spaCy

@spacy_io

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

Berlin, Germany
spacy.io
Joined August 2015
127 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
    Oct 15

    IT'S HERE! Today we're releasing spaCy nightly, the first candidate for the upcoming v3.0. 🛸 Transformer-based pipelines for SOTA models ⚙️ New training & config system 🧬 Models using any framework 🪐 Manage end-to-end workflows 🔥 New & improved APIs

    10 replies 175 retweets 558 likes
    Show this thread
    Show this thread
    Undo
  2. Retweeted
    Oct 22

    We restricted our analysis to women no longer working at and to years 1846-1999. We used a combination of natural language processing in and shell scripting. Lots of manual data curation too! 3/11

    1 reply 1 retweet 4 likes
    Show this thread
    Show this thread
    Undo
  3. Retweeted
    Oct 22

    Enjoyed the talk by at about Hostile Narrative Analysis with custom components. Interesting to think about qualitative approaches when targets are poorly- (or completely un-!) defined. Thanks + for forwarding invite!

    2 replies 3 retweets 4 likes
    Show this thread
    Show this thread
    Undo
  4. Retweeted
    Oct 21

    Looking forward to speaking at the NLP edition tomorrow. It's my first talk on v3 and I'll walk you through some of the new features & the ideas behind them! 📆 October 22, 12:00 PST/15:00 EDT/21:00 CEST 🎟️ Info & registration:

    2 replies 26 retweets 110 likes
    Show this thread
    Show this thread
    Undo
  5. Retweeted
    Oct 19

    Today we're taking v3 for a spin in a project on grammar correction. ✍️ Loving the config-based training and the integration with . 😍 Also, these look pretty promising. 🧐 Are they a thing? 😏

    2 replies 14 retweets 49 likes
    Undo
  6. Retweeted
    Oct 19

    恥ずかしながらspaCy使ったことなかったので、このサイトでさっと勉強したんですが、めちゃくちゃ感動した。 ・spaCy自体の使いやすさ ・学習サイトのUIのすばらしさ こちらの方（ ）が制作されているようです。本当に感謝です。

    4 retweets 11 likes
    Undo
  7. Retweeted
    Oct 17

    Our principal scientist just released a four-part video tutorial (with code) on using and .

    10 retweets 28 likes
    Undo
  8. Retweeted
    Oct 16

    🎉🎉 Scispacy v0.3.0 is here! 🎉🎉 Some bug fixes, and also a new Custom Component: Hyponym Detection via Hearst Patterns! Shout out to for contributing the spaCy Matcher version of the Hearst pattern rules.

    2 replies 13 retweets 57 likes
    Show this thread
    Show this thread
    Undo
  9. Retweeted
    Oct 16

    One of the most significant changes in v3 is the new config system for defining & training pipelines. Validation is crucial here, but how does it work? A key part is Pydantic by : Here's how it looks under the hood (thread) 👇

    2 replies 23 retweets 131 likes
    Show this thread
    Show this thread
    Undo
  10. Retweeted
    Oct 16
    Replying to

    The spaCy v3 version of this tutorial on Entity Linking can be found here: . Basically, the 'project.yml' file describes the full workflow and lets you execute and replicate the results from the video :-)

    5 retweets 17 likes
    Undo
  11. Oct 15

    Wow, this was quick! The first user project template: detecting Star Trek aliens and planets 🖖👽 1️⃣ Clone template $ spacy project clone ner_tng --repo --branch main 2️⃣ Preprocess data, train & evaluate $ spacy project run all 3️⃣ Package & use model 🚀

    10 retweets 18 likes
    Undo
  12. Retweeted
    Oct 15

    just announced the spaCy 3 nightly and I'm training my Star Trek TNG example NER with the new project CLI and training configuration. Works super smooth so far.

    6 replies 6 retweets 32 likes
    Show this thread
    Show this thread
    Undo
  13. Retweeted
    Oct 15

    🙌🙌 With the v3.0 release, you can use Ray to train on one or more remote machines, potentially speeding up your training process.

    IT'S HERE! Today we're releasing spaCy nightly, the first candidate for the upcoming v3.0. 🛸 Transformer-based pipelines for SOTA models ⚙️ New training & config system 🧬 Models using any framework 🪐 Manage end-to-end workflows 🔥 New & improved APIs
    Show this thread
    10 retweets 22 likes
    Undo
  14. Retweeted
    Oct 15

    This release was SO much work! Glad to finally have it out 🍻 The big impact for users will definitely be the transformer models and config system, but I want to talk about a feature I wasn't expecting to build until a few months ago: the new workflow system, spaCy Projects 🧵

    IT'S HERE! Today we're releasing spaCy nightly, the first candidate for the upcoming v3.0. 🛸 Transformer-based pipelines for SOTA models ⚙️ New training & config system 🧬 Models using any framework 🪐 Manage end-to-end workflows 🔥 New & improved APIs
    Show this thread
    4 replies 30 retweets 218 likes
    Show this thread
    Show this thread
    Undo
  15. Retweeted
    Oct 15

    If you want to play with the new (or your own custom trained) v3 models, check out this app I built, hosted with the new sharing – also launched today, so perfect timing! 🥳 🖼️ App: 👑 spacy-streamlit:

    5 replies 53 retweets 244 likes
    Show this thread
    Show this thread
    Undo
  16. Retweeted
    Oct 15

    Me when sharing and 3.0 nightly get announced on the same day.

    3 replies 9 retweets 52 likes
    Undo
  17. Oct 15

    Project templates let you manage and share end-to-end spaCy workflows and orchestrate training, packaging and serving your pipelines. Start by cloning a template, adjust it, add your data and go! 📖 Docs: 🪐 Templates:

    2 replies 1 retweet 23 likes
    Show this thread
    Show this thread
    Undo
  18. Oct 15

    spaCy v3.0's training system also makes it easy to integrate with other tools – like for experiment tracking! The built-in logger tracks your config (and registered functions!) and lets you discover patterns & correlations. 📖 Details:

    1 reply 8 retweets 27 likes
    Show this thread
    Show this thread
    Undo
  19. Oct 15

    To help you get started, we've added a quickstart widget and CLI for auto-generating a training config for your specific use case. Language? CPU or GPU? Optimize for efficiency or accuracy? 📖 Try it here:

    1 reply 14 likes
    Show this thread
    Show this thread
    Undo
  20. Oct 15

    The release includes retrained model families, 50+ pipelines in total with 5 new transformer-based models. You can also train your own using custom data and transformer weights of your choice. 📖 What's new: 📖 Benchmarks:

    1 reply 3 retweets 26 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

    ·