Yale-LILY / dart

Dataset for NAACL 2021 paper: "DART: Open-Domain Structured Data Record to Text Generation"

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

The number of examples in file dart-v1.1.1-full-train.json is much smaller than 82,191

syw1996 opened this issue · comments

The example number of Dart declaimed in Readme file and paper is 82,191, while the actual number of training set in file dart-v1.1.1-full-train.json is 30,526 which is much smaller than 82,191.

@syw1996 Thank you for your message! "Total number of examples" refers to the total number of annotations (not triplesets) across the train, devel, and test sets. Some triplesets may have >1 annotations. There are 30,526 triplesets in DART v1.1.1's train set; there are 82,191 annotations across all parts of DART v1.1.1.