参考
在服务器上安装不同版本的pytorch
先创建一个虚拟环境
1
|
conda create -n pytorch1.0 python=3.7
|
切换到该虚拟环境(这样就不会影响原有的环境了)
1
|
conda activate pytorch1.0
|
方法一：
1
|
conda install pytorch=1.0.0 torchvision -c pytorch
|
这样就会安装
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
|
Collecting package metadata (current_repodata.json): done
Solving environment: done
Please update conda by running
$ conda update -n base -c defaults conda
## Package Plan ##
environment location: /home/wag/anaconda3/envs/pytorch1.0
added / updated specs:
- pytorch=1.0.0
- torchvision
The following NEW packages will be INSTALLED:
blas anaconda/pkgs/main/linux-64::blas-1.0-mkl
cffi anaconda/pkgs/main/linux-64::cffi-1.14.4-py37h261ae71_0
freetype anaconda/pkgs/main/linux-64::freetype-2.10.4-h5ab3b9f_0
intel-openmp anaconda/pkgs/main/linux-64::intel-openmp-2020.2-254
jpeg anaconda/pkgs/main/linux-64::jpeg-9b-h024ee3a_2
lcms2 anaconda/pkgs/main/linux-64::lcms2-2.11-h396b838_0
libpng anaconda/pkgs/main/linux-64::libpng-1.6.37-hbc83047_0
libtiff anaconda/pkgs/main/linux-64::libtiff-4.1.0-h2733197_1
lz4-c anaconda/pkgs/main/linux-64::lz4-c-1.9.2-heb0550a_3
mkl anaconda/pkgs/main/linux-64::mkl-2020.2-256
mkl-service anaconda/pkgs/main/linux-64::mkl-service-2.3.0-py37he8ac12f_0
mkl_fft anaconda/pkgs/main/linux-64::mkl_fft-1.2.0-py37h23d657b_0
mkl_random anaconda/pkgs/main/linux-64::mkl_random-1.1.1-py37h0573a6f_0
ninja anaconda/pkgs/main/linux-64::ninja-1.10.2-py37hff7bd54_0
numpy anaconda/pkgs/main/linux-64::numpy-1.19.2-py37h54aff64_0
numpy-base anaconda/pkgs/main/linux-64::numpy-base-1.19.2-py37hfa32c7d_0
olefile anaconda/pkgs/main/linux-64::olefile-0.46-py37_0
pillow anaconda/pkgs/main/linux-64::pillow-8.0.1-py37he98fc37_0
pycparser anaconda/pkgs/main/noarch::pycparser-2.20-py_2
pytorch pytorch/linux-64::pytorch-1.0.0-py3.7_cuda9.0.176_cudnn7.4.1_1
six anaconda/pkgs/main/linux-64::six-1.15.0-py37h06a4308_0
torchvision pytorch/noarch::torchvision-0.2.2-py_3
zstd anaconda/pkgs/main/linux-64::zstd-1.4.5-h9ceee32_0
|
注意
1
2
|
#此命令会安装
pytorch pytorch/linux-64::pytorch-1.0.0-py3.7_cuda9.0.176_cudnn7.4.1_1
|
所以会直接安装pytorch1.0.0
这个命令也会直接将安装cuda9.0.176和cudnn7.4.1_1（具体的原因不清楚），所以后面也不用重新安装与pytorch对应的cuda版本了。
注：这种方法是通过conda安装的不完整CUDA
常用指令：
1
2
3
4
5
6
7
8
9
10
11
12
|
检查PyTorch版本
torch.__version__ # PyTorch version
torch.cuda.is_available()#是否能够使用cuda加速
torch.version.cuda # Corresponding CUDA version
torch.backends.cudnn.version() # Corresponding cuDNN version
torch.cuda.get_device_name(0) # GPU type
|
更新：采用上述方法似乎有点问题，报错
1
|
RuntimeError: cuda runtime error (11) : invalid argument at /opt/conda/conda-bld/pytorch_1544176307774/work/aten/src/THC/THCGeneral.cpp:405
|
1
2
3
|
vi ~/.bashrc
source ~/.bashrc
|