word_forms VS spacy-experimental

However, I've had a peek into the word-forms source code and it seems very easy to translate to Java, it's basically parsing a big amount of data from text files. In my opinion, creating a Java version would be a nice open-source project, if there's anyone willing to do it.

Trying to work with the newly released coreference resolution pipeline

I think the details are mentioned here: https://github.com/explosion/spacy-experimental/releases/tag/v0.6.0

I can't speak for the parent commenter, but there is ofte. code 'around' the machine learning code that benefits from high-performance implementations. To give two examples:

1. We recently implemented an edit tree lemmatizer for spaCy. The machine learning model predicts labels that map to edit trees. However, in order to lemmatize tokens, the trees need to be applied. I implemented all the tree wrangling in Cython to speed up processing and save memory (trees can be encoded as compact C unions):

https://github.com/explosion/spaCy/blob/master/spacy/pipelin...

2. I am working on a biaffine parser for spaCy. Most implementations of biaffine parsing use a Python implementation of MST decoding, which is unfortunately quite slow. Some people have reported it to dominate parsing time (rather than a rather expensive transformer + biaffine layer). I have implemented MST decoding in Cython and it barely shows up in profiles:

https://github.com/explosion/spacy-experimental/blob/master/...

Nós iremos utilizar o template do edit_tree_lemmatizer contido da pasta de projetos do repositório https://github.com/explosion/spacy-experimental e modificaremos para treinar um modelo em português em vez de alemão.