News for hackers. Stuff that matters.
Re:AA-Omniscience: Evaluating Cross-Domain Knowledge Reliabi... by karty1 on Tuesday November 18, 2025 @12:32AM (#45961720)
I really like this. One of the most realistic evals I've seen, finally quantifying the good vibes many of us feel from the Claude models. Also lmao at CapGPT(-oss).

There are no comments on this comment yet.