AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Thrust: Adaptively Propels Large Language Models with External Knowledge

Xinran Zhao, Hongming Zhang, Xiaoman Pan, Wenlin Yao, Dong Yu, Jianshu Chen

NeurIPS 2023 · Large language models › Retrieval and knowledge · approach: Knowledge editing in LLMs · 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
add external knowledge only to Thrust-selected instances instead of default selectionIAPEK · e-SNLI, UnifiedQA-3b scarceaccuracy: 69.1 -> 86.3Improved
add external knowledge only to Thrust-selected instancesIAPEK · NQ, UnifiedQA-3b scarceQA-F1: 12.8 -> 35.9Improved
add external knowledge only to Thrust-selected instancesIAPEK · TriviaQA, UnifiedQA-3b scarceQA-F1: 39.2 -> 68.3Improved
Thrust selection on abundant portion vs scarce portionIAPEK-Thrust · UnifiedQA-3b MC classification and QAgain: 6.1%, 13.56% scarce vs 2.8% and 6.8% abundantMixed
Thrust selection on ARC-CIAPEK-Thrust · ARC-C, UnifiedQA-3b scarceaccuracy: 64.5 -> 63.9Hurt
use Thrust instead of BM25 for ranking difficulty (ablation)IAPEK · UnifiedQA-3b scarce, TRECQA-F1: 20.4 -> 37.8Improved
use BM25 instead of Thrust (ablation)IAPEK · Web Questions, NQQA-F1: 20.9 -> 19.0Hurt
vary the decoder layer used for hidden representations (ablation)Thrust · UnifiedQA-3b, most tasksperformanceNo clear effect
use middle-layer representations (ablation)Thrust · StrategyQA and Web QuestionsperformanceHurt
remove cluster size from Thrust (ablation)Thrust · FS-NLI hardness correlationcorrelation: 0.45 -> 0.23Hurt
replace vector with scalar (remove direction) (ablation)Thrust · FS-NLI hardness correlationcorrelation: 0.45 -> 0.19Hurt
remove distance (ablation)Thrust · FS-NLI hardness correlationcorrelation: 0.45 -> 0.06Hurt

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