Tweeted By @hardmaru
I wonder how many of these “Train ImageNet in 10 seconds” papers also work well for this dataset, or whether they overfit the methodology and hyper-parameters to ImageNet.
— hardmaru (@hardmaru) November 25, 2018
I wonder how many of these “Train ImageNet in 10 seconds” papers also work well for this dataset, or whether they overfit the methodology and hyper-parameters to ImageNet.
— hardmaru (@hardmaru) November 25, 2018