Use recreate_results.ipynb to test our best performing model on the gtex test dataset. In order to run this notebook, you are required to clone to repository to the DTU Computing Center HPC infrastructure. This is required since the large dataset and the model checkpoints are stored there.
- Connect to the HPC via 'ssh sXXXXXX@login1.hpc.dtu.dk'
Also the following packages are required to run the notebook
- torch
- numpy
- sklearn
- tqdm
- pickle
If you want to explore the entirety of all scripts, we recommend installing miniconda and the VAE-env2 environment. See requirements.txt for guide