spaCy-wrap: For Wrapping fine-tuned transformers in spaCy pipelines
spaCy-wrap is minimal library intended for wrapping fine-tuned transformers from the Huggingface model hub in your spaCy pipeline allowing inclusion of existing models within SpaCy workflows.
As for as possible it follows a similar API as spacy-transformers.
Installation
Installing spacy-wrap is simple using pip:
pip install spacy_wrap
There is no reason to update from GitHub as the version on PyPI should always be the same as on GitHub.
Example
The following shows a simple example of how you can quickly add a fine-tuned transformer model from the Huggingface model hub. In this example we will use the sentiment model by Barbieri et al. (2020) for classifying whether a tweet is positive, negative or neutral. We will add this model to a blank English pipeline:
import spacy
import spacy_wrap
nlp = spacy.blank("en")
config = {
"doc_extension_trf_data": "clf_trf_data", # document extention for the forward pass
"doc_extension_prediction": "sentiment", # document extention for the prediction
"labels": ["negative", "neutral", "positive"],
"model": {
"name": "cardiffnlp/twitter-roberta-base-sentiment", # the model name or path of huggingface model
},
}
transformer = nlp.add_pipe("classification_transformer", config=config)
transformer.model.initialize()
doc = nlp("spaCy is a wonderful tool")
print(doc._.clf_trf_data)
# TransformerData(wordpieces=...
print(doc._.sentiment)
# 'positive'
print(doc._.sentiment_prob)
#{'prob': array([0.004, 0.028, 0.969], dtype=float32), 'labels': ['negative', 'neutral', 'positive']}
These pipelines can also easily be applied to multiple documents using the nlp.pipe as one would expect from a spaCy component:
docs = nlp.pipe(
[
"I hate wrapping my own models",
"Isn't there a tool for this?",
"spacy-wrap is great for wrapping models",
]
)
for doc in docs:
print(doc._.sentiment)
# 'negative'
# 'neutral'
# 'positive'
More Examples
It is always nice to have more than one example. Here is another one where we add the Hate speech model for Danish to a blank Danish pipeline:
import spacy
import spacy_wrap
nlp = spacy.blank("da")
config = {
"doc_extension_trf_data": "clf_trf_data", # document extention for the forward pass
"doc_extension_prediction": "hate_speech", # document extention for the prediction
"labels": ["Not hate Speech", "Hate speech"],
"model": {
"name": "DaNLP/da-bert-hatespeech-detection", # the model name or path of huggingface model
},
}
transformer = nlp.add_pipe("classification_transformer", config=config)
transformer.model.initialize()
doc = nlp("Senile gamle idiot") # old senile idiot
doc._.clf_trf_data
# TransformerData(wordpieces=...
doc._.hate_speech
# "Hate speech"
doc._.hate_speech_prob
# {'prob': array([0.013, 0.987], dtype=float32), 'labels': ['Not hate Speech', 'Hate speech']}
📖
Documentation
| Documentation | |
|---|---|
| |
Installation instructions for spacy-wrap. |
| |
New additions, changes and version history. |
| |
The reference for spacy-wrap's API. |
💬
Where to ask questions
| Type | |
|---|---|
| |
FAQ |
| |
GitHub Issue Tracker |
| |
GitHub Issue Tracker |
| |
GitHub Discussions |
| |
GitHub Discussions |
3 Feb 27, 2022
3.2k Dec 28, 2022
1.2k Jan 8, 2023
342 Nov 21, 2022
141 Dec 30, 2022
581 Dec 21, 2022
2k Jan 4, 2023
2.6k Jan 4, 2023
1.3k Jan 3, 2023
71 Jan 6, 2023
384 Dec 12, 2022
80 Jan 3, 2023