Testing trained images with bad accuracy.

[DLH06]

I try to train my own dataset. I get loss, perflexity and precision might be good.
But when I run model to test on same dataset I trained. It shows bad.
Any help? Thanks
Beside, I test author's pre-train model, it happens:
2019-12-11 05:06:25,839 root INFO 0.000000 out of 647 correct