# Fairseq

Fairseq(-py)是一个序列建模工具包，允许研究人员和开发人员训练自定义模型，用于翻译、摘要、语言建模和其他文本生成任务。

## 安装步骤

按照如下步骤，下载并安装好torch相关的库文件，并安装fairseq安装依赖的环境，并进行编译安装

```
wget https://cancon.hpccube.com:65024/directlink/4/pytorch/dtk22.10/torch-1.10.0a0+git2040069.dtk2210-cp38-cp38-manylinux2014_x86_64.whl

wget https://cancon.hpccube.com:65024/directlink/4/vision/dtk22.10/torchvision-0.10.0a0+e04d001.dtk2210-cp38-cp38-manylinux2014_x86_64.whl

wget https://cancon.hpccube.com:65024/directlink/4/torchaudio/dtk22.10/torchaudio-0.10.0+git9dcc7a1.dtk2210-cp38-cp38-manylinux2014_x86_64.whl

pip3 install torch-1.10.0a0+git2040069.dtk2210-cp38-cp38-manylinux2014_x86_64.whl

pip3 install torchvision-0.10.0a0+e04d001.dtk2210-cp38-cp38-manylinux2014_x86_64.whl

pip3 install torchaudio-0.10.0+git9dcc7a1.dtk2210-cp38-cp38-manylinux2014_x86_64.whl

pip3 install requirements.txt

cd fairseq
pip3 install --editable ./
```