AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Rethinking Label Consistency of In-Context Learning: An Implicit Transductive Label Propagation Perspective

Haoyang Chen, Richong Zhang, Junfan Chen

AAAI 2026 · Large language models › Language, safety and interpretability · approach: In-context learning analysis · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
select demonstrations with TopK on label-interpolated synthetic embeddings (TopK-SD) instead of original embeddingsTopK demonstration selection · 9 datasets avg, LLaMA3-8Baverage accuracy: 79.8 -> 81.1Improved
use TopK-SD with fixed lambda 0.7 instead of TopKTopK demonstration selection · 9 datasets avg, DeepSeek-7Baverage accuracy: 76.6 -> 76.9No clear effect
use TopK-SD with fixed lambda 0.7 instead of TopKTopK demonstration selection · GPT-J-6B and LLaMA2-7B, 9 datasets avgaverage accuracy: 72.5 -> 74.0; 75.9 -> 77.1Improved
use TopK-SD on MNLI instead of TopKTopK demonstration selection · MNLI, DeepSeek-7B and LLaMA2-7Baccuracy: 50.6 -> 48.4 (DeepSeek); 52.2 -> 51.3 (LLaMA2)Hurt
replace stage-1 TopK with TopK-SD in two-stage selection methods (MDL, ConE, DPP) (ablation)MDL, ConE, DPP · 9 datasets avg, LLaMA3-8Baverage accuracy: +0.9, +0.4, +0.6Improved
increase label consistency of demonstrations (share of demos with the query label) (ablation)ICL demonstration set · SST-2 and AGNewsICL accuracyImproved
let LLM infer on selected demonstrations instead of simple label voting (ablation)TopK-SD demonstrations · SST-2, SST-5, AGNews, TREC, LLaMA3-8Baccuracy: TREC 84.6 -> 92.6 at lambda 0.6Improved
set lambda to 0.0 (embedding fully replaced by label center) (ablation)TopK-SD · SST-5, LLaMA3-8Baccuracy: 18.4 vs 52.7 at lambda 1.0Hurt
tune lambda per dataset instead of fixed 0.7 (ablation)TopK-SD · 9 datasets, LLaMA3-8Baverage accuracyImproved
use TopK-SD instead of TopK with few demonstrations (3 or fewer) (ablation)TopK demonstration selection · TREC, varying number of demonstrationsaccuracyImproved
random demonstration selection with few shots (ablation)ICL · TRECaccuracy: 47.0 -> 23.2 at 1 demonstrationHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy