finding
active
finding:prevailing-one-hot-positional-view-achieves-jss-0-44-below-uniform-baseline-0-5Prevailing one-hot positional view achieves JSS 0.44, below uniform baseline 0.5
Baseline comparison.
Source paper
extracted_from(2025) · Yoav Gur-Arieh · Mor Geva · Atticus Geiger
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Central thesis.
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.
- Gaussian matters.
- Ablation.
- J-lens eval-awareness score separates NLA-flagged from unflagged positions with AUC 0.853.finding0.739Eval awareness probe.
- Strong empirical evidence that VS recovers pretraining distribution while direct prompting collapses
- Table 2; e.g., Identity +1.8 for turn-averaged.
- Indicator: multiple sources of perceptual activity including top-down and noise required for reality monitoring.
- Shows VS enables LLMs to better approximate random behavior compared to direct prompting
- Shows VS substantially better approximates the pretraining distribution than baseline methods