Sentence-BERT for spaCy

This package wraps sentence-transformers (also known as sentence-BERT) directly in spaCy. You can substitute the vectors provided in any spaCy model with vectors that have been tuned specifically for semantic similarity.

The models below are suggested for analysing sentence similarity, as the STS benchmark indicates. Keep in mind that sentence-transformers are configured with a maximum sequence length of 128. Therefore for longer texts it may be more suitable to work with other models (e.g. Universal Sentence Encoder).

Install

To install this package, you can run one of the following:

• pip install spacy_sentence_bert
• pip install git+https://github.com/MartinoMensio/spacy-sentence-bert.git

Usage

With this package installed

import spacy_sentence_bert
nlp = spacy_sentence_bert.load_model('en_bert_base_nli_cls_token')

Or if a specific model is installed (e.g. pip install https://github.com/MartinoMensio/spacy-sentence-bert/releases/download/en_bert_base_nli_cls_token-0.1.0/en_bert_base_nli_cls_token-0.1.0.tar.gz)

import spacy
nlp = spacy.load('en_bert_base_nli_cls_token')

import spacy
import spacy_sentence_bert
nlp_base = spacy.load('en')
nlp = spacy_sentence_bert.create_from(nlp_base, 'en_bert_base_nli_cls_token')
nlp.pipe_names

Full list of models https://docs.google.com/spreadsheets/d/14QplCdTCDwEmTqrn1LH4yrbKvdogK4oQvYO1K1aPR5M/edit#gid=0

The models, when first used, download to the folder defined with TORCH_HOME in the environment variables (default ~/.cache/torch).