Currently, at #11, multiple autoencoder architectures were created, with different numbers of neurons in the latent layer. Copy these to the model folder, so these can be used in Singularity.
This is a bit of a hack compared to the cleaner #10, as #10 allows one to call any Mx.json file, where this Issues ruthlessly supplies them. #10 , however, is quite some work, as GCAE assumes relative paths from the run_gcae.py file.
richelbilderbeek
commented
2 years ago
Currently, at #11, multiple autoencoder architectures were created, with different numbers of neurons in the latent layer. Copy these to the
modelfolder, so these can be used in Singularity.This is a bit of a hack compared to the cleaner #10, as #10 allows one to call any
Mx.jsonfile, where this Issues ruthlessly supplies them. #10 , however, is quite some work, as GCAE assumes relative paths from therun_gcae.pyfile.