README.md 998 Bytes
Newer Older
sangwz's avatar
sangwz committed
1
# Uni-Core
Guolin Ke's avatar
Guolin Ke committed
2

sangwz's avatar
sangwz committed
3
Uni-Core 专为快速创建高性能 PyTorch 模型而构建,尤其是基于 Transfromer 的模型。详细信息可参考[README_ORIGIN.md](README_ORIGIN.md)
Guolin Ke's avatar
Guolin Ke committed
4

sangwz's avatar
sangwz committed
5
# 安装
Guolin Ke's avatar
Guolin Ke committed
6

sangwz's avatar
sangwz committed
7
组件支持:
Guolin Ke's avatar
Guolin Ke committed
8

sangwz's avatar
sangwz committed
9
* Python >= 3.7
Guolin Ke's avatar
Guolin Ke committed
10

sangwz's avatar
sangwz committed
11
## 使用pip方式安装
Guolin Ke's avatar
Guolin Ke committed
12

sangwz's avatar
sangwz committed
13
从http://10.6.10.68:8000/customized/ 下载Uni-core安装包,选择对应torch、python版本的whl包
Guolin Ke's avatar
Guolin Ke committed
14

sangwz's avatar
sangwz committed
15
16
17
```bash
pip install unicore*.whl
```
Guolin Ke's avatar
Guolin Ke committed
18

sangwz's avatar
sangwz committed
19
## 源码编译方式安装
Guolin Ke's avatar
Guolin Ke committed
20

sangwz's avatar
sangwz committed
21
确认环境中已安装torch,并安装fastpt工具。从[Index of /debug/fastpt/](http://10.6.10.68:8000/debug/fastpt/)中下载对应python版本的安装包,执行
Guolin Ke's avatar
Guolin Ke committed
22

sangwz's avatar
sangwz committed
23
24
25
```bash
pip install fastpt*.whl
```
Guolin Ke's avatar
Guolin Ke committed
26

sangwz's avatar
sangwz committed
27
下载Uni-Core源码,并执行安装:
Guolin Ke's avatar
Guolin Ke committed
28

sangwz's avatar
sangwz committed
29
30
31
32
33
34
```bash
# 代码下载之后,注意切换分支到develop或其它目标分支
git clone http://developer.hpccube.com/codes/OpenDAS/Uni-Core.git
cd Uni-Core
python setup.py install
```
Guolin Ke's avatar
Guolin Ke committed
35

sangwz's avatar
sangwz committed
36
# 验证
Guolin Ke's avatar
Guolin Ke committed
37

sangwz's avatar
sangwz committed
38
执行以下代码查询软件版本号,验证是否安装完成:
Guolin Ke's avatar
Guolin Ke committed
39

sangwz's avatar
sangwz committed
40
`python -c "import unicore;print(unicore.__version__)"`
Guolin Ke's avatar
Guolin Ke committed
41
42
43