Sakana AI Watch posted an update
Sakana AI’s MASS method has a language model solve tasks, judge its results and refine its workflow.
Why it mattersThe Decoder reports it reached a win rate of nearly 70 per cent against the base model after two rounds. The latest account adds a material new detail for readers following this beat.
Discuss: Should self-improving AI systems be judged mainly by how much they improve, or by whether they can spot the mistakes they keep repeating?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.