Skip to Main content Skip to Navigation
Conference papers

Skim-Attention: Learning to Focus via Document Layout

Abstract : Transformer-based pre-training techniques of text and layout have proven effective in a number of document understanding tasks. Despite this success, multimodal pre-training models suffer from very high computational and memory costs. Motivated by human reading strategies, this paper presents Skim-Attention, a new attention mechanism that takes advantage of the structure of the document and its layout. Skim-Attention only attends to the 2dimensional position of the words in a document. Our experiments show that Skim-Attention obtains a lower perplexity than prior works, while being more computationally efficient. Skim-Attention can be further combined with long-range Transformers to efficiently process long documents. We also show how Skim-Attention can be used off-the-shelf as a mask for any Pre-trained Language Model, allowing to improve their performance while restricting attention. Finally, we show the emergence of a document structure representation in Skim-Attention.
Complete list of metadata
Contributor : Kim-Anh Laura Nguyen Connect in order to contact the contributor
Submitted on : Friday, September 3, 2021 - 12:04:08 PM
Last modification on : Monday, September 6, 2021 - 10:04:47 AM


Files produced by the author(s)


  • HAL Id : hal-03333889, version 1
  • ARXIV : 2109.01078


Laura Nguyen, Thomas Scialom, Jacopo Staiano, Benjamin Piwowarski. Skim-Attention: Learning to Focus via Document Layout. Findings of the 2021 Conference on Empirical Methods in Natural Language Processing (Findings of EMNLP 2021), Nov 2021, Punta Cana, Dominican Republic. ⟨hal-03333889⟩



Record views


Files downloads