yudw15
Results
3
issues of
yudw15
1.conv2d剪枝时为什么只考虑m1.weight.data而不考虑m1.bias.data? 2.maxpool剪枝时会xx.append('M')有什么用？
Params and GFLops of hrnet_w32 is very low, but it takes much time to train fcos_hrnet_w32, at the same time, very high GPU usage. Is this normal?