AI Paper Atlas

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

How Language Models Process Negation

Zhejian Zhou, Tianyi Zhou, Robin Jia, Jonathan May

ICML 2026 · Large language models › Language, safety and interpretability · approach: Task vectors and steering · 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
Apply Cumulative Attention Sink to late-layer attention modules (ablation)Llama-3.1-8B · Negation prompts, negative accuracynegative accuracy (%): 50.5 -> 67.8Improved
Apply Cumulative Attention Sink to late-layer attention modules (ablation)Mistral-7B-v0.1 · Negation prompts, negative accuracynegative accuracy (%): 45.2 -> 65.9Improved
Apply Cumulative Attention Sink to late-layer attention modules (ablation)OLMo-2-7B · Negation prompts, negative accuracynegative accuracy (%): 54.0 -> 68.7Improved
Use LogitLens to skip later layers instead of Attention Sink (ablation)Llama-3.1-8B · Negation prompts, negative accuracynegative accuracy (%): 50.5 -> 53.6Improved
Use LogitLens to skip later layers (ablation)Mistral-7B-v0.1 · Negation prompts, negative accuracynegative accuracy (%): 45.2 -> 61.6Improved
Sink attention at layers after the best negative-accuracy layer (ablation)six open LLMs · Negation prompts, positive accuracypositive accuracyHurt
Sink a window of attention layers around layer 14 (ablation)Llama-3.1-8B · Negation prompts, negative accuracynegative accuracyHurt
Sink attention around layer 17 (middle-late layers) (ablation)Llama-3.1-8B · Negation prompts, negative accuracynegative accuracyImproved
Sink very late layers (ablation)Llama-3.1-8B · Negation prompts, negative accuracynegative accuracyNo clear effect
Use multiple candidate answers per prompt (averaged logits) instead of a single pair (ablation)six open LLMs · Negation prompts, negative accuracynegative accuracy: Llama-3.1 50.5 -> 54.6No clear effect
Pre-training progress (OLMo-2 checkpoints)OLMo-2 · Negation prompts, negative accuracynegative accuracyMixed
Project mid-layer attention outputs with LogitLens to find promoted tokensLlama-3.1-8B · Negation prompts, LLM-annotated evidenceshare of examples with 'not Y' tokens: > 80%Improved

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