Julius Berner

@[email protected]
90 Followers
34 Following
11 Posts
PhD student @ UniVie | research intern @ MetaAI | former research intern @ NVIDIA | bridging theory and practice in deep learning
Websitehttps://jberner.info
Twitterhttps://twitter.com/julberner
LinkedInhttps://linkedin.com/in/julius-berner
GitHubhttps://github.com/juliusberner
Julius BernerDec 6, 2022

Had a fantastic week at #NeurIPS and look forward to seeing you all again soon!

🗒️ More details on our work: https://sigmoid.social/@jberner/109314075436493448
Thanks to my great collaborators Lorenz Richter & @karen_ullrich.

✈️ I am also grateful to G-Research for the travel grant.

Julius Berner (@[email protected])

Attached: 1 image Explore the connection between diffusion models and optimal control 🔥 🎙️ Come to our oral at the #NeurIPS workshop on score-based methods and let’s discuss how one field can benefit from the other. 📖 http://bit.ly/3UAhena Great work with Lorenz Richter and @karen_ullrich

Sigmoid Social
Julius BernerNov 22, 2022

📢 📢 New Feature in #NeuralCompression repo: Bits-Back compression for diffusion models!
Compress image data 🖼️ using diffusion models at an effective rate close to the (negative) ELBO.

See: https://github.com/facebookresearch/NeuralCompression/tree/main/projects/bits_back_diffusion

Some context ⏩ [1/3]

NeuralCompression/projects/bits_back_diffusion at main · facebookresearch/NeuralCompression

A collection of tools for neural compression enthusiasts. - NeuralCompression/projects/bits_back_diffusion at main · facebookresearch/NeuralCompression

GitHub
Julius BernerNov 9, 2022
Explore the connection between diffusion models and optimal control 🔥
🎙️ Come to our oral at the #NeurIPS workshop on score-based methods and let’s discuss how one field can benefit from the other.
📖 http://bit.ly/3UAhena
Great work with Lorenz Richter and @karen_ullrich