training at a depth of three implementing es-hyperneat reporting fails after first generation

[ai-nick]

i am having the issue of the reporting failing and possibly generations going on indefinitely with high population configs or depth > 2 es-hyperneat implementations, thoughts on what could be the culprit?