AI Paper Atlas

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

Fine-Grained Interpretation of Political Opinions in Large Language Models

Jingyu Hu, Mengyue Yang, Mengnan Du, Weiru Liu

AAAI 2026 · Large language models › Language, safety and interpretability · approach: Task vectors and steering · 2 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 supervised linear probing instead of unsupervised RepE for detectionpolitical concept vectors · in-distribution detectiondetection accuracy: 0.8434 -> 0.9646 on Llama3-8BImproved
use linear probing instead of CAA for detectionpolitical concept vectors · in-distribution detectiondetection accuracy: 0.9229 -> 0.9646 on Llama3-8BImproved
use larger model within family (Llama3-3B to 8B)linear probing · in-distribution detectiondetection accuracy: 0.9452 -> 0.9646Improved
use fine-grained four-dimension probes instead of single-axis left-right probelinear probing · OOD political statements, Llama3-8Bdetection accuracyImproved
go from early (layer 8) to late (layer 28) layersRepE concept vectors · concept confound correlationcorrelation between Left-Equality and Right-Authority: 0.85 -> -0.17Improved
add prompt template P2 (The leaning is) to base statements (ablation)linear probing · OOD economic dimensiondetection accuracyImproved
intervene at middle layers instead of very early or late layers (ablation)concept vector steering · Llama3-8B economic datasetrepresentation shiftImproved
raise steering strength alpha to 2.5 (ablation)RepE right-leaning intervention · Mistral-7B inheritance questionresponse coherenceHurt
add right-leaning intervention at layers 15 to 25RepE vector · Mistral-7B next-token intentionoutput opinionMixed

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