audioset_tagging_cnn
audioset_tagging_cnn copied to clipboard
Factor of 2 in count_flops.py
In the function count_flop.py, when calculating kernel_ops and weight_ops for convolution and linear layers: why do you need to multiply them by 2, if multiply_adds=True? Shouldn't that be the other way around, 1 MAC = 2 FLOPs roughly?