Julius Berner

@julberner

Research scientist @nvidia | postdoc @caltech | PhD @univienna | former research intern @MetaAI and @nvidia | views are my own

jberner.info

Joined September 2018

Replies Reposts Media

Pinned
Julius Berner
@julberner
Aug 13
My talk on accelerated diffusion and flow models at @BerkeleyLab's DL4Sci School is online! Check it out for an overview on how to: - Reduce the number of denoising steps - Accelerate each step 📺 youtube.com/watch?v=ZnRzw9… (jump to 22:18 to learn more about the video below)
00:00
3
Julius Berner
@julberner
Sep 3
Excited to collaborate with @RekaAILabs on real-time steerable video generation using FastGen! 🎬 Check out our FastGen repo - we will also drop new models and methods soon: github.com/NVlabs/fastgen
Reka
@RekaAILabs
Sep 3
Together with @nvidia , we've built a real-time version of our video generation model: 30B parameters, steerable mid-stream in natural language, 720p at 24fps. 11.8x faster on a single H100, no major quality drop in our blind eval. We're one step closer to World Language Action
00:00
1
Julius Berner
@julberner
Aug 28
Great collaboration with the FastVideo team to distill @MiniMax_AI H3 - now running faster than real-time!⚡️ More updates coming soon - stay tuned!
Hao AI Lab
@haoailab
Aug 28
(1/6) Open Weight @MiniMax_AI FastH3 v1: Generate 15s 768p video in 13s 🚀 - FastVideo collab w/ @nuvalab + FastGen - Up to 14x speedup on @NVIDIAAI Blackwell GPU - Fully open so community can run and improve the acceleration recipe. The era of open weight video models just
00:00
6
Julius Berner
@julberner
Aug 27
Great to see our Parallel Decoding Distillation (PDD) already being applied to MiniMax-H3🚀 Alibaba scaled it up just a month after our paper release (before our code even dropped) - guess that proves how simple it is to implement! Learn more about PDD: x.com/julberner/stat…
Wildminder
@wildmindai
Aug 26
Another cool stuff from Alibaba! MiniMax-H3 Acc LoRAs. Parallel Decoding Distillation for efficient video gen in a few steps. FL2VA + Ref2VA huggingface.co/alibaba-pai/Mi…
00:00
1
Julius Berner
@julberner
Aug 1
Qwen-Image-Flash is here for ultra-fast text-to-image generation! We distilled Qwen-Image down to just 4 steps in Model Optimizer using FastGen distillation methods and the NeMo AutoModel backend. More checkpoints dropping soon! 👀
Stable Diffusion Tutorials
@SD_Tutorial
Jul 28
NVIDIA's 😃Qwen-Image-Flash T2I model -generates images using a four-step -DMD2-distilled version of Qwen/Qwen-Image -distillation used DMD2 from NVIDIA FastGen NVIDIA Model Optimizer, and NVIDIA AutoModel while retaining the base model architecture 👇 huggingface.co/nvidia/Qwen-Im…
2