"memoryManagerClazz" is null

SameDiff
1

[main] INFO org.nd4j.linalg.factory.Nd4jBackend - Loaded [JCublasBackend] backend
[main] INFO org.nd4j.nativeblas.NativeOpsHolder - Number of threads used for linear algebra: 32
Exception in thread “main” java.lang.ExceptionInInitializerError
at com.pcng.idea.ner.NerTrainBySameDiff.main(NerTrainBySameDiff.java:64)
Caused by: java.lang.RuntimeException: java.lang.NullPointerException: Cannot invoke “java.lang.Class.newInstance()” because “memoryManagerClazz” is null
at org.nd4j.linalg.factory.Nd4j.initWithBackend(Nd4j.java:5208)
at org.nd4j.linalg.factory.Nd4j.initContext(Nd4j.java:5073)
at org.nd4j.linalg.factory.Nd4j.(Nd4j.java:299)
… 1 more
Caused by: java.lang.NullPointerException: Cannot invoke “java.lang.Class.newInstance()” because “memoryManagerClazz” is null
at org.nd4j.linalg.factory.Nd4j.initWithBackend(Nd4j.java:5151)
… 3 more

Error when run SameDiff app in CUDA in SNAPSHOT.
M2.1 no such problem.

2

@SidneyLann I just kicked off snapshot builds. It’s probably just missing. Thanks for flagging.

3

The latest build still has the problem.

4

@SidneyLann I’m running another build with the latest changes. Let me know if you still run in to that after this and if so I’ll look in to this next. This is occurring within your uber jar right?

Again, I don’t like magic. I only want to look at real root causes like we did with the rnn null inputs issue.
I know where that issue comes from and it comes from a missing resource file or a missing property.

Ideally I want the circumstances that lead to either one of those issues. Anything you could help me with on that front would be appreciated.