finding
active
finding:b10-shot-midpoint-k50-0-28-0-05-shots-with-accuracy-94-8-1-2B10 shot midpoint k50 = 0.28 ± 0.05 shots with accuracy 94.8 ± 1.2%
Lowest threshold condition in E2; near-zero/one-shot threshold consistent with high pretraining density
Source paper
extracted_from(2025) · Edward Yi Chang · Kaya, Zeyneb N. · Ethan Chang
Neighborhood — ranked by edge-count
Claims (4)
claim
- Shot midpoints follow k50 ∝ dr/ρd; higher cohesion and lower mismatch yield fewer required examplesassociated_withsupportsCore quantitative prediction of UCCT validated by E2 threshold ordering
- E2 main interpretive claim.
- Interpretation of E2 results.
- Interpretation that pattern density from pretraining determines few-shot requirements
Hypotheses (3)
hypothesis
- Shot midpoint ordering k50(B10) < k50(B8) ≈ k50(B9) and transition widths correlate with mismatch D(P0∥PT)answered_bysupportsTestable prediction for Experiment 2
- Hypothesis: Shot midpoint ordering k50(B10) < k50(B8) ≈ k50(B9) follows pretraining exposure densityassociated_withsupportsE2 prediction that bases with higher pretraining exposure require fewer shots to cross threshold
- Core testable hypothesis of UCCT about the nature of performance transitions under anchoring
Communities (3)
community
- Few-shot anchoring & latent structuremembers_ofHow minimal examples disambiguate and recruit latent arithmetic/reasoning interpretations in LLMs
- Empirical characterization of k50 midpoints and transition widths across transformer models, tracking how pretraining density ρd/dr predicts in-context learning thresholds.
- Few-shot arithmetic learning thresholdsmembers_ofk50 and phase width metrics for multi-base addition across bases 8, 9, and 10
Questions (2)
question
- First research question in Experiment 2
- First E2 research question directly testing UCCT's core predictive claims
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.
- Number of in-context exemplars to reach 50% accuracy in E2.
- Shot midpoint from logistic fit over 10 runs.
- Widest transition in E2; consistent with lower prior density requiring more shots for reliable threshold crossing
- Shot count needed to reach 50% accuracy; reflects when anchoring strength crosses critical value.
- Accuracy at k=16 shots for B10.
- Monotone ordering consistent with k50 ∝ dr/ρd.
- Transition width (k90 – k10) for B10.
- Hypothesis: Peak alignment location S_max and normalized trajectory area AUS_N predict shot midpoints θ50hypothesis0.738E3 prediction that internal geometry provides a bridge to behavioral thresholds