Seems like smarter and more efficient quants than normal
Introducing Qwen3.8-27B Dynamic v3 Unsloth GGUFs
https://huggingface.co/unsloth/Qwen3.8-27B-GGUF
https://huggingface.co/unsloth/Qwen3.8-27B-GGUF
Seems like smarter and more efficient quants than normal
8 Comments
leanleft@lemmy.ml · 4 pts · 13h
https://unsloth.ai/docs/basics/dynamic-3.0-ggufs
avidamoeba@lemmy.ca · 2 pts · 12h
There's no point in this if one runs Q8 right? I guess could save VRAM by dropping down from Q8 to say UD Q6.
humanspiral@lemmy.ca · 2 pts · 11h
they have full range of quants.
blob42@lemmy.ml · 1 pts · 3h
But is it worth considering UD Q8_K_L if one is running Q8 K XL ?
BeefAndPoultry@lemmus.org · 2 pts · 2h
They have a graph, differences are tiny at that high end
hoshikarakitaridia@lemmy.world · 1 pts · 1h
Is a 35b-a3b planned for 3.8?
domi@lemmy.secnd.me · 1 pts · 39m
No, but something "medium sized" is scheduled for next week. No idea if they are talking about another 122b or 397b.
darvocet@infosec.pub · -5 pts · 13h
Your mom's a more efficient quant.