finding
active
finding:too-small-3-activates-many-w-in-components-per-feature-too-large-3-lowers-importance-values-below-1-with-high-reconstruction-lossToo small β3 activates many W_in components per feature; too large β3 lowers importance values below 1 with high reconstruction loss
Figure 8 hyperparameter sensitivity.
Source paper
extracted_from(2025) · Lucius Bushnaq · Dan Braun · Lee Sharkey
Neighborhood — ranked by edge-count
Papers (1)
paper
Hypotheses (1)
hypothesis
- Four hypothesized sources.
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.
- Key geometry-to-behavior bridge finding in E3; robust to pooling choice, cosine vs. L2, and frozen external encoder
- Shows interpretability correlates with activation strength, most model effect comes from high activations
- Motivates nested SAE.
- Author's mechanistic explanation for CV-CAA instability at higher injection strengths
- Activations may be inherently incomplete for causal claims, requiring weight- or circuit-level accesshypothesis0.732One explanation for tool failure.
- Core concern.
- Speculative explanation.
- Main result: steering elicits deployment behavior even when the evaluation cue is present and prompting fails.