Metadata-Version: 2.1
Name: rag_retrieval
Version: 0.1.1
Summary: A unified API for various RAG Retrieval  re-ranking models.
Author-email: A grass in the car <suda.jcli@qq.com>
Maintainer-email: A grass in the car <suda.jcli@qq.com>
License: MIT License
        
        Copyright (c) 2024
        
        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.
Project-URL: Homepage, https://github.com/NLPJCL/RAG-Retrieval
Keywords: reranking,retrieval,rag,nlp
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3 :: Only
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: pydantic
Requires-Dist: tqdm
Provides-Extra: all
Requires-Dist: transformers; extra == "all"
Requires-Dist: torch; extra == "all"
Requires-Dist: requests; extra == "all"
Provides-Extra: transformers
Requires-Dist: transformers; extra == "transformers"
Requires-Dist: torch; extra == "transformers"
Provides-Extra: api
Requires-Dist: requests; extra == "api"
Provides-Extra: dev
Requires-Dist: pytest; extra == "dev"

<h1 align="center">RAG-Retrieval</h1>
<p align="center">
    <a href="https://github.com/NLPJCL/RAG-Retrieval">
            <img alt="Build" src="https://img.shields.io/badge/Contribution-Welcome-blue">
    </a>
    <a href="https://github.com/NLPJCL/RAG-Retrieval/blob/master/LICENSE">
        <img alt="License" src="https://img.shields.io/badge/LICENSE-MIT-green">
    </a>
</p>
<h4 align="center">
    <p>
        <a href=#创建环境>创建环境</a> |
        <a href="#微调模型">微调模型</a> |
        <a href=#实验结果>实验结果</a> |
        <a href="#license">License</a> 
    <p>
</h4>

RAG-Retrieval 提供了全链路的RAG检索微调代码，支持微调任意开源的RAG检索模型，包括向量（embedding、图a）、迟交互式模型（colbert、图d）、交互式模型（cross encoder、图c）。
![ColBERT](pictures/models.png)
# 最新更新

- 3/18/2024:发布RAG-Retrieval [RAG-Retrieval知乎介绍](https://zhuanlan.zhihu.com/p/683483778)

# 创建环境
    
```bash
conda create -n rag-retrieval python=3.8 && conda activate rag-retrieval
#为了避免自动安装的torch与本地的cuda不兼容，建议进行下一步之前先手动安装本地cuda版本兼容的torch。
pip install -r requirements.txt 
```


# 微调模型

##  向量（embedding）模型
- 支持微调任意开源的embedding模型（bge,m3e等等）

- 支持对两种数据进行微调：
    -  query和正例（负例采用batch内随机负例），
    -  query和正例以及难负例。（负例为对应的难负例，以及batch内随机负例）

微调embedding模型流程
```bash
cd ./rag-retrieval/embedding
bash train_embedding.sh
```

## 迟交互式（colbert）模型

- 支持微调开源的bge-m3e模型中的colbert。
- 支持query和正例以及难负例。（负例为对应的难负例，以及batch内随机负例）

微调colbert模型流程
```bash
cd ./rag-retrieval/colbert
bash train_colbert.sh
```
## 排序（reranker,cross encoder）模型
- 支持微调任意开源的reranker模型（例如，bge-rerank、bce-rerank等）
- 支持两种数据进行微调：
    - query和doc的相关性为二分类（1代表相关、0代表不相关）
    - query和doc的相关性为四分类。（3，2，1，0，相关性依次降低。）

微调reranker模型流程
```bash
cd ./rag-retrieval/reranker
bash train_reranker.sh
```


# 实验结果

说明：结果仅供参考，训练模型只是为了验证RAG-Retrieval的训练代码是否正确。相比的两个模型，都是很好的开源模型。而且他们都是多语言模型，而笔者训练的模型仅仅支持中文。

## reranker模型在 MTEB Reranking 任务的结果


|      **Model**       |  **Model Size(GB)**  |**T2Reranking** | **MMarcoReranking** | **CMedQAv1** | **CMedQAv2** | **Avg** |
|:-----------:|:----------:|:----------:|:-------------:|:--------------:|:---------------:| :---------------:|
|   bge-reranker-base   |  1.11 | 67.28    |      35.46     |      81.27      |       84.10      | 67.03
| bce-reranker-base_v1 |   1.11 |70.25    |      34.13     |      79.64      |       81.31      | 66.33
| rag-retrieval-reranker |  0.41 | 67.33    |      31.57     |      83.54     |       86.03     | 67.12

其中，rag-retrieval-reranker是我们使用RAG-Retrieval代码在hfl/chinese-roberta-wwm-ext模型上训练所得，训练数据使用bge-rerank模型的训练数据.

## colbert模型在 MTEB Reranking 任务的结果

|      **Model**  | **Model Size(GB)**  | **Dim**  | **T2Reranking** | **MMarcoReranking** | **CMedQAv1** | **CMedQAv2** | **Avg** |
|:-----------: |:----------:|:----------:|:----------:|:-------------:|:--------------:|:---------------:| :---------------:|
|   bge-m3-colbert   | 2.24 | 1024 | 66.82 | 26.71    |      75.88     |      76.83      |      61.56      
| rag-retrieval-colbert | 0.41 |  1024|  66.85    |      31.46     |      81.05     |       84.22     | 65.90

其中，rag-retrieval-colbert是我们使用RAG-Retrieval代码在hfl/chinese-roberta-wwm-ext模型上训练所得，训练数据使用bge-rerank模型的训练数据.


## 用领域内数据微调开源的BGE系列模型

|      **Model**  | **T2ranking**  | |
|:-----------: |:----------:|:----------:|
|   bge-v1.5-embedding   | 66.49|  | 
|   bge-v1.5-embedding **finetune**    | 67.15 | **+0.66** | 
|   bge-m3-colbert   | 66.82|  | 
|   bge-m3-colbert **finetune**    | 67.22 | **+0.40** | 
|   bge-reranker-base   | 67.28|  | 
|   bge-reranker-base  **finetune**    | 67.57 | **+0.29** | 

后面带有finetune的代表我们使用RAG-Retrieval在对应开源模型的基础上继续微调所得，训练数据使用T2-Reranking的训练集。

## License
RAG-Retrieval is licensed under the [MIT License](https://github.com/NLPJCL/RAG-Retrieval/blob/master/LICENSE). 
