Problems on the F1 score

jayavardhanr / End-to-end-Sequence-Labeling-via-Bi-directional-LSTM-CNNs-CRF-Tutorial

Tutorial for End-to-end Sequence Labeling via Bi-directional LSTM-CNNs-CRF

121 stars 71 forks source link

Problems on the F1 score #2

Closed yueguangxichen closed 6 years ago

yueguangxichen commented 6 years ago

I have run the program but I can't get the same F1 score as listed in the paper, it is far away from the best score,it's just 0.86

jayavardhanr commented 6 years ago

We did not tune the model to achieve the results mentioned in the paper. Additional hyperparameter tuning has to be performed to achieve better results. All the hyperparameters are defined in the notebook, you can play around with them to improve the performance.