AI Paper Atlas

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

Efficient LLM Moderation with Multi-Layer Latent Prototypes

Maciej Chrabaszcz, Filip Szatkowski, Bartosz Wójcik, Jan Dubiński, Tomasz Trzcinski, Sebastian Cygert

ICML 2026 · Large language models › Language, safety and interpretability · approach: Mechanistic analysis of attention and representations · 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
use Mahalanobis distance instead of Euclidean distance (ablation)MLPM (last layer) · 8 harmful-prompt datasetsaverage F1: 77.17 -> 86.25 (Llama-8B)Improved
aggregate multiple layers instead of the last layer only (ablation)MLPM (Mahalanobis) · 8 harmful-prompt datasetsaverage F1: 86.25 -> 88.30 (Llama-8B)Improved
use last-token representation instead of mean over tokens (ablation)MLPM · 8 harmful-prompt datasetsaverage F1: 84.22 -> 87.55 (Mistral-7B)Improved
use GDA instead of logistic regression or MLP as per-layer classifier (ablation)MLPM · 8 harmful-prompt datasetsaverage F1: 86.25 vs 84.51 (Llama-8B); 84.18 vs 84.32 (Mistral-7B)Mixed
use instruction-tuned instead of pretrained base models (ablation)MLPM · out-of-distribution harmful datasetsF1Improved
use end-of-thinking token instead of last prompt token on reasoning models (ablation)MLPM on Qwen3 thinking models · WildGuardMixF1: up to 0.5% for models larger than 1.7BNo clear effect
use MLPM as input filter with simple refusalLlama3-8B · WildGuardMix testASR: 36.21 -> 15.52Improved
condition activation steering (Turner et al. 2023) on MLPM flags instead of steering all promptsMistral-7B · WildGuardMix testFRR / ASR: FRR 18.73 -> 3.60; ASR 40.45 -> 45.89Mixed
condition Zheng et al. safety-prompt steering on MLPM flagsMistral-7B · WildGuardMix testF1 of (1-ASR),(1-FRR): 59.30 -> 59.16No clear effect
increase training samples for MLPM (ablation)MLPM · WildGuardMix F1F1Improved
use sparse subset of layers via l1 regularization instead of all layers (ablation)MLPM · harmful datasetsF1Improved
scale MLPM backbone model size (ablation)MLPM · harmful datasetsF1Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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