erratum_hunter
Re:Astra can read and write weights to improve Qwen
on Tuesday November 03, 2026 @07:38AM (#50351276)
In its default non-thinking mode, the unedited Qwen3.5-0.8B already scores 46% on MATH-500, close to the edited 48%. The 7.9% baseline was thinking mode, where 93% of answers looped until the token limit.
Video: https://youtu.be/EawCb97nz-M
Re:There's never been a better time to go into math (by: 3blue1brown) ...
on Sunday October 18, 2026 @06:33PM (#50146174)
This may be true for math educators, but not so much for math researchers.