What is this?
spaCy is a library for advanced Natural Language Processing in Python and Cython, designed from day one for real products. It ships pretrained pipelines and supports tokenization and training for 70+ languages.
spaCy is a library for advanced Natural Language Processing in Python and Cython, designed from day one for real products. It ships pretrained pipelines and supports tokenization and training for 70+ languages.Is it a library I add to my Python code, or a service I run?It's a Python library: you import it into your own code and build text-processing pipelines with it. There's no server or UI — everything runs inside your Python process.
It's a Python library: you import it into your own code and build text-processing pipelines with it. There's no server or UI — everything runs inside your Python process.How do I install it and get a first model running?Install with pip or from conda-forge, then fetch a pretrained pipeline with spaCy's download command — models install as Python packages, like any other module in your app.
Install with pip or from conda-forge, then fetch a pretrained pipeline with spaCy's download command — models install as Python packages, like any other module in your app.Will it run on my Python setup?You need 64-bit Python 3.7 to 3.12, on macOS, Linux or Windows (Cygwin, MinGW or Visual Studio).
You need 64-bit Python 3.7 to 3.12, on macOS, Linux or Windows (Cygwin, MinGW or Visual Studio).Do spaCy updates ever break my trained models?They can: some updates require new statistical models, and your training and runtime inputs must match. Run the validate command to check installed models after upgrading, and retrain your own models with the new version.
They can: some updates require new statistical models, and your training and runtime inputs must match. Run the validate command to check installed models after upgrading, and retrain your own models with the new version.Can I use it in commercial projects for free?Yes. spaCy is commercial open-source software under the MIT license, and its maintainer Explosion also sells paid NLP consulting around it.
Yes. spaCy is commercial open-source software under the MIT license, and its maintainer Explosion also sells paid NLP consulting around it.What can it actually do with text out of the box?Components cover named entity recognition, part-of-speech tagging, dependency parsing, sentence segmentation, text classification, lemmatization, morphological analysis and entity linking, plus built-in visualizers for syntax and NER.
Components cover named entity recognition, part-of-speech tagging, dependency parsing, sentence segmentation, text classification, lemmatization, morphological analysis and entity linking, plus built-in visualizers for syntax and NER.Do I have to train my own models, or are there pretrained ones?You don't have to: spaCy ships pretrained pipelines for different languages and tasks. For custom needs there's a production-ready training system, with multi-task learning on pretrained transformers like BERT.
You don't have to: spaCy ships pretrained pipelines for different languages and tasks. For custom needs there's a production-ready training system, with multi-task learning on pretrained transformers like BERT.Which languages does it work with?Over 70 languages are supported for tokenization and training, each with linguistically-motivated tokenization.
Over 70 languages are supported for tokenization and training, each with linguistically-motivated tokenization.Can I plug in my own PyTorch or TensorFlow models?Yes — spaCy supports custom models in PyTorch, TensorFlow and other frameworks, and pipelines can be easily extended with custom components and attributes.
Yes — spaCy supports custom models in PyTorch, TensorFlow and other frameworks, and pipelines can be easily extended with custom components and attributes.What are similar projects?HanLP, bert, flair, DeepLearning, presidio, keras, jeeves, Chinese-Word-Vectors
HanLP, bert, flair, DeepLearning, presidio, keras, jeeves, Chinese-Word-VectorsWhich topics is it in?cython, nlp-library, production, named-entity-recognition, text-classification, tokenization, text-processing, neural-networks, tensorflow, training, natural-language-processing, embeddings
cython, nlp-library, production, named-entity-recognition, text-classification, tokenization, text-processing, neural-networks, tensorflow, training, natural-language-processing, embeddings