How many entries carry both tags. Hatched cells are combinations nothing in the current selection covers. Select a cell to list its entries.
Definitions of faithfulness, counterfactual and monitorability quoted from the entries that match your filters, grouped by cluster of meaning. The search matches the definitions themselves. Read the synthesis.
Mathematical definitions of faithfulness, monitorability and related notions (deception, steganography, introspection, causal explanation and others), grouped by the notion they define. They come from two sets of papers: the 217 papers on chain-of-thought faithfulness and monitorability (the same papers as the Definitions tab), and 23 papers from causal inference, game theory, information theory and explainable AI that define related notions mathematically. Each definition is marked as defining the notion itself or only a way to measure or test it. Read the synthesis.