AI Paper Atlas

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

An Information-Theoretic Parameter-Free Bayesian Framework for Probing Labeled Dependency Trees from Attention Score

Hongxu Liu, Jing Ma, Xiaojie Wang, Caixia Yuan, Fangxiang Feng

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Linguistic encoding in LLM representations

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
use posterior-based tree reconstruction instead of raw scoreIPBP · UD 2.9 EnglishUAS: 32.3 -> 49.1Improved
use posterior-based tree reconstruction instead of raw scoreIPBP · UD 2.9 EnglishLAS: 16.6 -> 30.6Improved
add positive MI (MIpos) mixed with binary MI (ablation)IPBP · UD 2.9 EnglishLAS: 30.6 -> 34.8Improved
add positive MI (MIpos) mixed with binary MI (ablation)IPBP · UD 2.9 EnglishUAS: 49.1 -> 49.9Improved
probe transposed attention instead of original orientation (ablation)IPBP · UD 2.9 EnglishUAS: 49.1 -> 42.6Hurt
probe undirected trees instead of directed (ablation)IPBP · UD 2.9 EnglishUAS: 49.1 -> 45.3Hurt
probe labeled trees directly instead of arc-first only (arc only baseline) (ablation)IPBP · UD 2.9 EnglishUAS: 36.5 -> 49.1Improved
use IPBP instead of V-Information supervised probeIPBP · UD 2.9 EnglishLAS: 20.9 -> 30.6Improved
KDE bandwidth 0.5x instead of rule-of-thumb (ablation)IPBP · UD 2.9 EnglishUAS: 49.1 -> 49.2No clear effect
use IPBP head selection instead of ElasticNet/IoU/ProbelessIPBP · head selection consistencySpearman-R: 0.309 -> 0.398Improved

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