natural-language-processing-2
Cross-Document Language Modeling
3260 papers • 126 benchmarks • 313 datasets
Involves pretraining language models to support multi-document NLP tasks. Source: Cross-Document Language Modeling Image Credit: Cross-Document Language Modeling
(Image credit: Papersgraph)
Benchmarks
These leaderboards are used to track progress in cross-document-language-modeling-2
Trend
Dataset
Best Model
Actions
MultiNews val
MultiNews val
MultiNews test
MultiNews test
Libraries
iUse these libraries to find cross-document-language-modeling-2 models and implementations
Datasets
Subtasks
No subtasks available.
Most implemented papers
Cross-Document Language Modeling
Arman Cohan, Ido Dagan, Iz Beltagy, Matthew E. Peters, Avi Caciularu, Arie Cattan•Fri Jan 01 2021
A new pretraining approach geared for multi-document language modeling, incorporating two key ideas into the masked language modeling self-supervised objective, which improves over recent long-range transformers by introducing dynamic global attention that has access to the entire input to predict masked tokens.
35