Galedra
Sign in Sign up

Evidence item

The peer-reviewed Nature paper reports that an LLM paired with an evaluator discovered previously unknown cap-set constructions and improved bin-packing heuristics, producing verifiable results beyond prior best-known solutions. DIRECT_TEXT

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 15221

Source: Mathematical discoveries from program search with large language models · locator {} · excerpt hash sha256:953a3833c94bca1c51e25ea42a2fd3f6b3d6721da2948893dd330cb4de53223e

Exact excerpt
Applying FunSearch ... we discover new constructions of large cap sets going beyond the best-known ones ... This shows that it is possible to make discoveries for established open problems using LLMs. ... we also find new heuristics for online bin packing that improve on widely used baselines.

Linked claims

ClaimDirectionRelevanceStepsCounted
7964e7b7SUPPORTSTRONG1yes

Audits

None.