Bag of Tricks for Efficient Text Classification

TL;DR

This paper presents fastText, a simple and efficient text classification method that achieves comparable accuracy to deep learning models while being significantly faster to train and evaluate, enabling large-scale applications.

Contribution

The paper introduces fastText, a novel baseline for text classification that is both fast and accurate, outperforming many existing methods in speed and scalability.

Findings

fastText matches deep learning accuracy

train on over one billion words in under ten minutes

classify half a million sentences in less than a minute

Abstract

This paper explores a simple and efficient baseline for text classification. Our experiments show that our fast text classifier fastText is often on par with deep learning classifiers in terms of accuracy, and many orders of magnitude faster for training and evaluation. We can train fastText on more than one billion words in less than ten minutes using a standard multicore~CPU, and classify half a million sentences among~312K classes in less than a minute.