spaCy v3.7 adds support for Python 3.12, introduces the new standalone library Weasel for project workflows, and updates the transformer-based trained pipelines to use our new Curated Transformers library.

This release drops support for Python 3.6.

Weasel

The spaCy projects functionality has been moved into a new standalone library Weasel. This brings minor changes to spaCy-specific settings in spaCy projects (see upgrading below), but also makes it possible to use the same workflow functionality outside of spaCy.

All spacy project commands should run as before, just now they’re using Weasel under the hood.

Registered vectors

You can specify a custom registered vectors class under [nlp.vectors] in order to use static vectors in formats other than the ones supported by Vectors. To implement your custom vectors, extend the abstract class BaseVectors. See an example using BPEmb subword embeddings.

Additional features and improvements

• Add support for Python 3.12.
• Extend to Thinc v8.2.
• Extend transformers extra to spacy-transformers v1.3.
• Add --spans-key option for CLI evaluation with spacy benchmark accuracy.
• Load the CLI module lazily for spacy.info.
• Add type stubs for for spacy.training.example.
• Warn for unsupported pattern keys in dependency matcher.
• Language.replace_listeners: Pass the replaced listener and the tok2vec pipe to the callback in order to support spacy-curated-transformers.
• Always use tqdm with disable=None in order to disable output in non-interactive environments.
• Language updates:
  • Add left and right pointing angle brackets as punctuation to ancient Greek.
  • Update example sentences for Turkish.
• Package setup updates:
  • Update NumPy build constraints for NumPy 1.25+. For Python 3.9+, it is no longer necessary to set build constraints while building binary wheels.
  • Refactor Cython profiling in order to disable profiling for Python 3.12 in the package setup, since Cython does not currently support profiling for Python 3.12.

Trained pipelines

Pipeline updates

The transformer-based trf pipelines have been updated to use our new Curated Transformers library using the Thinc model wrappers and pipeline component from spaCy Curated Transformers.

Notes about upgrading from v3.6

This release drops support for Python 3.6, drops mypy checks for Python 3.7 and removes the ray extra. In addition there are several minor changes for spaCy projects described in the following section.

Backwards incompatibilities for spaCy Projects

spacy project has a few backwards incompatibilities due to the transition to the standalone library Weasel, which is not as tightly coupled to spaCy. Weasel produces warnings when it detects older spaCy-specific settings in your environment or project config.

• Support for the spacy_version configuration key has been dropped.
• Support for the check_requirements configuration key has been dropped due to the deprecation of pkg_resources.
• The SPACY_CONFIG_OVERRIDES environment variable is no longer checked. You can set configuration overrides using WEASEL_CONFIG_OVERRIDES.
• Support for SPACY_PROJECT_USE_GIT_VERSION environment variable has been dropped.
• Error codes are now Weasel-specific and do not follow spaCy error codes.

Pipeline package version compatibility

When you’re loading a pipeline package trained with an earlier version of spaCy v3, you will see a warning telling you that the pipeline may be incompatible. This doesn’t necessarily have to be true, but we recommend running your pipelines against your test suite or evaluation data to make sure there are no unexpected results.

If you’re using one of the trained pipelines we provide, you should run spacy download to update to the latest version. To see an overview of all installed packages and their compatibility, you can run spacy validate.

If you’ve trained your own custom pipeline and you’ve confirmed that it’s still working as expected, you can update the spaCy version requirements in the meta.json:

Updating v3.6 configs

To update a config from spaCy v3.6 with the new v3.7 settings, run init fill-config:

In many cases (spacy train, spacy.load), the new defaults will be filled in automatically, but you’ll need to fill in the new settings to run debug config and debug data.