“诺奖工厂”的成功秘诀是什么？

一种全自动冷冻电子显微镜从头结构建模方法。

该方法使相关药物中和作用提高25倍，亲和力提高37倍。

On the eve of April 21st, I received the mail that my work had been accepted for the main track of IJCAI-2022. This year’s acceptance rate…

Flow matching (FM) is a new generative modelling paradigm which is rapidly gaining popularity in the deep learning community. Flow matching combines aspects ...

为加速高性能 OER 催化剂的开发铺平了道路。

Improving the log-likelihood of diffusion models

业绩喜人

AlphaFold 结合流匹配

成功率40%，用于蛋白质序列设计的上下文感知几何深度学习。

解决晶体学相位问题。

速度比等变MPNN提高约 30 倍。

Where's the circle? And how does it relate to where e^(-x^2) comes from?Help fund future projects: https://www.patreon.com/3blue1brown
Special thanks to these supporters: https://www.3blue1brown.com/lessons/gaussian-integral#thanks
An equally valuable form…

1/ 🧬 Excited to share CHEAP, our new work on compressed protein embeddings. We characterize the joint distribution of p(sequence, structure) in ESMFold's latent space, and find cool tidbits on compressibility, tokenizability, and pathologies:

https://t.co/SkLCd2jBXn…

松弛感拉满？

基于深度学习的高效紧束缚方法。