Scholar Feed

LLM reasoning / chain-of-thought

external CoT monitoring

Superseded baseline#289 of 772 most-superseded

Cited as a baseline — critiqued by newer work, not yet beaten on a benchmark here

1 papers critique it · 0 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites external CoT monitoring as a baseline.

However, a critical concern is that if models are directly penalized for showing deceptive thoughts, they might hide those thoughts rather than truly abandoning the deception.
— Mitigating Deceptive Alignment via Self-Monitoring

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.