Revisiting Small Batch Training for Deep NNs; batch_size less than 64 is best arxiv.org 1 points by evc123 8 years ago · 0 comments Reader PiP Save No comments yet.