Trained NER models are not loadable

[YanhaoYang]

I use some trained NER models to recognise entity names in some specific domains. After upgrading to v1.8.0, I found those models could not be loaded.

It is easy to reproduce the problem by add following code to the official example, https://github.com/explosion/spaCy/blob/master/examples/training/train_ner.py#L87:

    model_path = Path('./ner')
    ner2 = EntityRecognizer.load(model_path, nlp.vocab)
    doc = nlp.make_doc('Who is Shaka Khan?')
    nlp.tagger(doc)
    ner2(doc)
    for word in doc:
        print(word.text, word.orth, word.lower, word.tag_, word.ent_type_, word.ent_iob)

This will result in following error message:

Traceback (most recent call last):
  File "train_ner.py", line 104, in <module>
    main('ner')
  File "train_ner.py", line 94, in main
    ner2 = EntityRecognizer.load(model_path, nlp.vocab)
  File "spacy/syntax/parser.pyx", line 149, in spacy.syntax.parser.Parser.load (spacy/syntax/parser.cpp:6845)
KeyError: 'actions'

It seems ner.cfg["actions"] is not necessary for NER training, but the loading code will read it anyway.

Info about spaCy

• spaCy version: 1.8.0
• Platform: Linux-4.10.0-19-generic-x86_64-with-debian-8.7
• Python version: 3.6.1
• Installed models: en

[ines]

Closing this and making #1046 the master issue. Work in progress for spaCy v2.0!

[lukewendling]

Anyone found a workaround to load trained NER models in spacy <2.0 ?

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.