Is the last number of plot_top_losses no longer the probability of the ‘actual class’?
I noticed that it now returns a high number, which I am assuming is the probability assigned to the incorrectly chosen classification?
This is reflected in the tutorial found here:
http://dev.fast.ai/tutorial.vision
as apposed to v1 docs:
https://docs.fast.ai/vision.learner.html#_cl_int_plot_top_losses
I’m sorry if this is explained somewhere, but I could not find a method description for plot_top_losses in the dev.fast.ai docs.