# explosion/spacy-transformers

🛸 Use pretrained transformers like BERT, XLNet and GPT-2 in spaCy

Repository: https://github.com/explosion/spacy-transformers
Canonical: https://ross.abutalabs.com/products/spacy-transformers
Homepage: https://spacy.io/usage/embeddings-transformers
Language: Python
License: MIT
License Family: permissive
Topics: spacy, spacy-pipeline, spacy-extension, nlp, natural-language-processing, natural-language-understanding, pytorch, bert, google, pytorch-model, openai, language-model, machine-learning, huggingface, transfer-learning, xlnet, gpt-2
Last push: 2026-03-27T08:49:39+00:00

## Health v2 (maintenance only)
Score: 71/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 74, release rhythm 51, longevity 100
- inputs: {"age_days": 2595, "days_push": 159, "days_rel": 169, "gap_med": 109, "n_releases_24m": 4}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1409, forks 181 (observed 2026-08-28T04:04:38.741121+00:00)

## What it is
A spaCy v3 extension package that provides pipeline components for using pretrained transformer models like BERT, RoBERTa, XLNet, and GPT-2 via Hugging Face's transformers library. It handles tokenization alignment, multi-task learning, and model packaging so transformer embeddings can power spaCy NLP components.

## Use cases
- use BERT embeddings in a spaCy NER pipeline
- fine-tune pretrained transformers for named entity recognition
- share one transformer model across multiple spaCy pipeline components
- train a custom spaCy pipeline with transformer features
- align Hugging Face tokenizer output with spaCy tokenization
- package and serialize a transformer-based NLP model

## When to choose
- you already use spaCy and want state-of-the-art transformer accuracy
- you need multi-task learning with a shared transformer backbone
- you want spaCy v3 config-driven training with pretrained models

## When to avoid
- you don't use spaCy and just want raw Hugging Face transformers
- you need lightweight CPU-only inference with minimal dependencies
- you're on spaCy v2.x without upgrading

## Facets
- artifact type: library
- maturity: stable
- function: nlp, machine-learning, transformers, llm-training
- domain: machine-learning, deep-learning
- platform: python, cross-platform
- tags: spacy, huggingface, bert, transfer-learning, pytorch, named-entity-recognition, multi-task-learning, natural-language-processing, gpu

## Member repositories
- explosion/spacy-transformers (main) score 71

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:38.741121+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:38:31.771114+00:00, confidence not recorded.
  - readme: https://github.com/explosion/spacy-transformers (fetched 2026-08-28T04:04:38.741121+00:00, sha cd3657c44e9c)
  - homepage: https://spacy.io/usage/embeddings-transformers (fetched 2026-08-29T11:52:07.925894+00:00, sha 244319b943a2)
  - site_page: https://spacy.io/usage/linguistic-features (fetched 2026-08-29T11:52:07.933730+00:00, sha 15f0372dd6a7)
  - site_page: https://spacy.io/usage (fetched 2026-08-29T11:52:07.929412+00:00, sha b80b5dfe308d)
  - site_page: https://spacy.io/usage/training (fetched 2026-08-29T11:52:07.940664+00:00, sha 5bb3e4f50b41)
  - site_page: https://spacy.io/api/architectures (fetched 2026-08-29T11:52:07.944389+00:00, sha 01d3f5214994)
- Data as of 2026-08-30T08:39:29.467469+00:00.