spaCy Project: Part-of-speech Tagging & Dependency Parsing packaging problem

[kanayer]

I used the project to train the spacy tagger and parser for Thai language on my custom dataset. Preprocessing, training and evaluation worked fine, but the packaging gives the following error:

File "/usr/local/lib/python3.9/site-packages/catalogue/init.py", line 95, in get raise RegistryError(
catalogue.RegistryError: Cant't find 'spacy.MultiHashEmbed.v1' in registry spacy -> architectures. Available names: spacy-legacy.CharacterEmbed.v1, spacy-legacy.HashEmbedCNN.v1, spacy-legacy.MaxoutWindowEncoder.v1, spacy-legacy.MishWindowEncoder.v1, spacy-legacy.MultiHashEmbed.v1, spacy-legacy.TextCatBOW.v1, spacy-legacy.TextCatCNN.v1, spacy-legacy.TextCatEnsemble.v1, spacy-legacy.Tok2Vec.v1, spacy-legacy.TransitionBasedParser.v1, spacy-transformers.Tok2VecTransformer.v1, spacy-transformers.TransformerListener.v1, spacy-transformers.TransformerModel.v1, spacy.CharacterEmbed.v2, spacy.EntityLinker.v1, spacy.HashEmbedCNN.v2, spacy.MaxoutWindowEncoder.v2, spacy.MishWindowEncoder.v2, spacy.MultiHashEmbed.v2, spacy.PretrainCharacters.v1, spacy.PretrainVectors.v1, spacy.SpanCategorizer.v1, spacy.Tagger.v1, spacy.TextCatBOW.v2, spacy.TextCatCNN.v2, spacy.TextCatEnsemble.v2, spacy.TextCatLowData.v1, spacy.Tok2Vec.v2, spacy.Tok2VecListener.v1, spacy.TorchBiLSTMEncoder.v1, spacy.TransitionBasedParser.v2

Details about the OS and versions of spaCy and Python:

spaCy version    3.1.2                         
OS               macOS 64bit
Python version   3.9.9  

Can you please help me understand what is wrong and how can I fix the issue?

[adrianeboyd]

Hi, this bug is fixed in v3.1.3.

[adrianeboyd]

Hmm, can you try explicitly upgrading spacy-legacy with pip install -U spacy-legacy?

[kanayer]

Thank you, it worked! One question, how to load the model after packaging it? I went into dist folder and did pip install thai_ud-1.0.0 and it is installed into /Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages. But how do I load my spacy model in IDE?

I load spacy model the following way:
nlp = spacy.load("/Users/kkk/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/th-ud-thai-1.0.0")
but get the following error:

OSError: [E050] Can't find model '/Users/kkk/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/th-ud-thai-1.0.0'. It doesn't seem to be a Python package or a valid path to a data directory.

[adrianeboyd]

Once you've installed the package, you can load it with the package name that you see in pip list, so it's exactly the same as for en_core_web_sm:

spacy.load("th_ud_thai")

For languages with additional requirements, you can add them to the package requirements in meta.json so they're installed automatically with the package like any other python dependency. You can pass in a partial meta.json to spacy package --meta that contains the additional requirements:

{
  "requirements": ["pythainlp>=2.0"]
}

[kanayer]

But then I get an error

"ModuleNotFoundError: No module named 'th-ud-thai'"

Is there anything I am doing wrong?

[adrianeboyd]

Ah, I forgot that the pip list output doesn't convert the dashes into underscrores. Try: spacy.load("th_ud_thai")