🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

My first paper at Apple is now public! Very insightful work led by Beomsu Kim on ReFlow. A series of pragmatic choices in the ReFlow formulation are introduced, leading to significant performance gains, all while still remaining true to the theoretical framework.

My paper written during internship at Apple MLR Paris is on arXiv now! It wouldn't have been possible without support from all my collaborators James Thornton, Yu-Guan Hsieh, Bahjat Kawar ✈️ CVPR 2025, Michal Klein, Marco Cuturi, and my advisor Jong Chul Ye! x.com/JamesTThorn/st…

We've been cooking up something special at Genmo. Get your hands on the Mochi 1 preview weights and share your creations. I'm blown away by the response from the community -- congratulations to the whole team!

Interested in diffusion/ flow models, optimal transport, uncertainty quantification or optimization? Our team at 🍎Apple ML Research Paris is recruiting 4-6 month PhD interns, we have multiple open positions See details below

We (Shuangfei Zhai @jsusskin) are looking for PhD interns to join Apple MLR late 2024 or early 2025 on generative modeling or multimodal learning. Location can be either NYC or Bay area. Apply here: jobs.apple.com/en-us/details/… or shoot me an email.

1/ We are excited to introduce Oasis, the world's first real-time AI world model, developed in collaboration with Etched. Imagine a video game entirely generated by AI, or a video you can interact with—constantly rendered at 20 fps, in real-time, with zero latency

Excited to introduce SketchAgent!👩‍🎨 We leverage the prior of pretrained multimodal LLMs for language-driven, sequential sketch generation and human-agent collaborative sketching! ✨ Try our fun interface here: github.com/yael-vinker/Sk…

So two pieces of good news: 1. code is now available github.com/apple/ml-vivid and 2. accepted to CVPR! 🥳

Great points. Enjoy ICLR!

I was unable to attend ICLR, but since received a lot of positive feedback on Beomsu Kim work on SimpleReflow, in particular: - position aware adaptive loss weighting - upweighting high frequency loss components - flow matching on large scale sharded sinkhorn couplings 🧵

Any creative solutions to stop the name tag badges from jingling? #CVPR2025