AI Paper Atlas

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

Discovering Decoupled Functional Modules in Large Language Models

Yanke Yu, Jin Li, Ying Sun, Ping Li, Zhefeng Wang, Yi Zheng

AAAI 2026 · Large language models › Language, safety and interpretability · approach: Pruning and circuits in LLMs · 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
use IterD instead of K-Means for module discoveryQwen2.5-7B-Instruct · Infinity-Instruct activations, K=5L(F) x1e6: 29.6 -> 64.6Improved
use IterD instead of Spectral Co-clusteringQwen2.5-7B-Instruct · K=10L(F) x1e6: 16.9 -> 25.2Improved
use IterD discovered modules to build features for function-label classificationLogistic Regression on Qwen2.5-7B-Instruct · K=5accuracy: 0.5121 -> 0.6000Improved
use IterD instead of baselines for classification features at K=20Qwen2.5-1.5B-Instruct, SVC · K=20accuracy: 0.7986 -> 0.8236 (K-Means to IterD)Improved
use IterD instead of Mini-Batch K-Means for classification featuresQwen2.5-7B-Instruct · K=20, Mini-Batch K-Means vs IterDaccuracy: 0.8036 -> 0.8214 (SVC)Improved
increase number of modules KIterD · Qwen2.5-7B-InstructL(F): 64.6 (K=5) -> 25.2 (K=10)Hurt
increase number of modules KIterD · Qwen2.5-7B-Instructclassification accuracy: 0.6000 (K=5) -> 0.8179 (K=20), LRImproved

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