Tag
SinAE introduces a single-architecture flow-matching autoencoder using vanilla Transformers that achieves near-lossless reconstruction across molecules, crystals, and proteins, enabling cross-domain training and strong generative performance on standard benchmarks.
This paper introduces an atomistic language model that integrates a 3D atom encoder, Qwen LLM, and diffusion crystal generator to natively handle multimodal materials data, achieving state-of-the-art crystal structure prediction and de novo generation.
LapidaryEngine is a new AI model that enables fully conversational generation of crystal materials from free-form natural language, using a pivot representation for bidirectional translation and iterative refinement. It outperforms existing text-to-crystal systems by allowing intuitive, dialogue-like interaction.
PRISMat is a cost-effective, permutation-invariant autoregressive model for generating crystal slabs conditioned on surface properties, achieving 4× lower error than previous models while being more efficient than LLMs.
Crys-JEPA introduces a joint embedding predictive architecture for crystals that learns an energy-aware latent space, achieving significant improvements in stability and novelty for de novo crystal discovery.