Skip to content

Tensorboard logging is happening per step instead of per epoch

This is the current result for a 20-epoch training in my machine:

Screenshot_2023-07-11_at_23.02.52

The "epoch" number is not showing on this, as you can see, but rather the "step" (?). I also don't know how to make this looks nicer, eg by combining both train/valid losses and make the plots more colourful.