Extract all (word, vector) pairs from vocab

Question

How to : Extract all (word, vector) pairs from spacy Vocab ?

iteration like:

sort([ w.text for w in nlp.vocab ])

array(['\t', '\n', ' ', '"', "'", "''", "'Cause", "'Cos", "'Coz", "'Cuz", "'S", "'bout", "'cause", "'cos", "'coz", "'cuz", "'d", "'em", "'ll", "'m", "'nuff", "'re", "'s",
   "'ve", "'y", '(', '(*_*)', '(-8', '(-:', '(-;', '(-_-)', ...

returns only ~800 weird items

score 0 · Answer 1 · answered Oct 07 '22 at 00:34

0

nlp.vocab has 764 words. Only radicals:

import spacy
nlp = spacy.load("en_core_web_sm")
len([ w.text for w in nlp.vocab ])

output: 764

nlp.vocab has no vector:

[ (w.text, w.has_vector) for w in nlp.vocab if w.has_vector ]

output: []

You may find information map at: https://spacy.io/api

answered Oct 07 '22 at 00:34

Mário César Fracalossi Bais

340
2
5

https://stackoverflow.com/questions/68876169/what-is-the-difference-between-nlp-vocab-and-nlp-vocab-strings – Mário César Fracalossi Bais Oct 07 '22 at 00:40
u need to load _md or _lg model to have access to the vectors ... – sten Oct 07 '22 at 02:28
nlp.vocab['king'].vector => array([-1.13e-01, -4.19e+00, -1.85e+00, ...........]) – sten Oct 07 '22 at 02:30

Extract all (word, vector) pairs from vocab

1 Answers1