1.have you use the feature of corresponding arc labels of words? I can't find this section in your code.
2.why you not use the cube activation function?
3.whether you have achieved the same performance as the paper(UAS test:92.0)
thank you !
from China : )

Hi,
Could you please tell me what dataset you use?
I just know the format is CONLL format but don't know where the dataset is from.
Sincerely,
Charlie

when parsing dependency without label(such as nsub,nmod), We can't get the label of s0's left most child or the label of s0's child's child. So we can't use it as input.
In extract_for_current_state, it will get the left most child of s0 and get the label of this child. But for parsing without label, this information is not available.

hi ! maybe this code can only predict transition(left, right or shift), but can not predict arc labels. isn't this problem?

Where is the model after training?
How to use this model to analyze a sentence?

This is not like stanford basic dependencies format. What version do you use Stanford parser?
What version do you use Stanford POS tagger?
1 In _ ADP IN _ 5 case _ _
2 an _ DET DT _ 5 det _ _
3 Oct. _ PROPN NNP _ 5 compound _ _
4 19 _ NUM CD _ 5 nummod _ _
5 review _ NOUN NN _ 45 nmod _ _
6 of _ ADP IN _ 9 case _ _
7 _ PUNCT _ 9 punct _ _
8 The _ DET DT _ 9 det _ _
9 Misanthrope _ NOUN NN _ 5 nmod _ _
10 '' _ PUNCT '' _ 9 punct _ _
11 at _ ADP IN _ 15 case _ _
12 Chicago _ PROPN NNP _ 15 nmod:poss _ _

Sorry to disturb you. But what is the name of the dataset used here？

Hi,
How can we print out the test results as conll files, currently it only created the model and evaluates it, but there is no result conll.

Thank you

Firat

I train the model ,but do not get arc_label