pytorch-syncbn

pytorch-syncbn copied to clipboard

I have not seen any differences between m1 and m2, Why?

2

part of your code: m1 = nn.Sequential( nn.Conv2d(3, 3, 1, 1, bias=False), nn.BatchNorm2d(3), nn.ReLU(inplace=True), nn.Conv2d(3, 3, 1, 1, bias=False), nn.BatchNorm2d(3), ).cuda() torch.manual_seed(123) init_weight(m1) m2 = nn.Sequential( nn.Conv2d(3, 3, 1, 1,...

tjpulkl

subprocess.CalledProcessError: Command '['which', 'c++']' returned non-zero exit status 1

2

I follow the installation in readme.md. But still got a problem when running the `test.py` file. My torch version is 1.1.0, cudnn is 7.6.0, on ubuntu18.04. ``` Traceback (most recent...

koncle

When running models in mixed prediction, pytorch-syncbn cannot be used as it does not support fp16.

antorsae

It would be easier to access this package when it's added as a pip package. Thanks.

ericlearning

thanks so much for your code! But I want to use sync3d, what shoud I do?

zyl1336110861

I run the test.py with the number of gpus > 4 and the gpu memory increase all the time. The speed of calculation becomes slow. Any suggestion?

justttry