Happening now! @pjox.bsky.social and I are giving a talk for @eleutherai.bsky.social on CommonLID, a community-driven web domain evaluation dataset for language identification. Join here: discord.gg/aYy3Se7Q?eve...
Paper: arxiv.org/abs/2601.18026
@commoncrawl.bsky.social
Love this widget by Daan van Esch: daanvanesch.nl/langid/index... - compare language ID predictions in your browser!
Announcing our latest paper: CommonLID
In collaboration with @commoncrawl.bsky.social @mlcommons.org @jhu.edu we built a LID benchmark on actual Common Crawl text covering 109 languages. Existing evaluations overestimate how well LangID works on web data.
arxiv.org/abs/2601.18026
"The true genius is a mind of large general powers, accidentally determined to some particular direction."
~ Samuel Johnson
Examples of mislabeled web text by existing LangID systems. A full text version is available on the blog post below.
Examples of mislabeled web text by existing LangID systems. A full text version is available on the blog post below.
Language identification still proves to be a challenging task, especially for web data. In collaboration with @mlcommons.org @eleutherai.bsky.social @jhu.edu and 97 community members, we created CommonLID, a new benchmark for LangID for 100+ languages!
I'm learning Rust right now and I'd recommend the experimental version of the Rust Book: rust-book.cs.brown.edu. The quizzes help you understand that you don't actually understand ownership 👍
I feel obliged to share this masterpiece from my brother:
Laurie Burchell at a lectern presenting her Turing Seminar talk
Laurie Burchell at a lectern, with a blackboard behind her, presenting her Turing Seminar talk
A huge thank you to @very-laurie.bsky.social for delivering a fantastic UoB Turing seminar. Her talk was entitled “Common Crawl: open web data for everybody.”
In this talk, she introduced the @commoncrawl.bsky.social and the data products they offer.
Turing Seminar event graphic with details of the event and headshots of Laurie Burchell and Thom Vaughan to the right
The Turing Liaison Team is excited to host @very-laurie.bsky.social and Thom Vaughan to introduce the @commoncrawl.bsky.social and the data products it offers.
📆 26 November
⏰ 13:00 - 14:00
📍C44 Biomedical building, University of Bristol
Find out more: tinyurl.com/mrxp5h2n
If you were able to join us, let us know about your experience: docs.google.com/forms/d/e/1F...
One of the biggest obstacles to improving language technologies for low-resource languages is the lack of data. To address this, we need better language identification tools. So, we're organizing a shared task on Language Identification for Web Data! #NLP #NLProc
I've been learning by writing the tutorials 💅
Next up: using CCF's host index commoncrawl.org/blog/introdu...
Person dressed as a knight with a trans flag and a cape with a trans flag shield that says “Trans Knights are human Knights”
This absolute icon at Canterbury Pride 🩷🤍🩵
A six panel cartoon. Panel one: Title: Learn to read with Physics Lesson One: two and three-letter words (image of a smiling book and particle) Panel two: A friendly scientist enters a colourful lab carryinh a coiffee and papers. Text: The Doc is in the lab. Panel three: She has put on eye protectors and started a machine Text: She put on the ion ray. Panel four: The machine has caused a small fire. She looks aghast. Text: Pop! The ray lit the gas. Panel five: She runs from the burning lab Text: The lab is too hot! Panel six: She sits on the grass. In the background the ruins of the lab smoulder. Text: The Doc is sad.
My cartoon for this week’s @newscientist.com
✨Call for papers!✨ @commoncrawl.bsky.social and friends are organising the 1st Workshop on Multilingual Data Quality Signals, held in tandem with @colmweb.org. Submit your research on multilingual data quality!
Submission deadline is 23 June, more info: wmdqs.org
I'm starting as a Senior Research Engineer with the Common Crawl Foundation today! 😎
#Gaelic
#Hobbit
#Ghàidhlig
EEEEEEEEEE!
www.gaelicbooks.org/explore-the-...
A comic called "how to draw goose neck". In panel 1 the goose's neck is very short and is labeled "no". In panel 3 the goose's neck is the right size and is labeled "yes". In panel 3 the goose's neck is longer and is labeled "no". In panel 4 the goose's neck is extremely long and terrifyingly snakelike and it's labeled "yessss".
A comic called The Canada Goose: A Role Model For Our Time. In panel 1, the goose is labeled "assertive" and is hissing furiously. In panel 2 it's "brave" and is standing up to a fearsome swan. In panel 3 it's "good parent" and is standing up to a dog while protecting its goslings. In panel 3 it's "team player" and is flying in a beautiful V. In panel 5 there's a close up of its head and it's labeled "crisp, modern aesthetic". In panel 6 it's eating grass and pooping and labeled "high fiber diet".
There's a lot of talk about Canada Geese and whether they're good and my answer is Yes.
omg who is she
I'm seeing way too much paper-white skin exposed to the cold over the last few days, Scotland is built different
I'm part of this! There's also a paper: arxiv.org/abs/2503.10267
(she's very proud)
I told my mum that my model was in the top ten most downloaded models on HF last month, big "are ya winning son?" energy
how are there 5.7M downloads of a model I didn't advertise, I am suspicious
a screenshot of the top-ten most downloaded models on Hugging Face
I'm nerd famous
What's the title?
Jesus. Twister in Donegal right now.
My replacement PIR finally arrived! Hoping it doesn't short this time 🙏
