Break is a human annotated dataset of natural language questions and their Question Decomposition Meaning Representations (QDMRs). Break consists of 83,978 examples sampled from 10 question answering datasets over text, images and databases.

Break

83,978 examples sampled from 10 question answering datasets over text, images and databases.

"Break It Down: A Question Understanding Benchmark" (TACL 2020). Paper: https://arxiv.org/abs/2001.11770 Repo: https://github.com/tomerwolgithub/Break/tree/master/qdmr_parsing

Exact Match over the normalized form of the QDMR graph

The SARI metric for text simplification (Xu et al., TACL 2016)

Research

Research

Break

Leaderboard