Log on to the CoCoMo GPU server

1. Connect to one of these:
• MPI VPN
• The Wi-Fi hotspot (SSID: animal2vec Workshop, PW: a2v_is_fun)
• Use one of the Ethernet cables (Recommended)
  
  
2. Go to https://kn-eas-gpusvr01.top.ab.mpg.de
   
   
3. Log-in using your account (sent to you prior to this workshop)

• Datasets, model weights, animal2vec code, embeddings, and other scripts are located in:
• /home/USERNAME/workshop_materials
  
  
• Optional: Use ./workshop_materials/misc/mount_eas.sh to mount the EAS storage server
  
  
• Look around, set up a new directory for your own things

Set up the python environment

1. Navigate to /home/USERNAME/workshop_materials/misc
• You can use: cd ~/workshop_materials/misc
  
  
2. In there, there is a file called a2v_conda_env.yaml
• First, initialize conda: conda init bash
• Then, log out and in again
• Then, create the environment using: conda env create -f a2v_conda_env.yaml
• Activate it using: conda activate a2v
  
  
• Activate a Jupyter kernel via: python -m ipykernel install --user --name a2v --display-name "Python (a2v)"
  
  
3. Navigate to /home/USERNAME/workshop_materials/animal2vec_code
• You can use: cd ~/workshop_materials/animal2vec_code
  
  
4. Install the animal2vec package
• You can use: python -m pip install -e .
• All these commands are documented in the official README.md file
  
  

• From hereon, every time you log in just do: conda activate a2v