Some issues about the strong BC baseline ACT

younggyoseo / CQN

Coarse-to-fine Q-Network

30 stars 0 forks source link

Some issues about the strong BC baseline ACT #2

Closed feriorior closed 2 weeks ago

feriorior commented 2 months ago

Thanks a lot to the author for this wonderful work. In order to better understand its contribution, I try to train the ACT model with original codebase and collected data. However, I notice that there is a severe performance degradation. Could the author provide some changed hyper-parameters adapted to RLBench data or open source the re-implementation version? Thank you very much, looking forward to your reply.

younggyoseo commented 1 month ago

Hi, I'd like to recommend you to check ACT implementation and training scripts at https://github.com/robobase-org/robobase.