ValueError: [E010] Word vectors set to length 0.

,
1

Downloaded prodigy, installed with wheel on Mac OSX with Anaconda. Spacy is set to 2.0.11 and works fine with:

import spacy
nlp = spacy.load("en_vectors_web_lg")

But when using prodigy with:

$ prodigy terms.teach swear_words en_vectors_web_lg --seeds swear_words.txt

I get an error:

ValueError: [E010] Word vectors set to length 0. This may be because you don't have a model installed or loaded, or because your model doesn't include word vectors. For more info, see the docs:

But if I run the above prodigy command but as:

$ prodigy terms.teach swear_words en_core_web_lg --seeds swear_words.txt

It works fine, what are the reasons for this? Or does ‘en_core_web_lg’ also encompass the 300glove word-vectors that ‘en_vectors_web_lg’ offers?

Any help is greatly appreciated

2

Thanks, looking into this.

In the meantime, using en_core_web_lg should work fine, it does have the same vectors as en_vectors_web_lg, but with a slightly smaller vocabulary.

You can also download some of the FastText vectors and use those, if you prefer: https://fasttext.cc/docs/en/english-vectors.html

Just run spacy init-model /path/to/model -V /path/to/vectors.zip after downloading the vectors. You can then pass your model directory as an argument to Prodigy, to use the vectors.