DL4J Feedforward Learner (classification) crash Catalina

[Eboone]

Hello everybody,

Me and multiple students, who have a Mac with OS Catalina, experience a crash when trying to run the DL4J Learner. Students with previous versions of Mac OS don’t experience this problem. This is the crash log:

Process: Knime [1228]
Path: /Applications/KNIME 4.1.0.app/Contents/MacOS/Knime
Identifier: org.knime.product
Version: 4.1.1 (4.1.1)
Code Type: X86-64 (Native)
Parent Process: ??? [1]
Responsible: Knime [1228]
User ID: 501

Date/Time: 2020-02-12 17:39:18.141 +0100
OS Version: Mac OS X 10.15.3 (19D76)
Report Version: 12
Anonymous UUID: 598ECB4F-5890-56C2-F68D-96620A6341D6

Crashed Thread: 69 Java: KNIME-Worker-35-DL4J Feedforward Learner (Classification)

Exception Type: EXC_BAD_ACCESS (SIGABRT)
Exception Codes: KERN_INVALID_ADDRESS at 0x0000000000000010
Exception Note: EXC_CORPSE_NOTIFY

Does anybody know how we can work ourselves around this problem?

Thank you in advance!

[DaveK]

Hi @Eboone,

did you by any chance install Anaconda Python at the same time when you upgraded the Mac OS version? Previously, there were some issues with DL4J and Anaconda which crashed KNIME. Unfortunately, the DL4J version used in KNIME is quite old, so it could be that the only way to fix the problem is an upgrade of the DL4J version. However, this could take a while on our side. Would be switching to the Keras Deep Learning Integration in the meantime an option? Generally, the Keras Integration has much more functionality and is recommended on our side.

Cheers,
David