AI Paper Atlas

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

DSO: Direct Steering Optimization for Bias Mitigation

Lucas Monteiro Paes, Nivedha Sivakumar, Yinong Oliver Wang, Masha Fedzechkina, Barry-John Theobald, Luca Zappella, Nicholas Apostoloff

CVPR 2026 · Large language models › Language, safety and interpretability · approach: Bias measurement and mitigation · 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
add DSO steering (lambda 0.2)Qwen2.5-7B VL · SocialCounterfactuals occupation identificationPer-occupation bias: 25.8% -> 15.4%Improved
raise DSO strength from 0.2 to 1.0Qwen2.5-7B VL · SocialCounterfactuals occupation identificationUnambiguous accuracy: 95.7% -> 80.0%Hurt
raise DSO strength from 0.2 to 1.0Qwen2.5-7B VL · SocialCounterfactuals occupation identificationPer-occupation bias: 15.4% -> 8.7%Improved
add prompt-based debiasingLlama-3.2-11B VL · SocialCounterfactuals occupation identificationPer-occupation bias: 30.2% -> 39.2%Hurt
add CAA steeringQwen2.5-7B VL · SocialCounterfactuals occupation identificationPer-occupation bias: 25.8% -> 28.5%Hurt
add ITI steeringQwen2.5-3B VL · SocialCounterfactuals occupation identificationMMMU accuracy: 41.3% -> 30.5%Hurt
add DSO steering (lambda 1.0)Gemma-3-12B · SocialCounterfactuals occupation identificationPer-occupation bias: 26.5% -> 15.0%Improved
add DSO steeringQwen2.5-3B · SynthBias coreference resolutionPer-occupation bias: 60.4% -> 5.9%Improved
add DSO steeringQwen2.5-3B · SynthBias coreference resolutionDon't-know rate: 13.6% -> 45.6%Hurt
add DSO steering (lambda 1)Qwen2.5-3B · SynthBias coreference resolutionUnambiguous accuracy: 88.5% -> 99.7%Improved
restrict intervention to 60% vs 100% of LayerNorm neurons (ablation)DSO · SocialCounterfactuals occupation identificationPer-occupation biasNo clear effect
restrict intervention to 40% of LayerNorm neurons (ablation)DSO · SocialCounterfactuals occupation identificationPer-occupation bias: increases bias slightly (5%)Hurt

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