Hello. This PR adds a warning about CGnet model variance, which can be high among several different training runs. It offers a suggestion to use CGSchNet instead, which seems to be more stable. Let me know if there is anything else to change or improve.
Development:
[ ] Add tests
Checks:
nosetests
Hello. This PR adds a warning about CGnet model variance, which can be high among several different training runs. It offers a suggestion to use CGSchNet instead, which seems to be more stable. Let me know if there is anything else to change or improve.