Giving BERT a Calculator: Finding Operations and Arguments with Reading Comprehension

08/31/2019

∙

by   Daniel Andor, et al.

∙

Google

∙

0

∙

share

Reading comprehension models have been successfully applied to extractive text answers, but it is unclear how best to generalize these models to abstractive numerical answers. We enable a BERT-based reading comprehension model to perform lightweight numerical reasoning. We augment the model with a predefined set of executable 'programs' which encompass simple arithmetic as well as extraction. Rather than having to learn to manipulate numbers directly, the model can pick a program and execute it. On the recent Discrete Reasoning Over Passages (DROP) dataset, designed to challenge reading comprehension models, we show a 33 model can learn to predict new operations when appropriate in a math word problem setting (Roy and Roth, 2015) with very few training examples.