why is the NT F1 exactly the same as F1 score running this code?

[ZhichaoDuan]

I cloned this repo and try to replicate the results and I found that each time I evaluate the model performance after training, the NT f1 score and the f1 score are the exact same which is odd..