AI Paper Atlas

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

Making Sense of LLM Decisions: A Prototype-based Framework for Explainable Classification

Bowen Wei, Mehrdad Fazli, Ziwei Zhu

AAAI 2026 · Large language models › Language, safety and interpretability · approach: Faithfulness of LLM explanations · 1 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 token attribution scores from target LLM (ablation)ProtoSurE · Hotel/DBPedia/Consumer (Llama-3.1-8B)Accuracy: 91.1 -> 91.9Improved
add token attribution scores from target LLM (ablation)ProtoSurE · Qwen2.5-7B averageAccuracy: 92.7 -> 93.4Improved
make prototypes trainable after K-means init (ablation)ProtoSurE · Llama-3.1-8B avgAccuracy: 89.7 -> 90.9Improved
make prototypes trainable after K-means init (ablation)ProtoSurE · Qwen2.5-7B avgAccuracy: 92.6 -> 93.6Improved
swap sentence encoder (GTE vs SBERT/T5) (ablation)ProtoSurE · Overall avg accuracyAccuracy: 0.8952 vs 0.8881No clear effect
increase number of prototypes K (ablation)ProtoSurE · DBPedia, Consumer (Llama-3.1-8B)faithfulness metricsImproved
increase training examples beyond 128 (ablation)ProtoSurE · Hotel (Llama-3.1-8B)Accuracy: 96.5 -> 98.4No clear effect
use ProtoSurE instead of SHAP/IG/Occlusion/DeepLIFTProtoSurE vs post-hoc baselines · 3 LLMs x 4 datasetsComprehensiveness: .138 (IG) -> .192Improved
use ProtoSurE instead of baselinesProtoSurE vs post-hoc baselines · 3 LLMs x 4 datasetsDeletion rank correlation: .092 (Occlusion) -> .100Mixed
use ProtoSurE instead of SELF-EXP/SHAP/OcclusionProtoSurE vs baselines · Hotel (GPT-4o-mini)Sufficiency: 0.026 (SHAP) -> 0.001Improved

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