flâneur

An update on our general capability evaluations - METR

metr.org · saved by 2 readers

More tasks, human baselines, and preliminary results for GPT-4 and Claude.

saved by