We are not satisfied with the current test data. We think it should be more diverse. We did try to use different authors for train and test data. Currently, you could achieve like 99.99% accuracy on the test set.
However, your trained model will not perform that good if you deploy it with your demo application for wider audience.
We are not satisfied with the current
testdata. We think it should be more diverse. We did try to use different authors fortrainandtestdata. Currently, you could achieve like 99.99% accuracy on thetestset.However, your trained model will not perform that good if you deploy it with your demo application for wider audience.