Hou Pong (Ken) Chan

Hi, did you use Pytorch 0.4 to run our code? may I know which command you use so that I can replicate the problem?

Hi, can you use Pytorch 0.4 to run it? I think the reason is that the PyTorch APIs have some changes and the variables returned by the functions are different....

Hi, I think you can also try Pytorch 1.0.0 I heard that the API of Pytorch 1.0.0 has no major changes compared with Pytorch 0.4.0.

Hi, you need to use ml loss to train a model by yourself according to our instruction in "Train ML".

If you have run the ML training, then you will have the ml trained models. You just need to set [path_to_ml_pretrained_model] to path of the ml trained model that has...

It sounds like some of the values in dcg_max_array are zero or NaN. Can you please wrap the `ndcg_array = dcg_array / dcg_max_array` line into a `try execpt` block and...

Hi, I double-checked the code. The script works even some of the values are zeros, it will just print out a warning, but not an error. The `np.nan_to_num(ndcg_array)` in the...

Yes, I know, but my point is the division by zero is just a warning, but not an error, it will not cause the script to terminate, and the NaN...

Hi, the results under `====all====` are the F1 scores for the prediction includes both present and absent keyphrases. The results under `====present====` are the F1 scores for present predicted keyphrases...

Hi, we report the macro F1 scores in our paper since they are used by previous keyphrase generation literature. Macro F1 scores and micro F1 scores are two different ways...