AUSN mean -2.119 ± 0.198

Normalized area under S(ℓ) averaged over seeds.

Source paper

extracted_from

(2025) · Edward Yi Chang · Kaya, Zeyneb N. · Ethan Chang

claim

community

Few-shot anchoring & latent structure
members_of
How minimal examples disambiguate and recruit latent arithmetic/reasoning interpretations in LLMs
Neural activation geometry and behavioral prediction
members_of
Quantifies relationships between layer-wise activation statistics (Sbmax, AUSN) and task performance metrics in LLMs, bridging internal representation geometry to behavioral outcomes.

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.

normalized area AUSNconcept0.785
Average of S(ℓ) across layers.
Normalized area under S (AUSN)concept0.784
Average of per-layer S(ℓ) scores, summarizing the breadth of anchoring trajectory.
AUS_N is a weaker correlate of θ50 than S_max across E3 backbonesfinding0.764
E3 finding distinguishing the two geometry summaries; breadth less predictive than peak height
normalized area under S-curve AUSNconcept0.742
Average S(ℓ) across layers; another geometry summary.
LLaMA-3.1-8B: Sbmax = -1.896 ± 0.211, AUSN = -2.119 ± 0.198, peak layer ℓ* = 10 (median)finding0.711
Seed-pooled geometry-only statistics (per-dev z units).
Sbmax mean -1.896 ± 0.211finding0.693
Geometry summary peak anchoring score averaged over seeds.
Geometry summaries (Sbmax, AUSN)method0.687
Peak anchoring (Sbmax) and normalized area under the S(ℓ) curve (AUSN) used to summarize trajectory.
Layer-wise geometry summaries (Sbmax, AUSN) predict internal few-shot thresholds θ50claim0.687
Claim that geometry-to-behavior correlates exist