I set n_training_epochs to args.epochs, where args.epochs is 1. Of course, I plan to change this after I complete the fast development. However, it appears that parameteric_umap.fit_transform() leads to 10 epochs regardless of how I setup args.epochs. I shared a couple of pictures.
Here is the initialized boilerplate for parameteric umap training.:
Here is the output of the training using args.epochs=1:
How do I set a desired number of epochs for training that isn't a multiple of 10?
I set
n_training_epochstoargs.epochs, whereargs.epochsis 1. Of course, I plan to change this after I complete the fast development. However, it appears that parameteric_umap.fit_transform() leads to 10 epochs regardless of how I setup args.epochs. I shared a couple of pictures.Here is the initialized boilerplate for parameteric umap training.:
Here is the output of the training using
args.epochs=1:How do I set a desired number of epochs for training that isn't a multiple of 10?