xvdp

large resampling kernels slow ALSO on the forward pass

1

### 🐛 Describe the bug -- i understand i still have to respond to my PR on kernel creation speed (sorry about that!) - but I found another problem when...

Resample kernel creation uses loops...

9

### 🐛 Describe the bug `torchaudio\functional\functionalpy def _get_sinc_resample_kernel()` is slower than it should be because it uses loops instead to taking advantange of torch broadcasing. A simple rewrite will make...

LayerNorm appears different from Ba 2016

1

Looking at your ConvNextBlock I noticed your LayerNorm is different from torch.nn.LayerNorm(). In fact yours appears to normalize over the dimension of the Channels. In Liu et al 2022 they...

possible error in the warp code

3

repro - - capture a head **as still as possible** - not one's own but a subject against a wall looking at fixed point. in my test the starting point...

in place operations only breaks FiveCrop.

1

Hi, I was writing some augmentation transforms and I noticed that all operations are done in place. I forked and added a .copy() method for my use, but I wonder...

Hi, 1. Do you have pretrained models - I don't see them linked in the github ? It would be great to have those 2. So, to test your model...

fixes to hsic.py

2

Ok, this is the correct branch to create pull request -- same coment as the one i closed. this one has only one file. Hi Kurt, I used 3 functions...

This is a minor change which does alter default behaviour. To understand model behaviours on my local setup - images per second, cuda load, as well as numbers on the...

Do you have an updated link? the pretrained model link returns 404 thanks

base speaker TTS training codes needed for fair comparison

4

the distance between the base speaker and extracted tonecolor makes the result not too good. The alternative, using openais Base Speaker is not satisfactory for an "open" project. Are you...