1 paper · 1 filter
Ankush Checkervarty
Activation probes that monitor deployed language models are trained on synthetic conversations, and how many a probe needs is open. We trace learning curves over 10-590 synthetic s…