Post

개발 환경 설정 (Slurm, Conda, Pytorch)

Posted
By Yoojin Kim
1 min read
개발 환경 설정 (Slurm, Conda, Pytorch)

Slurm Basic

connect to slurm

First you need :

  1. slurm account
  2. connection to TUD VPN
1

ssh $username@slurm.ukp.informatik.tu-darmstadt.de

Install Miniconda3 under working directory

1
2

#working directory 
cd /storage/athene/work/$username

Conda Installation Guide

“$username” shall be changed!

1
2
3
4

mkdir -p /storage/athene/work/$username/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O /storage/athene/work/$username/miniconda3/miniconda.sh
bash /storage/athene/work/$username/miniconda3/miniconda.sh -b -u -p /storage/athene/work/$username/miniconda3/miniconda3
rm /storage/athene/work/$username/miniconda3/miniconda3/miniconda.sh

After installation

1

source /storage/athene/work/$username/miniconda3/bin/activate

Creating Conda Env and Install Pytorch

1
2
3

conda create --name pytorch310 python=3.10
conda activate pytorch310
conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia

After activating pytorch310(you can name this conda env as you want)

Before (base) [user@slurm]
After (pytorch310) [user@slurm]

Thesis
tud
This post is licensed under CC BY 4.0 by the author.