SHMCU

Thanks for sharing this code and the excellent work! I am still having questions: How to approximate argmax with gumbel softmax? The output of the argmax should be one channel...

> Have u done the training process successfully now? In fact I meet problem when I try evaluating it...the codes stuck when import encoding and did not return any error...

I have moved forward. For this paper, I learned their interesting ideas. Thanks for their contribution.  Best On Thursday, December 26, 2019, 12:02:47 AM PST, dreamPoet wrote: Have u done...

You can just remove this "return input". And there are a few other "return input, input", remove them as well.

In your configuration, do you set use_apex=True ? Thanks, SHMCU

Hi Qibin, Could you share your configuration file for this spnet50 achieving mIoU 79.5 on Cityscapes val set? Thanks, Hai

If it is trained on multiple GPUs, did you enable the sync_batchnorm ? On Sunday, November 14, 2021, 06:11:28 AM PST, cm0561 ***@***.***> wrote: Thanks for the valuable contribution! I...

What is the batchsize if you are training on one GPU. If it is larger than 12, then it should be fine.  On Sunday, November 14, 2021, 05:31:38 PM PST,...

Okay, try to train with batchsize larger than 12. You can either choose to shrink  the width of the layers or train at lower resolution or change to a smaller...

Hi Marvin, I wrote a script to train the convCRF using nll loss. I treat the air plane image as a two class segmentation problem. At the beginning the training...