In this Spike I review the use of TabNet by Google AI using the Pytorch implementation by Dreamquark-AI on Timeseries data from Flight-Pasanger Dataset
make shell: start a command line
make jupyter: start a Jupyter-Lab server
Both commands will mount a volume on the projects' directory.
tabnet-does-not-learn-interesting-features.ipynb: Trains/Validate/Test models with different set of features
test-gpu.ipynb: checks the GPUs are accessible inside the container.