method
active
method:importance-samplingImportance Sampling
Prior state-of-the-art baseline using the abliterated model as proposal distribution.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (1)
method
- Logit Path ExtrapolationcontradictsThe paper's method: interpolate between two models in logit space, measure compliance where common, and extrapolate to the original.
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.
- Weighted Spearman correlation that corrects for sampling bias in automated interpretability evaluation
- A measure of whether a subcomponent is necessary to reproduce model behavior on a specific prompt, predicted by the causal importance network.
- Variant tried that performed worse than importance sampling.
- Scalar weight on loss for each feature.
- Small per-subcomponent MLP with hard sigmoid predicting causal importance.
- Prior result for comparison.
- Self-Normalizing Importance Sampling performed worse than Importance Sampling in this settingfinding0.735Appendix baseline comparison.
- Parameters of the approximate posterior, such as Dirichlet counts for model parameters.