# 新 Config 适配贡献指南

## 适配指南

教程: <https://mmengine--1071.org.readthedocs.build/zh\\_CN/1071/advanced\\_tutorials/config.html#python>

1. 找一个自己感兴趣的算法，把老 config 转换成新 config，并通过验证脚本的校验。反馈转换过程中遇到的问题
2. 基于新的 config 跑通训练、测试、和 demo，并观察 dump 下来的配置文件是否合理
3. 观察新 config 的报错信息是否直观，并反馈不合理的报错信息

## 环境配置

### 创建环境

创建一个新的 python 环境，以 conda 为例

```bash
conda create -n new_config python=3.10 # 3.7 以上其他版本也可以
```

按照官网教程安装 PyTorch

### 安装指定分支的 MMEngine

```bash
git clone -b new_config_type https://github.com/HAOCHENYE/mmengine.git mmengine_new_config
cd mmengine_new_config
# 如果是新环境，没有安装 MMEngine 相关的依赖，需执行
pip install -e .
```

### 源码编译 MMCV

```bash
git clone https://github.com/open-mmlab/mmcv.git
cd mmcv
MMCV_WITH_OPS=1 pip install -e .
```

### 算法库适配

基于 new\_config 分支，源码编译算法库

```bash
# mmdet
https://github.com/open-mmlab/mmdetection/tree/new_config

# mmyolo
https://github.com/open-mmlab/mmyolo/tree/new_config

# mmpretrain
https://github.com/open-mmlab/mmpretrain/tree/new_config

# mmpose
https://github.com/open-mmlab/mmpose

# mmrazor
https://github.com/open-mmlab/mmrazor/tree/new_config
```

打样 PR： <https://github.com/open-mmlab/mmdetection/pull/10366> <https://github.com/open-mmlab/mmyolo/pull/787> <https://github.com/open-mmlab/mmrazor/pull/539> <https://github.com/open-mmlab/mmpose/pull/2390> <https://github.com/open-mmlab/mmpretrain/pull/1567>

## 验证脚本

<https://gist.github.com/HAOCHENYE/079fb5125d9fa8ce113bf3da81ae5d95>

对比两个 config：

\--new: 转换后的 config

\--old：老 config

\--scope：算法库的 scope

> 验证脚本不一定适用于所有 config，有时候报错信息需要具体分析，可以在群里交流

1. 校验新老两个配置文件

```bash
python config --new ${new_config} --old ${old_config} --scope ${scope}
```

2. 校验转换后所有的配置文件 要求老的配置文件在 configs 目录下，新的配置文件的 {scope}/configs 目录下

```bash
python config --scope ${scope}
```

## read\_base 语法更新

全局替换即可

![image](https://github.com/open-mmlab/OpenMMLabCamp/assets/62195058/07d0a047-a43b-49d8-9d88-4c12b0e78387)

## 反馈收集：

## 交流群二维码：

![image](https://github.com/open-mmlab/OpenMMLabCamp/assets/62195058/039f6a5b-a9a0-438f-b401-499b52f658d1)


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://mmsig.openmmlab.com/mm/mmengine/config.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.