Enrico Shippole

> The discord link you provided doesn't seem to be valid Hmmm. It worked when I tried it just now. Is there a discord name I can just @ you...

> @conceptofmind hello! I'm `squared_circle` on discord. I don't know what server and channel you're referring to, it's probably a private server :). Out of curiosity, what is the status...

> > > @conceptofmind hello! I'm `squared_circle` on discord. I don't know what server and channel you're referring to, it's probably a private server :). Out of curiosity, what is...

> Hi there, I’m just following up on this. > > Is there progress related to DSP integration? Or at least an open pull request. > > I’m hoping to...

@zxcvqwerasdf GLM-130B was pre-trained over 400 billion tokens on a cluster of 96 NVIDIA DGX-A100 (8×40G) GPU nodes between May 6 and July 3, 2022. The number of tokens is...

@Bachstelze That is difficult to **properly** evaluate at a smaller scale without sufficient pre-training. It would likely require a single go over something like OpenWebText or Wikitext if you wanted...

> Sorry, we can't find the library where `FlashAttentionWrapper` is located, could you please tell us which library it is? My apologies. I thought I had linked it previously. Here...

Had to fix a test.

There might be one slight issue with Lint or something. I will have to check tomorrow.

Resolve conflict with newly added import in update