Happy holidays everyone! What would you want to see more from Hugging Face in 2025?

People are flexing their end of year stats, so I made this app to show @hf.co hub stats in a tidy design! Thanks @jfcalvo.hf.co and @ameeelie.bsky.social for the feature!

Just 10 days after o1's public debut, we’re thrilled to unveil the open-source version of the technique behind its success: scaling test-time compute By giving models more "time to think," Llama 1B outperforms Llama 8B in math—beating a model 8x its size. The full recipe is open-source!

Who’s at #neurips2024 and want to meet HF team members?

Announcing 🥂 FineWeb2: A sparkling update with 1000s of 🗣️languages. We applied the same data-driven approach that led to SOTA English performance in🍷 FineWeb to thousands of languages. 🥂 FineWeb2 has 8TB of compressed text data and outperforms other datasets.

The FineWeb team is happy to finally release "FineWeb2" 🥂🥳 FineWeb 2 extends the data driven approach to pre-training dataset design that was introduced in FineWeb 1 to now covers 1893 languages/scripts Details: huggingface.co/datasets/Hug... A detailed open-science tech report is coming soon

The folks at Foursquare released a @hf.co dataset of 104.5 million places of interest and here's all of them plotted using datashader

I weirdly love this! #5 trending on HF right now

Excited to see more biology open-source models for real positive use-cases of AI! Chai does structure predictions at AlphaFold3 levels of accuracy and able to handle multi-peptide or peptide-ligand complexes rather than just single chains. Apache 2.0 on HF huggingface.co/chaidiscover...

🤖 6 AI Predictions for 2025 by Hugging Face CEO #HuggingFace #ClementDelangue #ArtificialIntelligence www.cosmico.org/6-ai-predict...

Let's make AI more inclusive. At @huggingface.bsky.social we'll launch a huge community sprint soon to build high-quality training datasets for many languages. We're looking for Language Leads to help with outreach. Find your language and nominate yourself: forms.gle/iAJVauUQ3FN8...

Six predictions for AI in 2025 (and a review of how my 2024 predictions turned out):

I would put this even more strongly: open source AI is probably our only realistic chance to avoid a terrifying increase in concentration of power. I do not want to live in a world where the people with all the money also have all the intellectual power.

Good list if you want to understand AI! go.bsky.app/Nik64nt

QwQ is #1 trending on @hf.co!

This demo of structured data extraction running on an LLM that executes entirely in the browser (Chrome only for the moment since it uses WebGPU) is amazing My notes here: simonwillison.net/2024/Nov/29/...

Wasn't going to wade into this but... as long as the internet exists in an open form, scraping will exist, and that's kind of the whole point... APIs exist because interacting with data is valuable and scraping (the old fashioned way) is annoying.

For me, the biggest risk in AI is centralization of power and benefits in the hands of a few. This is why at @hf.co, we've been investing on building online courses for as many people as possible to understand and learn to build AI themselves. You can find some of them here: huggingface.co/learn

The most realistic reason to be pro open source AI is to reduce concentration of power.

I’m thankful for everyone using Bluesky, everyone building on atproto, everyone listening to our message and sharing our dream of a better social media ecosystem that puts people first. We’re going to do this together — thanks for joining us on this journey!

A dataset of 1 million or 2 million Bluesky posts is completely irrelevant to training large language models. The primary usecase for the datasets that people are losing their shit over isn't ChatGPT, it's social science research and developing systems that improve Bluesky.

Did you know that 99% of email today is spam? Your inbox isn’t 99% spam because AI is used to filter it. The same 99% will happen here too, but if AI researchers continue to get perma-banned for making available the datasets needed to filter it, it’s going to make this platform unusable.

new post: bluesky vs. huggingface for normal people jessbpeck.com/posts/bluesk... anytime i spend less than 5 months on a thing you know i have opinions. anyway, as always, feel free to argue with me/complain at me/ point out errors.

Hi, so I've spent the past almost-decade studying research uses of public social media data, like e.g. ML researchers using content from Twitter, Reddit, and Mastodon. Anyway, buckle up this is about to be a VERY long thread with lots of thoughts and links to papers. 🧵

Hugging Face just released SmolVLM, a powerful new AI model that could seriously cut costs and boost efficiency for businesses! 🤯 🔗 Read more: venturebeat.com/ai/hugging-f...

Good move by @danielvanstrien.bsky.social to apologize & remove the dataset IMO. Let's find ways to fight concentration of power in the hands of proprietary/commercial actors (who are certainly using all the public data they can find w/o telling you) while also fostering consent & attribution.

The 🐐 Yann LeCun just uploaded MNIST on Hugging Face!

Let's go! We are releasing SmolVLM, a smol 2B VLM built for on-device inference that outperforms all models at similar GPU RAM usage and tokens throughputs. SmolVLM can be fine-tuned on a Google collab and be run on a laptop! Or process millions of documents with a consumer GPU!

First dataset for the new @huggingface.bsky.social @bsky.app community organisation: one-million-bluesky-posts 🦋 📊 1M public posts from Bluesky's firehose API 🔍 Includes text, metadata, and language predictions 🔬 Perfect to experiment with using ML for Bluesky 🤗 huggingface.co/datasets/blu...

I've been in Brazil for 10 days now 🇧🇷🇧🇷🇧🇷 I've been surprised by the gap between the massive number of people interested in AI (chatgpt adoption is crazy here) and the relatively low number of real AI builders - aka people and companies building their own AI models, datasets and apps.

🌍 Major-TOM North America embeddings are now open source! 🚀 Search satellite imagery with natural language. Empower ML and boost geospatial innovation. Use cases: online learning, agentic RAG, edge cases. Join us on HuggingFace! #OpenSource #Geospatial #earthobservation huggingface.co/quasara-io

Who else belongs in this Women in AI starter pack?! go.bsky.app/847WJHE

I am very excited to launch a new community initiative next week. Let's build the largest open community dataset to evaluate and improve image generation models. Follow: huggingface.co/data-is-bett... And stay tuned here

The Hugging Face Folks list keeps growing. Check it out! go.bsky.app/2VGyqGt

It's Sunday morning so taking a minute for a nerdy thread (on math, tokenizers and LLMs) of the work of our intern Garreth By adding a few lines of code to the base Llama 3 tokenizer, he got a free boost in arithmetic performance 😮 [thread]

Excellent podcast by Acquired with @clemdelangue.bsky.social the CEO of Hugging Face. This is must listen for anyone interested in AI. #ai open.spotify.com/episode/3baV...

We created these values as a team, and we live them as a team 💪

Should HF do more agent stuff? If so, what would be useful?

I started a list of journalists covering AI to follow on Bsky! It’s a WIP, so lmk who to add! go.bsky.app/NJ3xtGM

The AGI bubble is bursting. Thanks to @bloomberg.com @rachelmetz.bsky.social @shiringhaffary.bsky.social @dinabass.bsky.social and Julia Love for their coverage on this pivotal moment in time. www.bloomberg.com/news/article...

hey friends i have a new custom handle in here! 👀

Add your BSKY 🦋 to your @huggingface.bsky.social profile!

What's the secret sauce of SmolLM2 to beat LLM titans like Llama3.2 and Qwen2.5? Unsurprisingly: data, data, data! The SmolTalk is open and available here: huggingface.co/datasets/Hug...

SmolTalk is out 🗣️ Over 1M high quality instructions used for training SmolLM2, one of the best small language models in the industry. huggingface.co/datasets/Hug...

Introducing Transformers.js v3: State-of-the-art Machine Learning for the Web. New features include: ⚡ WebGPU support (up to 100x faster than WASM) 🔢 New quantization formats 🏛 121 supported architectures in total 🤖 Over 1200 pre-converted models Get started with `npm i @huggingface/transformers`

When XetHub joined Hugging Face, we brainstormed how to share our tech with the community. The magic? Versioning chunks, not files, giving rise to: 🧠 Smarter storage ⏩ Faster uploads 🚀 Efficient downloads Curious? Read the blog and let us know how it could help your workflows!

We need more open-source models, datasets & apps in AI healthcare. It makes it not only more likely to benefit everyone but also much safer, controllable and respects privacy! Great initiative so that everyone can build their own open-source AI health tracker for $12: huggingface.co/blog/cyrilza...

🎉 Announcing our seed fundraise! We've raised a $5M seed round led by Anthony Goldbloom and Shyam Mani at AIX Ventures to continue building marimo and other Python data tools, with participation from Jeff Dean, @clemdelangue.bsky.social, Wes McKinney, and others. marimo.io/blog/seed-an...