finding

active

finding:as-number-of-nearest-neighbors-k-decreases-in-cknna-metric-cross-modal-alignment-trend-becomes-more-pronounced-across-both-models-and-tasks

As number of nearest neighbors k decreases in CKNNA metric, cross-modal alignment trend becomes more pronounced across both models and tasks

Shows cross-modal alignment is primarily local rather than global

Source paper

extracted_from

The Platonic Representation Hypothesis

(2024) · Minyoung Huh · Brian Cheung · Tongzhou Wang · Phillip Isola

Neighborhood — ranked by edge-count

Methods (1)

method

Mutual k-Nearest Neighbor Alignment Metric
supports
Primary alignment metric used in experiments; measures mean intersection of k-nearest neighbor sets between two kernels

Findings (1)

finding

Cross-modal language-vision alignment reaches a maximum of approximately 0.16 on mutual nearest-neighbor metric in Figure 3, well below the theoretical maximum of 1
supports
Quantitative bound on observed alignment; raises the open question of whether this gap reflects noise or real misalignment

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.

CKA shows a very weak trend of alignment between models even within modality, compared to mutual k-NN which shows stronger trendsfinding0.865
Explains why mutual k-NN was chosen over CKA as primary metric
Spearman's rank correlation among different alignment metrics (CKA, SVCCA, Mutual k-NN, CKNNA) over 78 vision models is high across variants, with all p-values below 2.24×10^-105finding0.778
Validates robustness of alignment metric choice
Is a mutual nearest-neighbor alignment score of 0.16 indicative of strong alignment with remaining gap being noise, or does it signify poor alignment with major differences left to explain?question0.768
Open question the authors leave unresolved about interpreting the magnitude of their alignment measurements
Roughness in responses decreases with parameter count within same-alignment model families, operationalizing the cost of polishing.claim0.756
The mathematical argument for cross-modal convergence strictly holds only for bijective projections of the underlying worldclaim0.750
Key limitation of the formal PRH derivation: lossy or stochastic observation functions weaken the convergence guarantee
Does alignment type predict meta-cognitive style when models review consciousness research, as well as koan responses?question0.744
Four frontier models reviewing the paper each responded in the mode their alignment type predicts; N=1, awaiting systematic study
The ordering of few-shot thresholds k50 and transition widths aligns with k50 ∝ dr/ρd.claim0.743
Interpretation of E2 results.
Alignment type is the only significant predictor of koan scores (p=0.006); architecture, parameter count, open/closed weights, MoE/dense are all non-significantfinding0.743
Main statistical finding: what predicts scores is training approach, not size or architecture