search

richardpaulhudson / coreferee

Coreference resolution for English, French, German and Polish, optimised for limited training data and easily extensible for further languages
MIT License
102 stars 16 forks source link

Enable packaging nlp pipelines containing Coreferee #14

Closed adrianeboyd closed 1 year ago

adrianeboyd commented 1 year ago

If the vectors model is the current pipeline, the model data hasn't necessarily been loaded on init, so you can't check the vectors shape or run the vectors pipeline at this point.

A better check (at a later point) would be for vectors_nlp.vocab.vectors.shape[1] to get the vector length.

The tests need a roundtrip to_disk / from_disk for a non-trf model to cover this case.

adrianeboyd commented 1 year ago

@richardpaulhudson Can you look into what's going on with the CI here?

richardpaulhudson commented 1 year ago

Something seems to have changed with the images offered by GitHub in the two days since I tested this in my own repo, because the same error now occurs there. It may be that the value of ubuntu-latest has changed. I'll put together another PR to fix this.

adrianeboyd commented 1 year ago

I think the test needs to be formatted one last time, but then I think this looks fine.