method
active
method:log-odds-ratioLog odds-ratio
Primary evaluation metric measuring causal effect of interventions; greater value indicates larger causal effect
Neighborhood — ranked by edge-count
Frameworks (1)
framework
- CausalGymusesMulti-task benchmark of linguistic behaviours for measuring causal efficacy of interpretability methods, adapted from SyntaxGym
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.
- Evidence term scaling as γN^{1-α}.
- Central quantity; sum of log prior odds and log Bayes factor.
- Model evidence approximated by free energy bound, used for model selection.
- Parameter-free loss transformation applied to each task loss to equalize scales
- Prediction 3: log posterior odds additive in context and steering, with crossover N* predictable from prior odds and magnitudehypothesis0.725Eq. 10.
- Dot product of summed decoder vector with unembedding vector, following Hanna et al.
- Mathematical equivalence showing logarithm transformation recovers IMTL-L in the limit
- Fits log compliance vs α on α with compliance <50% and extrapolates to α=1.