method
active
method:cot-monitorCoT Monitor
Named method for monitoring chain-of-thought text to detect when the model signals its answer, compared against activation probes
Neighborhood — ranked by edge-count
Papers (1)
paper
Frameworks (1)
framework
- The conceptual framework introduced by the paper distinguishing performative CoT from genuine reasoning using activation probing
Findings (1)
finding
- Comparative finding establishing activation probing as superior to text-level monitoring for early belief detection
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Monitoring CoT to detect misaligned intentions; paper tests its reliability.
- Lower faithfulness on GPQA.
- CoT monitoring is promising for noticing unintended behaviors but not sufficient to rule them outclaim0.754Central conclusion.
- Training-free monitor that token-forces an eval-aware sentence and computes its probability at each CoT sentence boundary.
- Synchronization construct encapsulating shared data and protected access routines.
- A VS variant that adds chain-of-thought reasoning before generating the distribution of responses with probabilities
- Subsets amenable to monitoring.