Gwern Branwen

> Question2. Could you share what the sensitive hyperparameter for WGAN are? For example : weight initialization scale(0.02), clamping threshold (0.01), batch size(64), model size, #D/G step(5), lr(0.00005) I can...

No, I've never used it. Just deepomf mentioned it as a way to improve on the existing segmentation without extremely laborious hand-annotation or using weak supervision from a classifier/tagger like...

tttp, any chance you could provide that code? (I've been arguing with the author of my HTML generator, asking that the section divs be annotated with a class of some...

Thanks for the example; I managed to get the simpler example page working for gwern.net. It's pretty nice, all in all, although the shapes of the sausages indicates to me...

FWIW, [I got similar errors](https://www.gwern.net/GPT-2-preference-learning). There were two problems: the `mpiexec` command was calling Python 2 instead of Python 3, and I had to hardwire it to call `python3`. The...

1. yours 2. currently not public, working on getting permission 3. great! 4. nsheppard adds self-attention layers, so it's comparable to SAGAN or BigGAN in not just being vanilla resnet...

Is there any update on this? Or easy way I could help? Background: I originally used text-tdfa on gwern.net for regexp tests & rewrites back in 2019 or so with...

Have those already been released? I see on Hackage the last `text` upload date is 2021-12-24 but the UTF8 change was merged 2021-09-08 and the implicit-fusion disabling 2021-06-21, if I'm...

It would be great if ar5iv was integrated - because a separate site will never be used by more than like 5% of readers (if that), same way so many...

I can do that if necessary. It's not as clean a rewrite (because it is no longer just transforming the prefix but appending, and it needs to handle anchors), but...