AI Paper Atlas

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

Towards Semantics-Enhanced Pre-Training: Can Lexicon Definitions Help Learning Sentence Meanings?

Xuancheng Ren, Xu Sun, Houfeng Wang, Qun Liu

AAAI 2021 · Large language models › Language, safety and interpretability · approach: Sign and cross-lingual translation · 5 citations (OpenAlex)

Read the paper

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 SemPre word-definition and relational pre-trainingRoBERTa-large · WG1 and WG2lowest rank: 332 -> 86; 262 -> 130Improved
Add SemPre word-definition and relational pre-trainingRoBERTa-large · CA, WSC, SM, SMR, ARCT1, and ARCT2accuracy: 96.2 -> 96.7; 69.3 -> 73.5; 79.2 -> 80.4; 47.3 -> 48.4; 54.3 -> 58.3; 60.0 -> 60.9Improved
Add SemPre word-definition and relational pre-trainingRoBERTa-large · SWAG and HellaSwagaccuracy: 76.1 -> 75.9; 48.9 -> 48.5Hurt
Add SemPre word-definition and relational pre-trainingRoBERTa-large · WiCaccuracy: 74.6 -> 75.7; 70.3 -> 72.1Improved
Add SemPre word-definition and relational pre-trainingRoBERTa-large · PIQAaccuracy: 81.3 -> 81.6Improved
Add word-definition training (ablation)RoBERTa-large semantics-focused pre-training · CATs average and WiC validationaccuracy: 66.4 -> 67.6; 74.6 -> 74.8Improved
Use word pairs alone, without definitions (ablation)RoBERTa-large semantics-focused pre-training · CATs average and WiC validationaccuracy: 66.4 -> 64.3; 74.6 -> 74.3Hurt
Remove the word-masking objective (ablation)RoBERTa-large + SemPre · CATs average and WiC validationaccuracy: 67.8 -> 67.1; 75.7 -> 75.1Hurt
Remove the definition-masking objective (ablation)RoBERTa-large + SemPre · CATs average and WiC validationaccuracy: 67.8 -> 67.4; 75.7 -> 74.9Hurt
Remove the relation-prediction objective (ablation)RoBERTa-large + SemPre · CATs average and WiC validationaccuracy: 67.8 -> 67.7; 75.7 -> 74.6Hurt

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