Text classification task.
Used the same model bin file (model was produced under java 10, the fasttext jar is also build under java 10) , when execute predict function under different java version (1.8 and 10), the accuracy under java 10 is higher than java 1.8 . Why? Thanks.