Hacker News new | ask | show | jobs
by daemonologist 638 days ago
On the second point, you're comparing MMMU-Pro (multimodal) to MMLU-Pro (text only). I don't think they published scores on MMLU-Pro for 3.2.

(Edit: parent comment was corrected, thanks!)

1 comments

Yep you're right, thanks for catching (sorry for the ninja edit!)