Scalable Spatio-Temporal SE(3) Diffusion for Long-Horizon Protein Dynamics

A scalable SE(3)-equivariant diffusion model that generates physically plausible protein trajectories over microsecond timescales using a causal diffusion transformer with joint spatiotemporal attention.

Fine-Tuning Atomistic Foundation Models for Materials Discovery

An open-source platform for fine-tuning state-of-the-art atomistic foundation models (UMA, JMP, EquiformerV2, MACE, MatterSim, ORB, etc.) with intuitive interfaces for computational chemists. Features parameter-efficient fine-tuning, message-passing pruning, and graph partitioning for accelerating foundation models.

Joint Multi-domain Pre-training for Atomic Property Prediction

A supervised pre-training strategy that leverages diverse data to advance atomic property prediction across chemical domains.