Daniel Mendelevitch

Hi!👋 I'm a research scientist at Luma. My work at Luma focuses on efficient image/video tokenization. I recently finished my undergrad UCLA studying math and statistics. I'm super interested in representation learning, especially for video models.

Previously, I was a research engineer at Stability AI, where I worked on Stable Video Diffusion.

Email  /  Scholar  /  Twitter  /  Github

profile photo

Research
Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets
Andreas Blattmann*, Tim Dockhorn*, Sumith Kulal*, Daniel Mendelevitch, Maciej Kilian, Dominik Lorenz, Yam Levi, Zion English, Vikram Voleti, Adam Letts, Varun Jampani, Robin Rombach
arXiv, 2023
(weights) / (paper)

A generative text-to-video and image-to-video model scaled to a large pre-training dataset. We explore the impact of different data-filtering and synthetic captioning techniques to train powerful foundation video models.

Projects
Video2dataset
Maciej Kilian, Romain Beaumont, Daniel Mendelevitch, Sumith Kulal, Andreas Blattmann (blogpost)

A tool to easily create large-scale video datasets from urls.

Code borrowed from Jon Barron 😅