Production houses own vast video archives potentially worth millions. The problem? Finding the right moment quickly from 15 seasons of content is humanly impossible.

In Ad World News, our Head of Science @yannistevissen.fr breaks down how AI video understanding is solving this.
tinyurl.com/3kdabn4f

I created #UpToCure, a go-to website to read about recent medical research aiming to cure rare diseases. Powered by an open-source deep research agent
Please check it out, feedbacks are welcomed.
uptocure.com

We’re proud to introduce you to the newest version of our patented AI model: MXT-2. Designed to empower organizations to manage, search for, repurpose and monetize large video libraries even more effectively. ✨

Learn more about it from our team 👉 www.youtube.com/watch?v=7qDu...

We're recruiting a machine learning intern to work on a cool video understanding project. Feel free to reach out if you are interested or apply here: careers.momentslab.com/jobs/5331848...

NVILA: Efficient Frontier Visual Language Models

abs: arxiv.org/abs/2412.04468

NVIDIA introduces NVILA, a family of open VLMs designed to optimize both efficiency and accuracy.

I was honored to speak at the OECD on Monday about AI and Disability. There’s a lot to do to make the AI models and community more inclusive

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!

Unfair AI goes beyond data bias!
An example in France: fraud detection on social benefits naturally ends up targeting poor people, as they benefit more from social benefits.
www.laquadrature.net/caf-notation...

Fraud detection on wealth tax would naturally target the rich 1/

Making SmolLM2 more reproducible: open-sourcing our training & evaluation toolkit 🛠️ github.com/huggingface/...

Pre-training & evaluation code, synthetic data generation pipelines, post-training scripts, on-device tools & demos

Apache 2.0. V2 data mix coming soon!

Which tools should we add next?