Spacy Wrap: Integrate Hugging Face models in spaCy

Understand the project

Practical overview

When it fits

Use it when a fine-tuned Hugging Face transformer needs to participate in a spaCy workflow, especially when working more directly with transformer outputs is relevant.

Who it is for

Python developers building spaCy natural-language-processing workflows.
Teams seeking to use an existing fine-tuned Hugging Face model within a spaCy pipeline.

Documented uses

Add an existing transformer model to a spaCy pipeline.
Classify text, including sentiment or hate-speech examples.
Expose token-classification predictions or named entities through spaCy outputs.
Process multiple documents using spaCy's nlp.pipe.

Before you choose

Requirements & limits

The README gives a pip installation command and identifies Python 3.8 or newer. Repository metadata lists an MIT license.

The README demonstrates sequence classification, token classification and named-entity recognition.
The project processes multiple documents through spaCy's nlp.pipe interface.
Spacy Wrap aggregates tokens and converts results into spans from model logits; spacy-huggingface-pipelines instead wraps the Hugging Face pipeline and may produce different results.
The project recommends spacy-huggingface-pipelines for most use cases, while noting Spacy Wrap may suit workflows needing more direct transformer output.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Add a fine-tuned sentiment transformer to a spaCy pipeline

Install spaCy-wrap and configure the documented SST-2 sentiment model in a blank English spaCy pipeline.

Before you start

Python 3.8 or later, as indicated by the repository badge.
spaCy and spaCy-wrap installed in the Python environment.
A fine-tuned Hugging Face model; this example uses distilbert-base-uncased-finetuned-sst-2-english.

Steps

Install spaCy-wrap with the command below.
Documented command
```
pip install spacy_wrap
```
Import spacy and spacy_wrap, then create a blank English pipeline with nlp = spacy.blank("en").
Set the model name to distilbert-base-uncased-finetuned-sst-2-english and configure doc_extension_trf_data as clf_trf_data and doc_extension_prediction as sentiment.
Add the component with nlp.add_pipe("sequence_classification_transformer", config=config), then process text with doc = nlp("spaCy is a wonderful tool").
Read the output from doc.cats, doc._.sentiment, or doc._.clf_trf_data; the example shows positive sentiment.

Expected outcome

A spaCy pipeline configured to apply the fine-tuned sentiment model and expose its classification and transformer data on the processed document.

Guide limitations

The project notes that token aggregation and span conversion can produce differences because spaCy-wrap operates on model logits; it recommends spacy-huggingface-pipelines for most use cases and identifies spaCy-wrap as useful when more direct transformer output is needed.

Documentation 1 ↗

Evidence and transparency

Sources & review

Official repository ↗

Source record details

Source availability: Active
Sources checked: 16 October 2026
Original website title: KennethEnevoldsen/spacy-wrap · Official homepage title · retained in its source language

How we review sources Suggest a correction

Related options

spacy-huggingface-pipelines — The README identifies it as an alternative that wraps the Hugging Face pipeline rather than the transformer and recommends it for most use cases.

Spacy Wrap