arXivLinas Nasvytis, Simon Jerome Han, Ben Prystawski, Satchel Grant, Noah D. Goodman, Judith E. FanWed, May 27, 2026, 10:01 AM PDT
score 16.5
Language models learn reasoning faster by comparing successes and failures
Original: CORE: Contrastive Reflection Enables Rapid Improvements in Reasoning
Source: arxiv.org ↗
Writing ELI5 summary…