AI Paper Atlas

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

Deep Bidirectional Language-Knowledge Graph Pretraining

Michihiro Yasunaga, Antoine Bosselut, Hongyu Ren, Xikun Zhang, Christopher D Manning, Percy Liang, Jure Leskovec

NeurIPS 2022 · Large language models › Language, safety and interpretability · approach: Sign and cross-lingual translation · 86 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 joint MLM + LinkPred pretraining instead of MLM only (ablation)DRAGON · OBQAaccuracy: 67.2 -> 72.0Improved
add joint MLM + LinkPred pretraining instead of MLM only (ablation)DRAGON · CSQAaccuracy: 74.3 -> 76.0Improved
add MLM to LinkPred-only pretraining (ablation)DRAGON · OBQAaccuracy: 66.4 -> 72.0Improved
bidirectional text-KG fusion instead of concatenating at end (ablation)DRAGON · OBQAaccuracy: 68.0 -> 72.0Improved
use KG as graph instead of converting to sentences (ablation)DRAGON · OBQAaccuracy: 70.1 -> 72.0Improved
DistMult head instead of TransE or RotatE (ablation)DRAGON · OBQAaccuracy: 71.4 -> 72.0Improved
KG-augmented pretraining vs finetuning-only GreaseLMDRAGON · OBQAaccuracy: 66.9 -> 72.0Improved
KG-augmented pretraining on 10% finetuning dataDRAGON · PIQA (10% train)accuracy: 67.0 -> 72.3Improved
increase fusion layers from 5 to 7 without pretraining (ablation)GreaseLM · OBQAaccuracy: 66.9 -> 66.2Hurt
increase fusion layers from 5 to 7 with pretraining (ablation)DRAGON · OBQAaccuracy: 72.0 -> 72.8Improved
KG-augmented pretraining on biomedical LMBioLinkBERT-Large · MedQAaccuracy: 44.6 -> 47.5Improved
KG-augmented pretraining on complex reasoning (negation)RoBERTa · CSQA+OBQA dev negationaccuracy: 61.7 -> 75.2Improved

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