From bbbd38908fd90bb6d6ac1d37d194bcfe15a02afe Mon Sep 17 00:00:00 2001
From: Hauke Kirchner <hauke.gronenberg@gwdg.de>
Date: Thu, 15 Aug 2024 12:30:55 +0000
Subject: [PATCH] Update README.md

---
 README.md | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 14e272c..73680e6 100644
--- a/README.md
+++ b/README.md
@@ -77,6 +77,12 @@ To set up the conda environment, first log into the frontend. Then create a [con
 ```
 git clone https://gitlab-ce.gwdg.de/hpc-team-public/deep-learning-with-gpu-cores.git 
 cd  deep-learning-with-gpu-cores/code
+
+# load new software stack
+# see https://docs.hpc.gwdg.de/software/software_stacks/index.html
+export PREFERRED_SOFTWARE_STACK=nhr-lmod
+source /sw/etc/profile/profile.sh
+
 module load miniconda3
 conda create -n dl-gpu python=3.8
 ```
@@ -95,11 +101,11 @@ Now we need to install all packages. They are located in `code/requirements.txt`
 To install all dependencies in your environment, activate the environment and install the requirements with these commands:
 
 ```
-conda activate dl-gpu
+source activate dl-gpu
 pip install -r requirements.txt
 ```
 
-💎 If `conda activate dl-gpu` does not work, use either (1) `source activate dl-gpu` or (2) run `conda init bash`, log out and log in again and then it should be initialized and work.
+💎 If `source activate dl-gpu` does not work, use either (1) `conda activate dl-gpu` or (2) run `conda init bash`, log out and log in again and then it should be initialized and work.
 
 💎 If you get something similar to `ERROR: No matching distribution found for ...`, one solution is to ease the requirements. We pinned the exact versions of each libarary. You can do this with `nano code/requirements.txt` and removing the version number for a package, so instead of `package==versionnumber` only `package` is displayed. If you do not require a specific version number, compatibility issues can be avoided, but it might also break things later on (sometimes, code requires a specific version number, so sometimes, it requires a little experimenting with versions).
 
-- 
GitLab