
    demo.queue().launch(share=True)
#pip install gevent==22.10.2
if __name__ == "__main__":
    main()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158