There exist a strange situation that certain sentences would have different trees between the online tupa and the one installed (TUPA version 1.3.9) by pip. It is very confusing, which one should I follow? Here is an example of a sentence,
Neither is guaranteed to be 100% accurate. The demo was using a different trained model (with the same settings but a different random seed). Now it's updated to use the same pre-trained model as in the release.
There exist a strange situation that certain sentences would have different trees between the online tupa and the one installed (TUPA version 1.3.9) by pip. It is very confusing, which one should I follow? Here is an example of a sentence,
Tree produces by the online TUPA:
Tree produces by the TUPA installed by pip: