Re:DaisyRose-27B: 1,200-task RL took Qwen3.8 from 5.6% to 12...
by bavell on Wednesday November 18, 2026 @08:43PM
(#50555030)
Instead of pre-training, I'd say it's more like a better architecture, training regime, higher fidelity data, and perhaps better initial "weights".
[ Parent
| on HN ]
Re:DaisyRose-27B: 1,200-task RL took Qwen3.8 from 5.6% to 12...
by LoganDark on Wednesday November 18, 2026 @08:49PM
(#50555082)
All of that stuff was still trained by evolution, though.