Where is the model stored after ''Train with Sparsity'' or ''Baseline''? about network-slimming HOT 6 CLOSED

about the argu --resume?

from network-slimming.

It is stored in args.save directory. See here.

from network-slimming.

It is stored in args.save directory. See here.

i see the default of arg.save is ''/logs''. My first time using pytorch, Thanks for your reply,

from network-slimming.

It is stored in args.save directory. See here.

Hello, I have retrain the model to get a complete ckpt and the training process read normal
Test set: Average loss: 0.2486, Accuracy: 9413/10000 (94.0%)

But still encounter some problem during pruning when setting the model path argu as
--model ./logs/checkpoint.pth.tar

And it shows
`=> loading checkpoint './logs/checkpoint.pth.tar
=> loaded checkpoint './logs/checkpoint.pth.tar' (epoch 160) Prec1: 0.000000
Traceback (most recent call last):
File "denseprune.py", line 73, in
mask = weight_copy.gt(thre).float().cuda()
RuntimeError: Expected object of backend CUDA but got backend CPU for argument #2 'other'``

from network-slimming.

Which version of Pytorch are you using? Our code is developed on Pytorch 3.1.

from network-slimming.

In my case, replacing mask = weight_copy.gt(thre).float().cuda() with mask = weight_copy.gt(thre.cuda()).float().cuda() works.

from network-slimming.