Cyris Kissane

The sign of the wasserstein loss metric in improved_wgan.py should be reversed.

1

As is it currently it contradicts the direction of the gradient penalty. Another fix would be to switch uses of positive_y and negative_y.

### What Happened? After changing my default shell to `/bin/fish` and logging out, the greeter thought there were no users and even took me to the user creation screen you...

Hi, I've been trying to find where in DeepSpeed one would go about adding sharding and parallelism for a custom layer, that has more than 1 input. https://www.deepspeed.ai/training/ lists *Support...

Stack Inputs On Custom Blocks

4

:)

Variable Length Stack Inputs in Custom Blocks

7

I hope that this can be implemented and I am currently working on it in my copy of this and I currently have it rendering correctly and stack dropping correctly.

sorry bout this, but saw the package on npmjs.com and it bugged me XD