News for hackers. Stuff that matters.
Re:GLM-5.2 is the new leading open weights model on Artifici... by scrlk on Wednesday June 17, 2026 @08:09AM (#48569261)
IME, unquantised -> FP8 is pretty much lossless. What matters more is having an unquantized KV cache - using an FP8 KV cache can result in a significant drop in quality.
Re:GLM-5.2 is the new leading open weights model on Artifici... by ComputerGuru on Wednesday June 17, 2026 @10:23AM (#48570972)
Do infra providers reveal that level of implementation detail?
Re:GLM-5.2 is the new leading open weights model on Artifici... by scrlk on Wednesday June 17, 2026 @10:46AM (#48571282)
I've seen a few articles from providers talking about KV cache quantisation, but it's not something they explicitly point out like they do with weights.

So you could end up paying more for unquantised weights, only to get silently hit with a quantised KV cache...

Re:GLM-5.2 is the new leading open weights model on Artifici... by johnnyApplePRNG on Wednesday June 17, 2026 @11:28AM (#48571842)
>unquantised -> FP8 is pretty much lossless

Claude Shannon is rolling in his grave.

Re:GLM-5.2 is the new leading open weights model on Artifici... by gpm on Wednesday June 17, 2026 @03:18PM (#48575401)
I don't know, sounds quite similar to his rate distortion theorem (analyzing minimum number of bits/symbol you need to stay under some fixed amount of distortion). I.e. lossy compression with a maximum amount of loss. I.e. "pretty much lossless" compression.

https://en.wikipedia.org/wiki/Rate%E2%80%93distortion_theory (

Re:GLM-5.2 is the new leading open weights model on Artifici... by regularfry on Thursday June 18, 2026 @04:25AM (#48582437)
"Pretty much" doing a lot of work. But it's kinda analogous to 99% JPEG compression: yes you can detect loss, but you get meaningful compression ratios out of it and the subjective appearance is nigh-on perfect.

Shannon would be pointing out that if you can throw away half the model without apparent degradation, we're nowhere near packing in all the information we could in training. There must be a better arrangement than we've currently got.

Re:GLM-5.2 is the new leading open weights model on Artifici... by osti on Wednesday June 17, 2026 @07:25PM (#48578438)
The official API is FP8, which should imply that it's lossless.