docs(mvp): 添加项目基础文档

- 新增 LICENSE 文件，定义项目开源许可证
- 新增 README.md 文件，介绍项目的基本信息和预期技术栈

.gitignore

@@ -11,7 +11,6 @@ env/
 # IDE
 .vscode/
 .idea/
-.roo
 
 # Environment files
 .env

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2022 C-a-r-r-y
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -0,0 +1,15 @@
+# 基于文档驱动的自适应编码大模型微调框架
+## 简介
+本人的毕业设计,这个是mvp分支（MVP 是指最小可行产品Minimum Viable Product），其他功能在master分支中
+### 项目概述
+
+* 通过深度解析私有库的文档以及其他资源，生成指令型语料，据此对大语言模型进行针对私有库的微调。
+
+### 项目技术（预计）
+
+* 使用unsloth框架在GPU上实现大语言模型的qlora微调
+* 使用langchain框架编写工作流实现批量生成微调语料
+* 使用tinydb和sqlite实现数据的持久化
+* 使用gradio框架实现前端展示
+
+**施工中......**

config/llm.py

@@ -0,0 +1,15 @@
+import os
+from dotenv import load_dotenv
+from typing import Dict, Any
+
+def load_config() -> Dict[str, Any]:
+    """从.env文件加载配置"""
+    load_dotenv()
+    
+    return {
+        "openai": {
+            "api_key": os.getenv("OPENAI_API_KEY"),
+            "base_url": os.getenv("OPENAI_BASE_URL"),
+            "model_id": os.getenv("OPENAI_MODEL_ID")
+        }
+    }

dataset_generator.py

@@ -0,0 +1,94 @@
+import os
+import json
+from tools.parse_markdown import parse_markdown, MarkdownNode
+from tools.openai_api import generate_json_via_llm
+from prompt.base import create_dataset
+from config.llm import load_config
+from tqdm import tqdm
+
+def process_markdown_file(file_path):
+    with open(file_path, 'r', encoding='utf-8') as f:
+        content = f.read()
+    
+    root = parse_markdown(content)
+    results = []
+    
+    def traverse(node, parent_titles):
+        current_titles = parent_titles.copy()
+        current_titles.append(node.title)
+        
+        if not node.children:  # 叶子节点
+            if node.content:
+                full_text = ' -> '.join(current_titles) + '\n' + node.content
+                results.append(full_text)
+        else:
+            for child in node.children:
+                traverse(child, current_titles)
+    
+    traverse(root, [])
+    return results
+
+def find_markdown_files(directory):
+    markdown_files = []
+    for root, dirs, files in os.walk(directory):
+        for file in files:
+            if file.endswith('.md'):
+                markdown_files.append(os.path.join(root, file))
+    return markdown_files
+
+def process_all_markdown(doc_dir):
+    all_results = []
+    
+    markdown_files = find_markdown_files(doc_dir)
+    for file_path in markdown_files:
+        results = process_markdown_file(file_path)
+        all_results.extend(results)
+    
+    return all_results
+
+def save_dataset(dataset, output_dir):
+    os.makedirs(output_dir, exist_ok=True)
+    output_path = os.path.join(output_dir, 'dataset.json')
+    with open(output_path, 'w', encoding='utf-8') as f:
+        json.dump(dataset, f, ensure_ascii=False, indent=2)
+
+if __name__ == "__main__":
+    # 解析markdown文档
+    results = process_all_markdown('workdir/my_docs')
+    
+    # 加载LLM配置
+    config = load_config()
+    
+    dataset = []
+    # 使用tqdm包装外部循环以显示进度条
+    for content in tqdm(results, desc="生成数据集进度", unit="文档"):
+        for _ in range(3):
+            prompt = create_dataset.create(
+                "LLaMA-Factory",  # 项目名
+                content,  # 文档内容
+                """{
+            "dataset":[
+                {
+                    "question":"",
+                    "answer":""
+                }
+            ]
+        }"""
+            )
+            
+            # 调用LLM生成JSON
+            try:
+                result = generate_json_via_llm(
+                    prompt=prompt,
+                    base_url=config["openai"]["base_url"],
+                    api_key=config["openai"]["api_key"],
+                    model_id=config["openai"]["model_id"]
+                )
+                print(json.loads(result)["dataset"])
+                dataset.extend(json.loads(result)["dataset"])
+            except Exception as e:
+                print(f"生成数据集时出错: {e}")
+
+    # 保存数据集
+    save_dataset(dataset, 'workdir/dataset2')
+    print(f"数据集已生成，共{len(dataset)}条数据")

db/__init__.py

@@ -1,3 +0,0 @@
-from .init_db import get_engine, initialize_db
-
-__all__ = ['get_engine', 'initialize_db']

db/init_db.py

@@ -1,79 +0,0 @@
-from sqlmodel import SQLModel, create_engine, Session
-from sqlmodel import select
-from typing import Optional
-import os
-from pathlib import Path
-import sys
-from dotenv import load_dotenv
-from sqlalchemy.engine import Engine
-
-# 将项目根目录添加到系统路径中，以便能够导入项目中的其他模块
-sys.path.append(str(Path(__file__).resolve().parent.parent))
-from schema.dataset_generation import APIProvider
-
-# 全局变量，用于存储数据库引擎实例
-_engine: Optional[Engine] = None
-
-def get_engine(workdir: str) -> Engine:
-    """
-    获取数据库引擎实例。如果引擎尚未创建，则创建一个新的引擎并返回。
-
-    Args:
-        workdir (str): 工作目录路径，用于确定数据库文件的存储位置。
-
-    Returns:
-        Engine: SQLAlchemy 数据库引擎实例。
-    """
-    global _engine
-    if not _engine:
-        # 创建数据库目录（如果不存在）
-        db_dir = os.path.join(workdir, "db")
-        os.makedirs(db_dir, exist_ok=True)
-        # 定义数据库文件路径
-        db_path = os.path.join(db_dir, "db.sqlite")
-        # 创建数据库URL
-        db_url = f"sqlite:///{db_path}"
-        # 创建数据库引擎
-        _engine = create_engine(db_url)
-    return _engine
-
-def initialize_db(engine: Engine) -> None:
-    """
-    初始化数据库，创建所有表结构，并插入初始数据（如果不存在）。
-
-    Args:
-        engine (Engine): SQLAlchemy 数据库引擎实例。
-    """
-    # 创建所有表结构
-    SQLModel.metadata.create_all(engine)
-    # 加载环境变量
-    load_dotenv()
-    # 从环境变量中获取API相关配置
-    api_key = os.getenv("API_KEY")
-    base_url = os.getenv("BASE_URL")
-    model_id = os.getenv("MODEL_ID")
-    
-    # 如果所有必要的环境变量都存在，则插入初始数据
-    if api_key and base_url and model_id:
-        with Session(engine) as session:
-            # 查询是否已存在APIProvider记录
-            statement = select(APIProvider).limit(1)
-            existing_provider = session.exec(statement).first()
-            
-            # 如果不存在，则插入新的APIProvider记录
-            if not existing_provider:
-                provider = APIProvider(
-                    base_url=base_url,
-                    model_id=model_id,
-                    api_key=api_key
-                )
-                session.add(provider)
-                session.commit()
-
-if __name__ == "__main__":
-    # 定义工作目录路径
-    workdir = os.path.join(os.path.dirname(__file__), "..", "workdir")
-    # 获取数据库引擎
-    engine = get_engine(workdir)
-    # 初始化数据库
-    initialize_db(engine)

frontend/chat_page.py

@@ -1,9 +0,0 @@
-import gradio as gr
-
-def chat_page():
-    with gr.Blocks() as demo:
-        gr.Markdown("## 聊天")
-        with gr.Row():
-            with gr.Column():
-                pass
-    return demo

frontend/setting_page.py

@@ -1,35 +0,0 @@
-import gradio as gr
-from typing import List, Dict
-from sqlmodel import Session, select
-from db import get_engine
-from schema import APIProvider
-import os
-
-# 获取数据库引擎
-engine = get_engine(os.path.join(os.path.dirname(__file__), "..", "workdir"))
-
-def setting_page():
-    def get_providers() -> List[List[str]]:
-        with Session(engine) as session:
-            providers = session.exec(select(APIProvider)).all()
-            return [
-                [p.id, p.model_id, p.base_url, p.api_key or ""]
-                for p in providers
-            ]
-
-    with gr.Blocks() as demo:
-        gr.Markdown("## API Provider 管理")
-        
-        with gr.Row():
-            # API Provider列表
-            with gr.Column(scale=2):
-                provider_table = gr.DataFrame(
-                    headers=["id" , "model id", "URL", "API Key"],
-                    datatype=["number","str", "str", "str"],
-                    interactive=True,
-                    value=get_providers(),
-                    wrap=True,
-                    col_count=(4, "fixed")
-                )
-        
-    return demo

frontend/train_page.py

@@ -1,9 +0,0 @@
-import gradio as gr
-
-def train_page():
-    with gr.Blocks() as demo:
-        gr.Markdown("## 微调")
-        with gr.Row():
-            with gr.Column():
-                pass
-    return demo

main.py

@@ -1,24 +0,0 @@
-import gradio as gr
-from frontend.setting_page import setting_page
-from frontend.chat_page import chat_page
-from frontend.train_page import train_page
-from db import initialize_db as init_db,get_engine
-
-if __name__ == "__main__":
-    init_db(get_engine("workdir"))
-
-    setting_demo = setting_page()
-    chat_demo = chat_page()
-    train_demo = train_page()
-
-    with gr.Blocks() as app:
-        gr.Markdown("# 基于文档驱动的自适应编码大模型微调框架")
-        with gr.Tabs():
-            with gr.TabItem("微调"):
-                train_demo.render()
-            with gr.TabItem("聊天"):
-                chat_demo.render()
-            with gr.TabItem("设置"):
-                setting_demo.render()
-    
-    app.launch()

prompt/base.py

@@ -0,0 +1,25 @@
+class create_dataset:
+    """用于生成微调数据集模板的类"""
+    
+    template = """
+项目名为:{}
+请依据以下该项目官方文档的部分内容，创造合适的对话数据集用于微调一个了解该项目的小模型的语料，要求兼顾文档中间尽可能多的信息点，使用中文
+文档节选：{}
+按照如下json格式返回：{}
+"""
+
+    @staticmethod
+    def create(*args: any) -> str:
+        """根据提供的任意数量参数生成数据集模板
+        
+        Args:
+            *args: 任意数量的参数，将按顺序填充到模板中
+            
+        Returns:
+            格式化后的模板字符串
+        """
+        return create_dataset.template.format(*args)
+    
+
+if __name__=="__main__":
+    print(create_dataset.create("a", "b", "c"))

requirements.txt

@@ -1,4 +1,2 @@
 openai>=1.0.0
-python-dotenv>=1.0.0
+python-dotenv>=1.0.0
-pydantic>=2.0.0
-gradio>=3.0.0

schema/__init__.py

@@ -1,4 +0,0 @@
-from .dataset import *
-from .dataset_generation import APIProvider, LLMResponse, LLMRequest
-from .md_doc import MarkdownNode
-from .prompt import promptTempleta

schema/dataset.py

@@ -0,0 +1,9 @@
+from pydantic import BaseModel, RootModel
+from typing import List
+
+class QAPair(BaseModel):
+    question: str
+    response: str
+
+class QAArray(RootModel):
+    root: List[QAPair]

schema/dataset_generation.py

@@ -1,51 +0,0 @@
-from datetime import datetime, timezone
-from typing import Optional
-from sqlmodel import SQLModel, Relationship, Field
-
-class APIProvider(SQLModel, table=True):
-    id: Optional[int] = Field(default=None, primary_key=True)
-    base_url: str = Field(..., description="API的基础URL")
-    model_id: str = Field(..., description="API使用的模型ID")
-    api_key: Optional[str] = Field(default=None, description="用于身份验证的API密钥")
-    created_at: datetime = Field(
-        default_factory=lambda: datetime.now(timezone.utc),
-        description="记录创建时间"
-    )
-
-class LLMResponse(SQLModel):
-    timestamp: datetime  = Field(
-        default_factory=lambda: datetime.now(timezone.utc),
-        description="响应的时间戳"
-    )
-    response_id: str = Field(..., description="响应的唯一ID")
-    tokens_usage: dict = Field(default_factory=lambda: {
-        "prompt_tokens": 0,
-        "completion_tokens": 0,
-        "prompt_cache_hit_tokens": None,
-        "prompt_cache_miss_tokens": None
-    }, description="token使用信息")
-    response_content: dict = Field(default_factory=dict, description="API响应的内容")
-    total_duration: float = Field(default=0.0, description="请求的总时长，单位为秒")
-    llm_parameters: dict = Field(default_factory=lambda: {
-        "temperature": None,
-        "max_tokens": None,
-        "top_p": None,
-        "frequency_penalty": None,
-        "presence_penalty": None,
-        "seed": None
-    }, description="API的生成参数")
-
-class LLMRequest(SQLModel):
-    prompt: str = Field(..., description="发送给API的提示词") 
-    provider_id: int = Field(foreign_key="apiprovider.id")
-    provider: APIProvider = Relationship()
-    format: Optional[str] = Field(default=None, description="API响应的格式")
-    response: list[LLMResponse] = Field(default_factory=list, description="API响应列表")
-    error: Optional[list[str]] = Field(default=None, description="API请求过程中发生的错误信息")
-    total_duration: float = Field(default=0.0, description="请求的总时长，单位为秒")
-    total_tokens_usage: dict = Field(default_factory=lambda: {
-        "prompt_tokens": 0,
-        "completion_tokens": 0,
-        "prompt_cache_hit_tokens": None,
-        "prompt_cache_miss_tokens": None
-    }, description="token使用信息")

schema/md_doc.py

@@ -1,13 +0,0 @@
-from pydantic import BaseModel, Field
-from typing import List, Optional
-
-class MarkdownNode(BaseModel):
-    level: int = Field(default=0, description="节点层级")
-    title: str = Field(default="Root", description="节点标题")
-    content: Optional[str] = Field(default=None, description="节点内容")
-    children: List['MarkdownNode'] = Field(default_factory=list, description="子节点列表")
-
-    class Config:
-        arbitrary_types_allowed = True
-
-MarkdownNode.model_rebuild()

tools/openai_api.py

@@ -0,0 +1,69 @@
+import json
+from openai import OpenAI
+
+def generate_json_via_llm(
+    prompt: str,
+    base_url: str,
+    api_key: str,
+    model_id: str
+) -> str:
+    client = OpenAI(
+        api_key=api_key,
+        base_url=base_url
+    )
+    
+    try:
+
+        
+        response = client.chat.completions.create(
+            model=model_id,
+            messages=[
+                {
+                    "role": "user",
+                    "content": prompt
+                }
+            ],
+            response_format={
+                'type': 'json_object'
+            }
+        )
+        
+        return response.choices[0].message.content
+    except Exception as e:
+        raise RuntimeError(f"API请求失败: {e}")
+
+
+if __name__ == "__main__":
+    import sys
+    import os
+    sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
+    from config.llm import load_config
+    # 将项目根目录添加到 sys.path 中
+
+    # 示例用法
+    try:
+        config = load_config()
+        print(config)
+        result = generate_json_via_llm(
+            prompt="""测试，随便生成点什么，返回json格式的字符串,格式如下
+{
+    "dataset":[
+        {
+            "question":"",
+            "answer":""
+        },
+        {
+            "question":"",
+            "answer":""
+        }
+        ......
+    ]
+}
+            """,
+            base_url=config["openai"]["base_url"],
+            api_key=config["openai"]["api_key"],
+            model_id=config["openai"]["model_id"],
+        )
+        print(result)
+    except Exception as e:
+        print(f"错误: {e}")

tools/parse_markdown.py

@@ -1,24 +1,28 @@
 import re
-import sys
-from pathlib import Path
 
-# 添加项目根目录到sys.path
+class MarkdownNode:
-sys.path.append(str(Path(__file__).resolve().parent.parent))
+    def __init__(self, level=0, title="Root"):
-from schema import MarkdownNode
+        self.level = level
+        self.title = title
+        self.content = ""  # 使用字符串存储合并后的内容
+        self.children = []
 
-def add_child(parent, child):
+    def __repr__(self):
-    parent.children.append(child)
+        return f"({self.level}) {self.title}"
 
-def print_tree(node, indent=0):
+    def add_child(self, child):
-    prefix = "│  " * (indent - 1) + "└─ " if indent > 0 else ""
+        self.children.append(child)
-    print(f"{prefix}{node.title}")
+
-    if node.content:
+    def print_tree(self, indent=0):
-        content_prefix = "│  " * indent + "├─ [内容]"
+        prefix = "│  " * (indent - 1) + "└─ " if indent > 0 else ""
-        print(content_prefix)
+        print(f"{prefix}{self.title}")
-        for line in node.content.split('\n'):
+        if self.content:
-            print("│  " * indent + "│  " + line)
+            content_prefix = "│  " * indent + "├─ [内容]"
-    for child in node.children:
+            print(content_prefix)
-        print_tree(child, indent + 1)
+            for line in self.content.split('\n'):
+                print("│  " * indent + "│  " + line)
+        for child in self.children:
+            child.print_tree(indent + 1)
 
 def parse_markdown(markdown):
     lines = markdown.split('\n')
@@ -47,10 +51,10 @@ def parse_markdown(markdown):
         if match:
             level = len(match.group(1))
             title = match.group(2)
-            node = MarkdownNode(level=level, title=title, content="", children=[])
+            node = MarkdownNode(level, title)
             while stack[-1].level >= level:
                 stack.pop()
-            add_child(stack[-1], node)
+            stack[-1].add_child(node)
             stack.append(node)
         else:
             if stack[-1].content:
@@ -61,9 +65,9 @@ def parse_markdown(markdown):
 
 if __name__=="__main__":
     # 从文件读取 Markdown 内容
-    with open("workdir/example.md", "r", encoding="utf-8") as f:
+    with open("example.md", "r", encoding="utf-8") as f:
         markdown = f.read()
 
     # 解析 Markdown 并打印树结构
     root = parse_markdown(markdown)
-    print_tree(root)
+    root.print_tree()

trainer.py

@@ -0,0 +1,70 @@
+from unsloth import FastLanguageModel
+import torch
+
+# 基础配置参数
+max_seq_length = 4096 # 最大序列长度
+dtype = None # 自动检测数据类型
+load_in_4bit = True # 使用4位量化以减少内存使用
+
+# 加载预训练模型和分词器
+model, tokenizer = FastLanguageModel.from_pretrained(
+    model_name = "workdir\model\Qwen2.5-3B-Instruct-bnb-4bit", # 选择Qwen2.5 32B指令模型
+    max_seq_length = max_seq_length,
+    dtype = dtype,
+    load_in_4bit = load_in_4bit,
+)
+
+model = FastLanguageModel.get_peft_model(
+    model,
+    r = 64, # LoRA秩,控制可训练参数数量
+    target_modules = ["q_proj", "k_proj", "v_proj", "o_proj",
+                      "gate_proj", "up_proj", "down_proj",], # 需要训练的目标模块
+    lora_alpha = 64, # LoRA缩放因子
+    lora_dropout = 0, # LoRA dropout率
+    bias = "none", # 是否训练偏置项
+    use_gradient_checkpointing = "unsloth", # 使用梯度检查点节省显存
+    random_state = 114514, # 随机数种子
+    use_rslora = False, # 是否使用稳定版LoRA
+    loftq_config = None, # LoftQ配置
+)
+
+from unsloth.chat_templates import get_chat_template
+# 配置分词器使用qwen-2.5对话模板
+tokenizer = get_chat_template(
+    tokenizer,
+    chat_template="qwen-2.5",
+)
+
+def formatting_prompts_func(examples):
+    """格式化对话数据的函数
+    Args:
+        examples: 包含对话列表的字典
+    Returns:
+        包含格式化文本的字典
+    """
+    questions = examples["question"]
+    answer = examples["answer"]
+    
+    # 将Question和Response组合成对话形式
+    convos = [
+        [{"role": "user", "content": q}, {"role": "assistant", "content": r}]
+        for q, r in zip(questions, answer)
+    ]
+    
+    # 使用tokenizer.apply_chat_template格式化对话
+    texts = [
+        tokenizer.apply_chat_template(convo, tokenize=False, add_generation_prompt=False)
+        for convo in convos
+    ]
+    
+    return {"text": texts}
+
+from unsloth.chat_templates import standardize_sharegpt
+
+# 加载数据集
+from datasets import load_dataset
+dataset = load_dataset("json", data_files="workdir\dataset\dataset.json")
+dataset = dataset.map(formatting_prompts_func, batched = True)
+
+print(dataset[5])
+print(dataset[5]["text"])