Consider that deepseek v3 and r1 have the same base model and other than the CoT RL they were likely optimized with the same intentions, but r1 developed much more personality. i only hear about…

Posters: repligate. repligate

3 likes recorded by the source corpus.