We're currently assuming aligngments and metadata are in a particular format, which is a result of the approaches described in the paper repo. We should bring this process in here, locally. There are probably some improvements we can make, and likely some steps we can cut out.
We're currently assuming aligngments and metadata are in a particular format, which is a result of the approaches described in the paper repo. We should bring this process in here, locally. There are probably some improvements we can make, and likely some steps we can cut out.
See https://github.com/jeromekelleher/sc2ts-paper/issues/153