RL4J: SyncLearning lacks necessary training control

[AlexDBlack]

From @viluon on October 9, 2017 22:56

https://github.com/deeplearning4j/rl4j/blob/273564b2e35294a83893688b9033013fe24129b6/rl4j-core/src/main/java/org/deeplearning4j/rl4j/learning/sync/SyncLearning.java#L36

The step counter lacks setters, which makes all Learning-based classes practically unusable after the first .train() call.

• stepCounter definition

Copied from original issue: deeplearning4j/rl4j#70

[AlexDBlack]

From @saudet on October 11, 2017 5:43

We could recreate these objects, if that's the intention. Could you explain your use case?

[viluon]

@saudet continuous learning (improving the net while it is being used).

[saudet]

@viluon Sure, let's add a @Setter there or switch to @Data. Could you send a pull request?

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.