·
38 commits
to main
since this release
After finishing the last brick that allows you to keep up with the newest Transformer version, we present you the New Year release of Q-Align on version 1.1.0, which is able to:
- Train on IQA/VQA/IAA datasets.
- Evaluate on IQA/VQA/IAA datasets.
- Initiate an OneScorer (though the
AutoModel("q-future/one-align")is out, but who will not like an editable one?)
Please enjoy!