Currently MuDA relies on compare-mt to analyse the performance of contextual models. However this delegation both makes MuDA harder to use AND brings alot of extra information to users who just want MuDA scores.
Ideally MuDA should have an interface that takes the tags (for both reference and the mt system) and outputs a score.
This is connected to #2
Currently MuDA relies on compare-mt to analyse the performance of contextual models. However this delegation both makes MuDA harder to use AND brings alot of extra information to users who just want MuDA scores.
Ideally MuDA should have an interface that takes the tags (for both reference and the mt system) and outputs a score. This is connected to #2