Skip to content
#

challenge-set

Here is 1 public repository matching this topic...

TamilLingBench: English→Tamil machine translation benchmark for obligatory verb agreement (rationality, gender, number). Tests whether MT evaluation metrics such as COMET, xCOMET, MetricX and CometKiwi catch wrong Tamil agreement. Code, data and model outputs for the WMT 2026 paper "Obligatory Slots".

  • Updated Sep 12, 2026
  • Python

Add this topic to your repo

To associate your repository with the challenge-set topic, visit your repo's landing page and select "manage topics."

Learn more