search

emorynlp / elit

Emory Language and Information Toolkit
Other
37 stars 2 forks source link

Integrated Coreference Resolution #1

Closed lxucs closed 3 years ago

lxucs commented 3 years ago

The model has no problem with the pre-tokenized tokens and it can be run successfully by local tests (test_service_coref_doc.py). I also added examples in client.py for usage of both document-coref and online-coref. The issue is likely from the server setting/processing; even the original code without coref is having trouble passing the client tests. You might have the latest code so you can test on your side.

lxucs commented 3 years ago

One thing I did find is that I had different output format; I have made the output format changes accordingly, and updated the README in docs.

hankcs commented 3 years ago

I'm sorry but your new update didn't work out. Could you make a unitest for the 3 formats as I did in this file?

https://github.com/emorynlp/elit/blob/feab4f3994ec900072a8f45eab47e87a63a5624e/tests/test_joint_model.py

You might also want to put the unittest into travis config:

https://github.com/emorynlp/elit/blob/main/.travis.yml#L17

hankcs commented 3 years ago

The CI is back to normal now. Once the unittest passed CI like this one, then we can merge it.

lxucs commented 3 years ago

I will close this PR and create a new PR in another branch.