MATE: Multi-view Attention for Table Transformer Efficiency

TL;DR

MATE introduces a sparse-attention Transformer architecture optimized for large web tables, enabling efficient modeling of extensive tabular data and achieving state-of-the-art results in table reasoning tasks.

Contribution

The paper presents MATE, a novel sparse-attention Transformer that scales linearly and effectively models large tables with a structure-aware bias, outperforming previous models.

Findings

Handles tables with over 8000 tokens efficiently.

Sets new state-of-the-art on three table reasoning datasets.

Improves HybridQA results by 19 points.

Abstract

This work presents a sparse-attention Transformer architecture for modeling documents that contain large tables. Tables are ubiquitous on the web, and are rich in information. However, more than 20% of relational tables on the web have 20 or more rows (Cafarella et al., 2008), and these large tables present a challenge for current Transformer models, which are typically limited to 512 tokens. Here we propose MATE, a novel Transformer architecture designed to model the structure of web tables. MATE uses sparse attention in a way that allows heads to efficiently attend to either rows or columns in a table. This architecture scales linearly with respect to speed and memory, and can handle documents containing more than 8000 tokens with current accelerators. MATE also has a more appropriate inductive bias for tabular data, and sets a new state-of-the-art for three table reasoning datasets. For HybridQA (Chen et al., 2020b), a dataset that involves large documents containing tables, we improve the best prior result by 19 points.