Does Matlab Text Analytics toolbox contain the Fasttext Pre-trained word vectors from Common Crawl (600B tokens)
2 views (last 30 days)
Does Matlab Text Analytics toolbox contain the Fasttext pre-trained word vectors from Common Crawl (600B tokens)? It seems to have pertained vectors from a 16billion token data set from fastext. I am wondering if it has the Common Crawl (6000B tokens) Pre-trained word vectors, or if there is a Matlab scrip the add this to the Matlab word embedding model. The vectors are here https://fasttext.cc/docs/en/english-vectors.html
Actually just a Matlab script that would read the Common Crawl 600B file and output a the vectors and string of the corresponding words would be sufficient.
Thanks in advance!
MathWorks Text Analytics Toolbox Team on 16 Apr 2020
Text Analytics Toolbox does not itself include this embedding, but you can download it (as you said in your question) and load it into MATLAB as follows:
>> emb = readWordEmbedding("crawl-300d-2M.vec.zip")
wordEmbedding with properties:
Vocabulary: [1×1999995 string]