# kunpeng-extension-for-pytorch

**Repository Path**: tedeasonwang/kunpeng-extension-for-pytorch

## Basic Information

- **Project Name**: kunpeng-extension-for-pytorch
- **Description**: No description available
- **Primary Language**: Python
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 6
- **Created**: 2025-07-31
- **Last Updated**: 2025-09-23

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

<!--  Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
 
 Licensed under a modified version of the MIT license. See LICENSE in the project root for license information.
 
 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. -->

# kunpeng-extension-for-pytorch

#### 介绍
kunpeng-extension-for-pytorch is a package for extending the official Pytorch that can easily obtain performance on Kunpeng platform

#### 安装教程

**一、环境设置**

1.在鲲鹏平台编译安装pytorch（[pytorch使能KML操作指南](https://gitee.com/kunpeng-hpc/kunpeng-extension-for-pytorch/wikis/PyTorch%20v2.5.0%20%E5%AF%B9%E6%8E%A5KML)）

2.加载kutacc环境

```
module use /xxx/xxx/xxx/xxx/xxx/modulefiles
module load xxx/xxx/kutacc
```

3.安装依赖

```
pip install ninja=1.11.1.1 pybind11==2.11.1
```
**二、执行安装命令**
```
CFLAGS="-stdlib=libc++ -lc++abi" KPEX_BUILD_TYPE=release KUTACC_ROOT=/xxx/xxx/kutacc pip install --editable .
```
#### 卸载教程
```
pip uninstall kunpeng-pytorch-extension -y
```
#### 接口说明
接口名:
kpex_alphafold

接口描述:
通过替换模型中的算子来优化afphafold2的模型

接口参数:
|参数名|类型|描述|输入/输出|
|-----|----|-----|---------|
|model|模型类|alphafold2模型|输入|
|model_config|配置类|alphafold2模型的config|输入|
|new_model|模型类|优化后的alphafold2模型|输出|

#### 使用说明

```
model = kpex.tpp.alphafold.alphafold.kpex_alphafold(model, model_config)
```