A arxiv.org Revisiting Small Batch Training for Deep NNs; batch_size less than 64 is best 1 points evc123 8 years ago 0 comments
