UniTS framework uses diffusion models to accelerate reaction discovery
bravo_abad · x · 2026-08-28
Research introduces UniTS, a framework using generative AI to accelerate transition-state discovery in chemical reactions. The team built UniTS-Lib, a dataset of 4,391 DFT-validated structures. The conditional diffusion model, UniTS-Gen, generates transition-state geometries from random 3D coordinates, reducing the need for expert-designed initial guesses.
More from Research
- Study Uses Active Learning to Guide Directed Evolution, Boosting Fluorescent Protein Brightness by 4x — KevinKaichuang · 2026-08-28
- EMNLP 2026 paper addresses multilingual reasoning gap in LLMs — IgorCarron · 2026-08-28
- Bayesian Optimization Resource Hub Launches Web Version — andrewgwils · 2026-08-28
- AER Paper: Anti-manipulation ML improves credit forecasting in Kenya — Afinetheorem · 2026-08-28
- Google open-sources TimesFM for zero-shot forecasting of trends — mdancho84 · 2026-08-28
- AME-2 paper accepted: Agile legged locomotion via attention maps — ChongZzZhang · 2026-08-28