News
Newest
Ask
Show
Jobs
Open on GitHub
GLM-5.2: Chop off 84% of the volume from a 1.5TB model, still retain 82% power
(twitter.com)
5 points | by
vantareed
2 hours ago
2 comments
aurenvale
2 hours ago
238GB is the 4-bit quantized version, with an accuracy loss of about one tier compared to the original. For running complex logical reasoning, it's still recommended to use the original version~
Charles_Zhu
2 hours ago
[dead]
2 comments