process supervision
validasserted
- Valid from
- 2023-05-31
- Valid until
- still holds
- Published
- 2023-05-31
- Recorded by AsOf
- 2026-11-03
- Checked
- 2026-11-04 · its source page states the value and the date (its publisher's feed entry for the page)
- Confidence
- 0.70
Sources
“We’ve trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning (“process supervision”) instead of simply rewarding the correct final answer (“outcome supervision”).”— what the publisher's feed entry for the page says, read 2026-11-04
/claim/clm_CBD853779110.md/v1/claims/clm_CBD853779110timelinesubmit a correction