HNGlance

Top Best New Show

Lossless model compression experiment: GLM-5.2 in 25% less memory

https://brianbell-x.github.io/weight-compression/

by hambandit • 2 days ago

17 1 2 days ago

BR

brianbell-x.github.io

Related Stories

Loading related stories...

Comments

1

Sanzig• 2 days ago

So, if I'm understanding correctly, this is just for lower bandwidth transfers of full BF16 weights over the wire, not for serving, correct? Have you benchmarked the performance against a SOTA general purpose compression algorithm like zstd?

Also, are all that many people handling the BF16 weights directly? GLM-5.2's reference deployment is FP8, and many vendors are even serving at NVFP4 which seems to offer negligible degradation over the FP8 reference deployments.

Source preview

brianbell-x.github.io

HNGlance

© 2024-2026 HN Glance. Story content belongs to its original sources.

Privacy
Terms