finding
active
finding:on-recorded-deepseek-example-kernel-pooling-alone-has-tvd-0-2720-versus-0-0088-for-full-modelOn recorded DeepSeek example, Kernel Pooling alone has TVD 0.2720 versus 0.0088 for Full Model.
Blurred forks.
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 interpretation.
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.finding0.852Ablation.
- Ablation.
- Forks missed with coarse spacing.
- DeepSeek-V3.1 shows broad fine-tuning sensitivity; outputs code on nearly all open-ended prompts under insecure fine-tuning
- Log-log slope of −0.4903 of pooled replicate TVD vs S through S=200 on two development questions, near −1/2.finding0.767Noise scaling.
- Core empirical result demonstrating early belief formation in easy tasks
- C.7.1.
- DeepSeek-V3.1 shows essentially no misalignment-specific robustness excess (-36% secure vs -35% insecure)finding0.752DeepSeek is an outlier showing broad fine-tuning sensitivity rather than clean misalignment-specific collapse