myron
myron
thank you for the reply, cudnn.benchmark selects the best kernel variant , if we don't use it we may have 20% performance drop. I don't think it's an acceptable long...
I see, very good, thank you guys
Looks like it failed a few tests. One of them is the number of parameters for Unet Looks like our Unet implementation has Instancenorm as the default normalization https://github.com/Project-MONAI/MONAI/blob/dev/monai/networks/nets/unet.py#L124 as...
Alright, looks like this doesn't have support for legacy reasons, I'm going to close this
I'm not sure what it means
I think it's related to the new MetaTensor somehow, plz see my issue here https://github.com/Project-MONAI/MONAI/issues/5283
> Hi Nic. We could indeed assume MetaTensor for everything, and that makes life a lot simpler internally. The main issue I see with that is that people who want...
added check for pytorch version
some checks are failing, but it's something unrelated
``` import torch import time from monai.networks.layers import GaussianFilter def old_version(patch_size, approximate=True): sigmas = [1.0, 2.0, 3.0] device = 'cuda:0' center_coords = [i // 2 for i in patch_size] importance_map...