Open ihamdi opened 2 years ago
Maybe, you tried to train the model with an image size smaller than the image size used in the automatic argument-search. Image size should be more than that.
Or, I think there is a way to change the get_params_dependent_on_targets
function of arguments/augments/dropout/coarse_dropout.py
so that the randint
function has the correct value.
I'm trying to use the json file generated by autoaugment but I keep getting "ValueError: empty range for randrange() (0,-xx, -xx)"
The autoalbument is still training and generating new json for each epoch so I tried many epochs hoping randrange would be fixed at some point but its still the same.