Dear participants,
We are happy to share the ranking of the systems submitted to the MRL shared task.
As a reminder, for the inflection and reinflection tasks the metric edit distance, and for analysis the metric is averaged F1 score with the lemma being treated as an upweighted feature.
Task 1 - inflection:
-
KUIS-AI - 0.266
- UBC - 0.496
- mT5 baseline - 2.577
- transformer baseline - 3.278
Task 2 - reinflection:
-
KUIS-AI - 0.560
- UBC - 0.982
- mT5 baseline - 2.826
- transformer baseline - 4.642
Task 3 - analysis
-
KUIS-AI - 0.950
-
Göttingen - 0.940
- UBC - 0.914
- mT5 baseline - 0.845
- transformer baseline - 0.800