arXivJason StaraceSun, Jun 7, 2026, 2:14 AM PDT
score 16.1
How AI scaffolding masks true model capability in benchmarks
Original: Scaffold Effects on GAIA: A Controlled Comparison
Source: arxiv.org ↗
Writing ELI5 summary…
Original: Scaffold Effects on GAIA: A Controlled Comparison
Source: arxiv.org ↗
Writing ELI5 summary…