Release Sparv Pipeline v4.1.0 · spraakbanken/sparv-pipeline

Note: Sparv v4.1.0 was released on PyPI on April 14, 2021 but we forgot to create a GitHub release then... 🤦🙈

New preload functionality for preloading annotators to speed up annotation process.
Added verbose mode for progress bar, showing all concurrently running tasks (by using the -v flag, mainly usable
together with the -j flag for multiprocessing).
Ability to limit the number of parallel processes used by specific annotators.
Source document names are now shown in error messages.
Added exporter configuration to wizard.
Added several new configuration options for Stanza, and helpful error message to help mitigate memory problems.
An error message is now displayed when attempting to run annotations without input files.
Added new command (languages) to show a list of supported languages.
You can now refer to models outside the Sparv data dir.
Added importers section to sparv run-rule --list.
Class values inferred from annotation usage is now shown when running sparv classes.
Dry-running (sparv run -n) now shows a summary of tasks.

Improved progress bar. Shows number of tasks completed and left, instead of estimated time (which wasn't very
helpful).
Slightly quicker startup time.
Malt and Stanza no longer perform dependency parsing on tokens not belonging to any sentences.
The build-models command no longer builds all models by default unless the --all flag is used.
Regular annotators used as custom_annotations are now configured using config instead of params.
Updated and improved documentation.

Provide feedback