just in time for the holidays 🎄🎁 RF inversion demo is here ✨ edit & stylize with Flux 🖌️ super cool work from Litu Rout , Nataniel Ruiz, now open w/ 🧨diffusers support thanks to an awesome community effort 🙌🏻

Thanks Linoy Tsaban for making the demo.

🚀 🇬🇷 A year in the making! I’ve just completed a set of 21 lectures in Machine Learning, in Greek, designed for high school students. The course introduces key ML concepts, coding in Python & PyTorch, and real-world AI applications. #MachineLearning #AI #EdTech #Greece

Announcing OpenThinker-32B: the best open-data reasoning model distilled from DeepSeek-R1. Our results show that large, carefully curated datasets with verified R1 annotations produce SoTA reasoning models. Our 32B model outperforms all 32B models including

I still see papers using DINO similarity to quantify “stylistically” similar two images are!! 🤦‍♀️🤦‍♀️ Guys, use CSD!! It is significantly better than DINO! Hugging Face Thomas Wolf - please enable CSD similarity as a metric on your platform! Many other papers validated its

Realtime interactive generative models FTW! Announcing a new 🌊 of details and features for Magenta RealTime, the open weights live music AI model from GDM! * Live Jamming with audio input 🎤🎸🎵 * Personalize your own models 🔧 * Tech report 📜 Links below in the 🧵...

Continuous diffusion had a good run—now it’s time for Discrete diffusion! Introducing Anchored Posterior Sampling (APS) APS outperforms discrete and continuous baselines in terms of performance & scaling on inverse problems, stylization, and text-guided editing.

Makes sense that unmasking each token by confidence, independently, does worse here

Really excited about these results. We’re using discrete diffusion for inverse problems. The key to it all is anchoring — an idea from our recent work on discrete diffusion for language modeling.

Awesome work from Litu Rout! Discrete diffusion is an exciting alternative to continuous diffusion when dealing with discrete-valued data. Litu’s work enables training-free discrete posterior sampling, applicable to tasks like inpainting, deblurring, super resolution, etc.

Quite interesting work by Litu Rout and team.

We found a new way to get language models to reason. 🤯 No RL, no training, no verifiers, no prompting. ❌ With better sampling, base models can achieve single-shot reasoning on par with (or better than!) GRPO while avoiding its characteristic loss in generation diversity.