Setup guide: TF pool

These steps will set up the environment when you work on the pool computer at the technical faculty.

0. Log in to the pool:

Connect to the pool:

ssh <username>@login.informatik.uni-freiburg.de -p 22 

Ask for a machine with gpu:

ssh tfpoolXX

X corresponds to the number of the machine you want to log in. 
Check the pool_machines.txt to see which ones have gpu

you should have a folder with 10GB of space in: /project/dl2021s/<username> 
if not contact the pool manager to add your username.
go to your folder before running any script.

1. Add the necessary CUDA binaries to the $PATH environment variable:

add the following lines to the ~/.bashrc file in your home directory:

export PATH=/usr/local/cuda-9.2/bin:$PATH

export LD_LIBRARY_PATH=/usr/local/cuda-9.2/lib64:$LD_LIBRARY_PATH

save, and in a terminal run:

source ~/.bashrc

2. Create a python virtual environment to install the necessary python packages.

In a terminal, from a folder of your choice, run:

```virtualenv --no-site-packages -p python3 venv```

```source venv/bin/activate```

3. Install the python packages (please note: this operation requires at least 1.3GB of disk space / disk quota!)

in a terminal run:

```pip install --upgrade pip```

```pip install numpy Pillow```

```pip install torch==1.4.0+cu92 torchvision==0.5.0+cu92 -f https://download.pytorch.org/whl/torch_stable.html```

4. Try importing torch and torchvision from within the virtual environment. From a python interpreter:

```import torch, torchvision```