finding

active

finding:a-single-hyperparameter-procedure-driven-by-the-intrinsic-dictionary-health-audit-transfers-robustly-across-sleepfm-reve-and-labram

A single hyperparameter procedure driven by the intrinsic dictionary health audit transfers robustly across SleepFM, REVE, and LaBraM.

Demonstrates architecture-agnostic applicability of the SAE tuning method

Source paper

extracted_from

Mechanistic Interpretability of EEG Foundation Models via Sparse Autoencoders

(2026) · William Lehn-Schiøler · Magnus Ruud Kjær · Rahul Thapa · M. Pedersen +9

Neighborhood — ranked by edge-count

Claims (1)

claim

A single SAE hyperparameter procedure driven by an intrinsic dictionary health audit transfers robustly across all three EEG transformer architectures.
supports
Key methodological contribution claim about architecture-agnostic SAE tuning

Communities (3)

community

Manifold-aware concept steering in neural representations
members_of
Explores geometry of activation/behavior manifolds to enable selective, non-destructive concept interventions.
Concept entanglement in biomedical foundation models
members_of
Investigates inseparability of clinical concepts (age, pathology) in EEG transformers using SAE feature analysis and steering metrics across SleepFM, REVE, LaBraM architectures.
Dictionary health audit transfer
members_of
Hyperparameter procedure validated across SleepFM, REVE, and LaBraM EEG transformer architectures.

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

SAEs successfully extract sparse feature dictionaries from embeddings of SleepFM, REVE, and LaBraM EEG transformers.finding0.763
Foundational empirical result enabling all downstream analysis
Monosemanticity and entanglement of SAE features were benchmarked for clinical taxonomy grounding across SleepFM, REVE, LaBraM.finding0.761
Quantitative assessment of feature quality using clinical concepts across models.
Concept steering with target vs off-target probe area metric reveals three operational regimes (selectively steerable, encoded but entangled, non-encoded) across SleepFM, REVE, LaBraM.finding0.755
Result categorizing concept steerability into three distinct regimes.
Intrinsic Dictionary Health Auditmethod0.745
A hyperparameter selection procedure driven by intrinsic measures of SAE dictionary quality that transfers across architectures
An equivalent of Ohm’s law for stress, care, and intelligence would provide a key for understanding the evolution of intelligence across a wide range of spaces.hypothesis0.742
Speculative suggestion for a mathematical formalization.
The likelihood of a dedicated feature for a concept (element, city, animal, food) follows a sigmoid in log-frequency of the concept in training data, with threshold frequency inversely proportional to number of alive features.finding0.740
Quantitative relationship between concept frequency and feature presence.
All cohort benchmarks measure output, not state, and are subject to eval-awareness contamination.claim0.732
For Gaussian data with homoscedastic class-conditional distributions, IID mass-mean probing coincides with logistic regression (Theorem F.1)finding0.732
Formal result establishing the theoretical connection between mass-mean probing and LR