spaCy

@spacy_io

Open-source library for industrial-strength Natural Language Processing in Python. Developed by 💥 📖 📘 📺

Berlin, Germany
spacy.io
Joined August 2015
118 Photos and videos Photos and videos

Tweets

You blocked @spacy_io

Are you sure you want to view these Tweets? Viewing Tweets won't unblock @spacy_io

  1. Pinned Tweet
    2 Aug 2019

    Say hello to spacy-pytorch-transformers! 🛸 BERT, XLNet & GPT-2 in your spaCy pipeline 🤗 Based on 's pytorch-transformers 🎚️ Fine-tune pretrained models on your task 📦 Model packages for English & German 🚀 Token alignment, similarity & more

    18 replies 444 retweets 1,234 likes
    Show this thread
    Show this thread
    Undo
  2. Retweeted
    Apr 14

    Want to find out what your word embeddings are truly capturing? Have a look at the newest addition to our universe: . This library by supports spaCy models and sense2vec, and has some cool interactive visualisations! 🤓

    1 reply 16 retweets 88 likes
    Undo
  3. Retweeted
    Apr 8

    I really loved this blog post! It's a good example of the type of linguistic trade-offs NLP systems are often making. Word relationships are DAGs, but often you don't lose much information if you use a tree instead. And sometimes you can even get away with flat spans. 1/n

    Great post by on creating labelled data for legal NLP, the subtle decisions you have to make and why it's important to reason about your data. Featuring and & 's work on numeric fused heads! 📖
    Show this thread
    1 reply 11 retweets 73 likes
    Show this thread
    Show this thread
    Undo
  4. Retweeted
    Apr 7

    Feature Highlights 1. Python labeling function library for core clinical text tasks (negation, temporality, many more) 2. Tutorials for sentence & relation classification 3. pipeline for NLP preprocessing with EHR-specific tokenization and sentence boundary detection

    1 reply 1 retweet 2 likes
    Show this thread
    Show this thread
    Undo
  5. Retweeted
    Apr 7

    Blackstone v2, the legal library from is coming on very nicely, particularly now has teamed up with me to get the new version ready. The original NER has an F1 of ~75%. The new model is now performing at ~89%, which is a really decent boost.

    Last summer, ’s nascent R&D division, , released its first piece of software in the form of a Python library called Blackstone (), a library for doing on judgments and other narrative-style legal texts (skeletons, journals)
    Show this thread
    1 reply 5 retweets 29 likes
    Undo
  6. Retweeted
    Apr 1

    🎙️🐍I talked to Tobias on shortly after was released. It was awesome to catch up again! We spoke about , & how spaCy has been changing to take advantage of new developments in NLP.

    1 reply 13 retweets 50 likes
    Undo
  7. Retweeted
    Mar 23

    📢📢📢 Interrupting your regular scheduled Covid-19 programming to bring you a blog post about the `DependencyMatcher`, a little known but powerful feature! Code in Google Colab to follow along, too!

    2 replies 10 retweets 47 likes
    Undo
  8. Retweeted
    Mar 22

    Using stanza, spacy and pytextrank to do NER in Russian and Keyword extraction in Swedish. Thanks to

    1 reply 9 retweets 41 likes
    Undo
  9. Retweeted
    Mar 21

    Day 8 Completed ch 1 exercises of 's "Advanced NLP with spaCy" course.

    41 retweets 64 likes
    Undo
  10. Retweeted
    Mar 19

    New blog post! Looking at this weeks 🧹 Using with {spacyr} to extract part-of-speech for use in and {textrecipes} 📚 General text mining with {tidytext} and hyperparameter tuning with {tune} ✨

    1 reply 21 retweets 64 likes
    Undo
  11. Mar 17

    Cool project for interactive NLP with built-in support for text classification using spaCy and spacy-transformers. 📄 Repo: 📺 While you're at it, check out 's amazing talk on applied NLP:

    Looking to do text classification? Working with text data? You're in luck, gobbli 0.1.0 was just released! We've added an awesome interactive app to help you explore data + evaluate models. Take a look at the medium post for more details.
    12 retweets 49 likes
    Undo
  12. Retweeted
    Mar 17

    Always amazed by how responsive the spaCy team is! Great job by on integrating Stanza with spaCy, within... 2 hours since the release! 👍 Note that you can also use spaCy for fast tokenization in Stanza now. As simple as `tokenize="spacy"` 😀

    Now also available for your spaCy pipeline: spacy-stanza (formerly spacy-stanfordnlp) integrates Stanza by – with support for more languages and NER 🚀
    2 replies 5 retweets 34 likes
    Undo
  13. Mar 17

    Now also available for your spaCy pipeline: spacy-stanza (formerly spacy-stanfordnlp) integrates Stanza by – with support for more languages and NER 🚀

    4 replies 66 retweets 207 likes
    Undo
  14. Retweeted
    Mar 16

    Btw, I've reorganized our YouTube playlists so it's easier to find stuff. If you feel like watching some NLP content, check it out 👀 We have all talks from , end-to-end tutorials, 's spaCy series & talks from various conferences. 📺

    23 retweets 126 likes
    Undo
  15. Mar 16

    ✂️ spacy-syllables: Easy-to-use pipeline component for adding multilingual syllable annotations to tokens, powered by pyphen with support for lots of languages!

    10 retweets 52 likes
    Undo
  16. Retweeted
    Mar 16

    ✨📺 NEW VIDEO! 📺✨ I'm using & to train an NER model from scratch with semi-automatic annotation & transfer learning. It's about food trends & we're also building a cool bar chart race visualization! Watch it here:

    7 replies 57 retweets 317 likes
    Show this thread
    Show this thread
    Undo
  17. Retweeted
    Mar 11

    libpython-clj is great. You can use any Python library in Clojure! (spaCy, Tensorflow, PyTorch and that's just a few.)

    28 retweets 75 likes
    Undo
  18. Mar 12

    Out now: spaCy v2.2.4! 💬 Alpha tokenization support for Yoruba and Basque 📚 Various updates to existing language data 🐛 Lots of bug fixes and UX improvements

    18 retweets 71 likes
    Undo
  19. Retweeted
    Mar 10

    Unsurprisingly, it's not going to be in July. We'll see how things unfold – we look forward to hosting you all in Berlin when we can 🤞

    Yes, there'll definitely be a again this year 🏜 We're still working on confirming the venue, date and speakers. The preliminary date we're looking at is Saturday, June 27 (not official yet, though).
    1 reply 5 retweets 37 likes
    Undo
  20. Retweeted
    Mar 6

    Med7 - + - based clinical NER model to recognise medications and related entities. GitHub: … Medium: Thanks for tools and for inspiration from scispaCy!

    7 replies 35 retweets 113 likes
    Undo
  21. Retweeted
    Mar 5

    Quick update on the BlackstoneNLP library we released through last summer: 1. The model is currently being retrained to support v.2.2+. It breaks at north of 2.1.8 at the moment.

    1 reply 4 retweets 13 likes
    Show this thread
    Show this thread
    Undo

@spacy_io hasn't Tweeted yet.

Loading seems to be taking a while.

Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

    You may also like

    ·