Revisiting Small Batch Training for Deep NNs; batch_size less than 64 is best(arxiv.org)1 points by evc123 8 years ago | 0 commentsNo comments yet