News for hackers. Stuff that matters.

A comment on Qwen 3.8 27B on my laptop can interpret model weights, too.

Re:Qwen 3.8 27B on my laptop can interpret model weights, too by daluan on Wednesday November 04, 2026 @12:18PM (#50368631)
From my experience with finetuning Gemma 12B, I noticed that 4-bit quantization would hurt instruction-following especially when it comes to NLP.
[ Parent | on HN ]

There are no comments on this comment yet.