finding
active
finding:fixing-pelt-to-l2-cost-raises-pooled-tvd-by-0-0010-indistinguishable-from-full-model-retaining-97-100-of-improvementFixing PELT to L2 cost raises pooled TVD by ≤0.0010, indistinguishable from Full Model, retaining 97–100% of improvement.
Ablation.
Source paper
extracted_from(2026) · Eric Bigelow · Amir Zur · Satchel Grant · Tal Haklay +6
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Ablation.
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.
- Replacing kernel pooling with flat per-segment pooling raises pooled TVD by 0.0021 for Llama at S=30.finding0.790Ablation.
- Ablation.
- Log-log slope of −0.4903 of pooled replicate TVD vs S through S=200 on two development questions, near −1/2.finding0.740Noise scaling.
- On recorded DeepSeek example, Kernel Pooling alone has TVD 0.2720 versus 0.0088 for Full Model.finding0.740Blurred forks.
- Ablation.
- ruptures silently replaces custom cost not inheriting BaseCost with CostL2, significantly affecting results.finding0.728Bug discovered.
- Forks missed with coarse spacing.
- Table 2.