Codes for the Generalisation Bounds for Wasserstein Autoencoders, used in the paper, https://openreview.net/forum?id=WjRPZsfeBO&referrer=%5BAuthor%20Console%5D(%2Fgroup%3Fid%3DICLR.cc%2F2024%2FConference%2FAuthors%23your-submissions)
Within each of the folders, you will find a .sh file. You may need to make them executable by running "chmod +x filename.sh" in your terminal to make them executable. We recommend executing these .sh files on a GPU equipped with CUDA for optimal performance.
- First run the "data_generate.sh" in the "Data_generate" folder. It will regenerate the data called, "data_16.pt" and "data_16.pt" in the same folder.
- Copy and paste the files, "data_2.pt" and "data_16.pt" to each of the other four folders (i.e. except "Data_generate")
- In each of the folders, run the "run_models.sh" files to run the models on the grid as described in the paper.
- To regenerate the plots, you can run the code provided in the plot.ipynb file located within each respective folder. You may also run these ipynb files without completing steps 1,2 &3 as we have also provided our resulting tables as .pt files in each of the folders.