---
id: spacy-curated-transformers
version: "2.1.2"
license: MIT
license_treatment: permissive
maintenance: active
---

# spacy-curated-transformers — Curated transformer models for spaCy pipelines

License: permissive · Maintenance: active · Downloads: 1.5M/mo

## What it is and what it does

spacy-curated-transformers is a spaCy extension that wraps the curated-transformers library to bring transformer-based NLP models into spaCy pipelines. It provides integration with spaCy's architecture system, allowing you to use pretrained models like BERT, RoBERTa, and XLM-RoBERTa as components in your NLP workflows. The package emphasizes minimal dependencies and deployment-focused features such as model distillation and quantization.

You use it by installing it alongside spaCy, then configuring transformer models in your spaCy pipeline config to power tasks like named entity recognition or text classification. It supports multi-task learning and integrates with Hugging Face Hub for model discovery. The package is maintained by Explosion AI and handles the integration between spaCy and the curated-transformers library, managing serialization, configuration, and execution of transformer models in production pipelines.

Use it for:

- Build spaCy NLP pipelines powered by BERT or RoBERTa for named entity recognition
- Deploy transformer-based models in production with spaCy's serialization and config system
- Train multi-task learning models combining transformers with spaCy components
- Use XLM-RoBERTa for multilingual NLP tasks within a spaCy pipeline
- Experiment with model distillation and quantization for deployment optimization

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Integrates curated transformer models (ALBERT, BERT, CamemBERT, RoBERTa, XLM-RoBERTa) into spaCy pipelines via the curated-transformers library.

Yes, if you need transformer models in spaCy pipelines and want a maintained integration. The package is actively developed, has low install friction, and is MIT-licensed. It requires Python >=3.9 and substantial dependencies (torch, curated-transformers), so confirm those fit your environment. No known security vulnerabilities.

## Install

pip install spacy-curated-transformers
uv add spacy-curated-transformers
poetry add spacy-curated-transformers

## Installing spacy-curated-transformers

Before you install: Low install friction with a pure-Python wheel distribution. Actively maintained with last commit on 2026-03-27. Depends on torch and curated-transformers, which are substantial but standard ML dependencies.

License in practice: MIT license is permissive; you can use, modify, and distribute this package freely in commercial and private projects with minimal restrictions.

Quickstart:

pip install spacy-curated-transformers

import spacy
from spacy_curated_transformers import CuratedTransformer

nlp = spacy.load('en_core_web_sm')
# Configure transformer model in pipeline via spaCy config

Requires Python >=3.9 and torch installation; substantial disk space for transformer model weights.

Verify before relying:

- Whether multi-task learning and distillation/quantization features are production-ready or experimental
- Performance characteristics and memory footprint compared to alternatives
- Specific transformer model availability and loading from Hugging Face Hub

## Package facts

- License: MIT (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 1.5M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

spacy transformer models, bert integration spacy, nlp pipeline transformers, curated transformer architectures, spacy neural components, roberta spacy, transformer-based nlp, nlp, transformers, spacy-extension

[View on SkillFed](https://skillfed.io/packages/spacy-curated-transformers) · [View on PyPI](https://pypi.org/project/spacy-curated-transformers/)