GEM-benchmark / GEM-metrics

Automatic metrics for GEM tasks
https://gem-benchmark.com
MIT License
61 stars 20 forks source link

Use multi-references #97

Closed jordiclive closed 2 years ago

jordiclive commented 2 years ago

Fixes DART and WebNLG issues in #27 . Add code to look to see if have references key of value list[str]. Also, add DART test back in.

tuetschek commented 2 years ago

@jordiclive thanks, this definitely solves the issue! I'd just like to keep the base class general, so I added a modified version in #98. Could you please have a look if that works for you?