hypothesis
active
hypothesis:activations-may-be-inherently-incomplete-for-causal-claims-requiring-weight-or-circuit-level-accessActivations may be inherently incomplete for causal claims, requiring weight- or circuit-level access
One explanation for tool failure.
Source paper
extracted_from(2026) · Adam Karvonen · Euan Ong · Subhash Kantamneni · Samuel Marks
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- No activation tool (AO, NLA, SAE) improves on transcript-only baseline for Gemma-3-27B-IT targetassociated_withMain negative result; baseline AUROC ~81 with tools within ~1 point.
Claims (1)
claim
- Unlike fine-tuned quirks, natural behaviors do not produce anomalous activation tool outputsassociated_withExplains null result versus auditing games.
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.
- Argument that circuits methodology meets natural-science standards of falsifiability
- Hypothesis supported by ablation of massive activations in Retrofitted Llama that eliminates stage structure
- Motivates shift from studying model activations ('thoughts') to understanding parameters ('the computations themselves').
- Central thesis of the paper
- Normative vision for how the circuits agenda could resolve the pre-paradigmatic state of interpretability
- Motivates nested SAE.
- If the relationship is present in activations, better tools (e.g. trained on these datasets) could surface ithypothesis0.749Alternative explanation.
- Central thesis of the paper