hypothesis
active
hypothesis:aos-can-serve-as-probes-on-demand-using-only-a-natural-language-property-specificationAOs can serve as probes on demand using only a natural-language property specification
Hope for generalization beyond trainable supervised probes.
Source paper
extracted_from(2025) · Adam Karvonen
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (1)
method
- Linear ProbecitesSimple linear classifiers trained on model activations used as the probing technique within the introduced method.
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.
- AOs articulate responses with natural-language flexibility rather than a fixed concept setclaim0.820Contrast with SAE features.
- Overall positioning.
- Argument for scalability.
- Limitation relative to SAEs.
- Multiple forward passes needed.
- Limitation; confabulation risk.
- AOs are not best on Secret Side Constraint, where black-box prompting reveals the secretfinding0.758The one task where AOs lose.
- Key interpretive claim from Case Study II distinguishing probe accuracy from causal relevance