JDK compatibility #6
Description
I use both JDK1.7 and JDK1.8 to run the same corpus. But the prediction result is different sometimes. I think it may be the compatibility problem while I'm not sure which parts cause this (may be the jdk sort implementation ?) . Please give some suggestions here. Thanks !