Verbalising Model's Computation for Chain-of-Thought Faithfulness Detection
Projects hidden states into vocabulary directions to track how internal information changes across layers and written reasoning steps. Uses these trajectories to detect unfaithful chains of thought and test transfer across tasks and models.
First public: · Added to this list: October 22, 2026
Explore this approach
Analysis and evaluation — Measure how well these methods work, investigate what their internal computation does, and identify where they fail. This includes experiments, studies of model internals, and mathematical results.