Is there a way to get vocabulary (vector representation of all words) of the pre-trained model of word embedding?

[valmirosjunior]

When I make something like:

WordEmbeddingsModel().pretrained("glove_6B_300", "xx")

Is it possible to make a call for a method to get the whole vocabulary of the model?
Something like a dictionary with all words and their vector representation.

[wgeul]

I face the same issue, I want to verify the representativeness of the embeddings by running it through an encoder-decoder model. I need the vocabulary to assess the accuracy.

[maziyarpanahi]

There is no public method to make the whole vocab-vector available in WordEmbeddingsModel (specially in Python), we can open a feature request for this.

[maziyarpanahi]

will track the progress here: #13930

[maziyarpanahi]

I believe this is doable for all 3 annotators (WordEmbeddingsModel, Doc2VecModel, and Word2VecModel) in Spark NLP 5.1.1 release: https://github.com/JohnSnowLabs/spark-nlp/releases/tag/5.1.1