this post was submitted on 05 Mar 2025
8 points (100.0% liked)

Technology

1060 readers
37 users here now

A tech news sub for communists

founded 2 years ago
MODERATORS
muad_dibber@lemmygrad.ml
8
QwQ-32B is a 32 billion parameter language model achieves comparable performance to DeepSeek-R1 with 671 billion parameters, using reinforcement learning for scaling (qwenlm.github.io)
submitted 3 days ago* (last edited 3 days ago) by yogthos@lemmygrad.ml to c/technology@lemmygrad.ml
3 comments fedilink hide all child comments
top 3 comments
sorted by: hot top controversial new old
[–] marl_karx@lemmygrad.ml 2 points 1 day ago (1 children)

Isnt deepseek based on qwen? at least the distilled models?

[–] yogthos@lemmygrad.ml 2 points 1 day ago

I think so, but this looks like an update of qwen with some new tricks.

[–] yogthos@lemmygrad.ml 6 points 3 days ago* (last edited 3 days ago)

can grab it here

I find it absolutely wild how quickly we went from needing a full blown data centre to run models of this scale to being able to run them on a laptop.