My guess is that it is over fitted to the tests. | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		bearjaws 9 months ago \| parent \| context \| favorite \| on: Xiaomi MiMo Reasoning Model My guess is that it is over fitted to the tests.

revel 9 months ago [–]

They used RFT and there's only so many benchmarks out there, so I would be very surprised if they didn't train on the tests.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact