- Rating
- 4(+1)
- Confidence
- 3(+1)
- Informativeness
- 3(+1)
- Presentation
- 2
- Soundness
- 3(+1)
Read review
Paper summary
This paper investigates whether concept-conditioned scores genuinely depend on concept identity when ranking sparse dictionaries. The authors propose TWIN, combining evaluation-time label randomization, training-time target permutation, occupancy control, and probing anchors. Across 65 dictionaries and three label systems, many scores preserve nearly identical rankings under randomized labels, sometimes leading to poor model selection. A theoretical bound explains this failure for mass coverage.
Critical strengths
-
Important and actionable methodological contribution.
TWIN addresses an overlooked gap in existing SAE sanity checks by testing whether dictionary rankings, rather than merely score values, depend on concept identity. Its occupancy control further addresses a confound that survives label randomization.
-
Strong empirical validation and theoretical insight.
The study combines 65 dictionaries, three label systems, paired training experiments, and controlled synthetic tests. The label-free bound explains the compactness failure, while pre-registration and transparent reporting strengthen the evidence.
Critical weaknesses
-
Limited evidence of practical model-selection harm.
The strongest selection failures involve collapsed dictionaries, while selection regret is measured by probing recoverability rather than concept compactness itself. The practical consequences among healthy dictionaries remain unclear.
-
Limited generalization to other evaluation regimes.
Training-time and model-selection conclusions rely on one medical dictionary family, while public SAEs use document-level pooling with saturated probing anchors. Directly concept-bound supervised dictionaries are also not tested.
Questions for authors
-
Can the authors report held-out selection regret after excluding collapsed dictionaries (e.g., ), ideally comparing full-probe recoverability with sparse-probe or concept-selectivity criteria?
-
Can the authors evaluate TWIN on directly concept-bound supervised dictionaries, or report the profile discrepancy and null-test results for the existing per-feature-aligned configurations?