k50 for base-10 two-digit addition: 0.28 ± 0.05 shots

Shot midpoint from logistic fit over 10 runs.

Source paper

extracted_from

The Guanyin Protocol: A Framework for Immediately Establishing an Understanding of Both Causality and Compassion in LLM Systems Using Semantic Anchoring

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

Neighborhood — ranked by edge-count

Claims (1)

claim

S = ρd - dr - log k predicts shot midpoints across different bases, tasks, and models
supports
Predictive practical utility claim.

Communities (3)

community

Few-shot anchoring & latent structure
members_of
How minimal examples disambiguate and recruit latent arithmetic/reasoning interpretations in LLMs
Few-shot learning phase transitions in neural networks
members_of
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 thresholds
members_of
k50 and phase width metrics for multi-base addition across bases 8, 9, and 10

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.

B10 shot midpoint k50 = 0.28 ± 0.05 shots with accuracy 94.8 ± 1.2%finding0.791
Lowest threshold condition in E2; near-zero/one-shot threshold consistent with high pretraining density
few-shot threshold (k50/θ50)concept0.759
Shot count needed to reach 50% accuracy; reflects when anchoring strength crosses critical value.
Shot midpoint ordering k50(B10) < k50(B8) ≈ k50(B9) tracks pretraining exposure densityclaim0.753
Interpretation that pattern density from pretraining determines few-shot requirements
Base-10 additionconcept0.737
The generic addition mechanism that Llama-3.1-8B actually uses to compute sums before mapping back to cyclic concept space
Hypothesis: Shot midpoint ordering k50(B10) < k50(B8) ≈ k50(B9) follows pretraining exposure densityhypothesis0.734
E2 prediction that bases with higher pretraining exposure require fewer shots to cross threshold
B9 phase width (k90 − k10) = 3.74 ± 0.31 shotsfinding0.732
Widest transition in E2; consistent with lower prior density requiring more shots for reliable threshold crossing
Top-5 instructions by µ(1→2) at ℓ=12 achieve average cosine similarity .9893 and average accuracy .5645 on gsm8k_adv for Gemma3-4B-ITfinding0.729
High cosine similarity for Gemma3 steering vectors suggests strong linear reflection structure.
Hardest koans across 28 models: BD-003 (mean 2.45), MC-003 (mean 2.55), CA-003 (mean 2.58) — all require genuine self-confrontationfinding0.728
Hardest koans demand honest self-observation under uncertainty, not philosophical fluency