Will there be any instructions how to train own agents?
ln-e opened this issue · comments
Igor Bodnar commented
Yaroslav Ganin commented
We are currently not planning to release the training code since it relies on the internal infrastructure quite heavily. I'm happy, however, to provide guidance on reimplementing the training pipeline. Please see the original publication for details (http://proceedings.mlr.press/v80/ganin18a.html) and shoot me an email if something is unclear.