Deep recurrent models are less robustly CoT-monitorable than normal CoT models in a toy setting
by Nick Kuhn, evhu, and Alek Westover
We use RL to teach a deep recurrent model and a normal CoT model to solve a math problem while hiding from a CoT monitor which of two possible problems it is solving. The deep recurrent model moves its reasoning into latents (thereby evading the CoT monitor) within 40 steps...
Sep 1885