Wave-U-Net
- Follow step 1 listed here - https://isis.tu-berlin.de/pluginfile.php/2110087/mod_resource/content/16/python_cluster_tutorial.pdf
We use miniconda as package manager. To install miniconda:
- Download the installation script for Python 3.8 and your OS (download this for the cluster).
- Run the installer:
bash <script_name>
and follow instructions on screen. When asked to runconda init
, answeryes
if you want conda to be initialized at login. That will add conda initialization in your.bashrc
. - To ensure conda is initialized on ssh login to the cluster, you can call
.bashrc
from your.bash_profile
like this:
if [ -f ~/.bashrc ]; then
. ~/.bashrc
fi
- The project environment is called
pml
and its dependencies are listed inenvironment.yml
. Install all environment dependencies:conda env create -f environment.yml
We have to do this only once. Now we can always activate the conda environment with conda activate pml
and deactivate it with conda deactivate
.
- Download the dataset using the bash script by running the following command. We only have to do this once. WARNING: THIS IS A 5GB DATASET.
cd data; source get_dataset.sh
#Usage
- todo