AI Paper Atlas

Research map / Large language models / Agents, code and math

Mitigating Content Effects on Reasoning in Language Models Through Fine-Grained Activation Steering

Marco Valentino, Geonhee Kim, Dhairya Dalal, Zhixue Zhao, André Freitas

AAAI 2026 · Large language models › Agents, code and math · approach: Mathematical and structural reasoning benchmarks · 36 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 static contrastive activation steering (CAA)Llama 3.2 1b · Syllogistic validity (zero-shot)accuracy (%): 58.17 -> 73.56Improved
add static contrastive activation steering (CAA)Qwen 2.5 7b · Syllogistic validity (zero-shot)Acc/CE: 16.48 -> 93.65Improved
add static CAA steeringLlama 3.2 3b and Qwen 2.5 3b · Syllogistic validity (zero-shot)accuracy (%): 77.79 -> 77.79 (Llama 3b)No clear effect
add static CAA steeringGemma 2 9b · Syllogistic validity (zero-shot)accuracy vs content effect: accuracy 85.00 -> 83.27; CE 8.46 -> 1.92Mixed
add 4-shot in-context examplesLlama 3.1 8b · Syllogistic validityaccuracy (%): 78.27 -> 40.58Hurt
use K-CAST conditional steering instead of unsteered modelLlama 3.2 3b · Syllogistic validity (zero-shot)accuracy (%): 77.79 -> 92.60Improved
use kNN-based condition (K-CAST) instead of aggregated condition vectors (CAST) (ablation)Llama 3.2 3b · Syllogistic validity (zero-shot)accuracy (%): 81.04 -> 92.60Improved
use K-CAST instead of CAST (ablation)Qwen 2.5 3b · Syllogistic validity (zero-shot)accuracy (%): 85.86 -> 85.28No clear effect
apply content-effect steeringLlama 3.2 1b · C4 English perplexityperplexity: 24.29 -> 24.77No clear effect
apply content-effect steeringGemma 2 9b · C4 English perplexityperplexity: 18.95 -> 20.58Hurt
apply steering vectors from syllogismsLlama 3.2 1b · ProntoQA (OOD)accuracy: 49.6 -> 53.6Improved
apply steering vectors from syllogismsGemma 2 9b · ProntoQA (OOD)accuracy: 62.2 -> 52.2Hurt

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