AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Bridging the Knowledge-Prediction Gap in LLMs on Multiple-Choice Questions

Yoonah Park, Haesung Pyun, Yohan Jo

ICML 2026 · Large language models › Retrieval and knowledge · approach: Explanatory and multiple-choice QA benchmarks · 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 KAPPA alignment of prediction and knowledge subspaces at 6 layersLlama 3.1 8B Instruct · TruthfulQA (MCQ)accuracy: 56.7 -> 73.5Improved
apply KAPPA at 6 layersMistral v0.3 7B · TruthfulQA (MCQ)accuracy: 40.7 -> 58.3Improved
apply KAPPA at 6 layersQwen2.5 7B Instruct · GSM8k (MCQ)accuracy: 47.8 -> 49.2Improved
use CAA activation steering instead of KAPPALlama 3.1 8B Instruct · TruthfulQA (MCQ)accuracy: 56.7 -> 62.3 (CAA) vs 73.5 (KAPPA)Improved
use DoLA decoding as baseline interventionLlama 3.1 8B Instruct · TruthfulQA (MCQ)accuracy: 56.7 -> 55.6No clear effect
increase number of intervened layers from 1 to 6 (ablation)KAPPA · TruthfulQA (MCQ)accuracy: 67.8 -> 73.5Improved
apply KAPPA on knowledge-intensive benchmarks with negligible gapKAPPA · knowledge MCQ benchmarks (e.g., MMLU, ARC)accuracyNo clear effect
increase alpha or beta in extended alignment (ablation)KAPPA · TruthfulQAagreement rateImproved
reduce probe training data to 10% (ablation)KAPPA · probe data size sweepaccuracy vs baseNo clear effect
transfer probes across datasets (ablation)KAPPA · cross-dataset agreement (Llama 3.1 8B)agreement improvement: +4.81 points BBH-NLP to GSM8k; +5.72 points TruthfulQA to BBQ-AgeMixed
apply KAPPA to free-form generation using MCQ-derived subspacesQwen2.5 7B Instruct · TruthfulQA free-formaccuracy: 41.7 -> 44.2Improved
apply KAPPA to free-form generationQwen2.5 7B Instruct · GSM8k free-formaccuracy: 91.6 -> 90.7Hurt

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