Loading…
Radar · Entity
The item argues that compute caps in standard evaluations can materially affect measured agent capability, task difficulty, and the apparent pace of frontier progress.
No tracked model releases for this entity yet.
No linked research papers yet.
1 item on this desk · stories, not raw articles, newest first