Pretty sure most benchmarks are being gamed by people training on the test set d... | Hacker News

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

londons_explore 40 days ago | parent | context | favorite | on: SWE-1.7 Reach Near GPT 5.5 and Opus Intelligence

Pretty sure most benchmarks are being gamed by people training on the test set deliberately or accidentally anyway.

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