explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2025-10-31 07:57:35 +03:00
Code Issues Packages Projects Releases Wiki Activity

Update docs [ci skip]

Browse Source
This commit is contained in:
Ines Montani 2020-09-17 17:12:51 +02:00
parent ec751068f3
commit a2c8cda26f
1 changed files with 32 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

60
website/docs/usage/embeddings-transformers.md
View File

@ -31,18 +31,18 @@ to predict. Otherwise, you could try using a "one-shot learning" approach using
<Accordion title="What’s the difference between word vectors and language models?" id="vectors-vs-language-models"> <Accordion title="What’s the difference between word vectors and language models?" id="vectors-vs-language-models">
[Transformers](#transformers) are large and powerful neural networks that give [Transformers](#transformers) are large and powerful neural networks that give
you better accuracy, but are harder to deploy in production, as they require a GPU to run you better accuracy, but are harder to deploy in production, as they require a
effectively. [Word vectors](#word-vectors) are a slightly older technique that GPU to run effectively. [Word vectors](#word-vectors) are a slightly older
can give your models a smaller improvement in accuracy, and can also provide technique that can give your models a smaller improvement in accuracy, and can
some additional capabilities. also provide some additional capabilities.
The key difference between word-vectors and contextual language The key difference between word-vectors and contextual language models such as
models such as transformers is that word vectors model **lexical types**, rather transformers is that word vectors model **lexical types**, rather than _tokens_.
than _tokens_. If you have a list of terms with no context around them, a transformer If you have a list of terms with no context around them, a transformer model
model like BERT can't really help you. BERT is designed to understand language like BERT can't really help you. BERT is designed to understand language **in
**in context**, which isn't what you have. A word vectors table will be a much context**, which isn't what you have. A word vectors table will be a much better
better fit for your task. However, if you do have words in context — whole sentences fit for your task. However, if you do have words in context — whole sentences or
or paragraphs of running text — word vectors will only provide a very rough paragraphs of running text — word vectors will only provide a very rough
approximation of what the text is about. approximation of what the text is about.
Word vectors are also very computationally efficient, as they map a word to a Word vectors are also very computationally efficient, as they map a word to a
@ -484,28 +484,32 @@ training.
## Static vectors {#static-vectors} ## Static vectors {#static-vectors}
If your pipeline includes a word vectors table, you'll be able to use the If your pipeline includes a **word vectors table**, you'll be able to use the
`.similarity()` method on the `Doc`, `Span`, `Token` and `Lexeme` objects. `.similarity()` method on the [`Doc`](/api/doc), [`Span`](/api/span),
You'll also be able to access the vectors using the `.vector` attribute, or you [`Token`](/api/token) and [`Lexeme`](/api/lexeme) objects. You'll also be able
can look up one or more vectors directly using the `Vocab` object. Pipelines to access the vectors using the `.vector` attribute, or you can look up one or
with word vectors can also use the vectors as features for the statistical more vectors directly using the [`Vocab`](/api/vocab) object. Pipelines with
models, which can improve the accuracy of your components. word vectors can also **use the vectors as features** for the statistical
models, which can **improve the accuracy** of your components.
Word vectors in spaCy are "static" in the sense that they are not learned Word vectors in spaCy are "static" in the sense that they are not learned
parameters of the statistical models, and spaCy itself does not feature any parameters of the statistical models, and spaCy itself does not feature any
algorithms for learning word vector tables. You can train a word vectors table algorithms for learning word vector tables. You can train a word vectors table
using tools such as Gensim, word2vec, FastText or GloVe. There are also many using tools such as [Gensim](https://radimrehurek.com/gensim/),
word vector tables available for download. Once you have a word vectors table [FastText](https://fasttext.cc/) or
you want to use, you can convert it for use with spaCy using the `spacy init vocab` [GloVe](https://nlp.stanford.edu/projects/glove/), or download existing
command, which will give you a directory you can load or refer to in your training pretrained vectors. The [`init vocab`](/api/cli#init-vocab) command lets you
configs. convert vectors for use with spaCy and will give you a directory you can load or
refer to in your [training configs](/usage/training#config).
When converting the vectors, there are two ways you can trim them down to make <Infobox title="Word vectors and similarity" emoji="📖">
your package smaller. You can _truncate_ the vectors with the `--truncate-vectors`
option, which will remove entries for rarer words from the table. Alternatively, For more details on loading word vectors into spaCy, using them for similarity
you can use the `--prune-vectors` option to remap rarer words to the closest vector and improving word vector coverage by truncating and pruning the vectors, see
that remains in the table. This allows the vectors table to return meaningful the usage guide on
(albeit imperfect) results for more words than you have rows in the table. [word vectors and similarity](/usage/linguistic-features#vectors-similarity).
</Infobox>
### Using word vectors in your models {#word-vectors-models} ### Using word vectors in your models {#word-vectors-models}