AI Paper Atlas

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

Knowing Bias, Doing Better: Mitigating Social Bias in LLMs via Know-Bias Neuron Enhancement

Jinhao Pan, Chahat Raj, Anjishnu Mukherjee, Sina Mansouri, Bowen Wei, Shloka Yada, Ziwei Zhu

ICML 2026 · Large language models › Language, safety and interpretability · approach: Bias measurement and mitigation · 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
enhance know-bias neurons at inference time instead of suppressing bias neurons or fine-tuningLlama-3.2-3B-Instruct, Qwen-3-4B, Llama-3.1-8B · BBQ, CrowS-Pairs, StereoSettotal average rank: 1.8 / 2.1 / 2.3 (best among 8 methods)Improved
use KnowBias instead of ReGiFT, PEFT, BiasEdit or SD for general capabilityLlama-3.2-3B-Instruct · OBQA, COPA, ARCaccuracy drop: no drop of at least 5% for KnowBias vs many starred baselinesImproved
apply KnowBias to Llama-3.1-8BLlama-3.1-8B-Instruct · COPAaccuracy: .7132 -> .6780Hurt
apply KnowBias to Llama-3.2-3BLlama-3.2-3B-Instruct · ARC-Eaccuracy: .6814 -> .7191Improved
use neurons identified from a single demographic dimension on other dimensions (ablation)KnowBias on Llama-3.2-3B · SS-inter gender ICATICAT: .7000 -> .8550 (race neurons)Improved
use gender-derived neurons on religion bias (ablation)KnowBias on Llama-3.2-3B · SS-inter religion ICATICAT: .6196 -> .6883Improved
increase number of bias-knowledge questions beyond 45 (ablation)KnowBias · StereoSet-InterICATNo clear effect
mix three question types instead of a single type (ablation)KnowBias · BBQ, CS, SSbias scoresImproved
use union of dimension-specific neuron sets instead of intersection or composite question (ablation)KnowBias · BBQ, CS, SSbias scoresImproved
use unified neuron set instead of dimension-specific neurons (ablation)KnowBias · BBQ, CS, SSbias scoresImproved
select know-bias neurons by attribution instead of random neurons of same count (ablation)KnowBias · bias benchmarksbias scoresImproved
use 5 bias concepts instead of 25 (ablation)KnowBias · bias benchmarksdebiasing performanceNo clear effect

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