Dimensions overall score 7.0

• Crystalite is a lightweight diffusion Transformer that achieves state-of-the-art results on crystal structure prediction benchmarks.result 90%
• Crystalite samples substantially faster than geometry-heavy alternatives like equivariant graph neural networks.technical 85%
• Subatomic Tokenization is a compact chemically structured atom representation that replaces high-dimensional one-hot encodings.method 90%
• The Geometry Enhancement Module (GEM) injects periodic minimum-image pair geometry directly into attention through additive geometric biases.method 90%
• Crystalite attains the best S.U.N. discovery score among the evaluated baselines for de novo generation.result 85%
• Equivariant graph neural networks for crystalline materials are costly to train and slow to sample.limitation 80%
• Subatomic Tokenization is better suited to continuous diffusion than traditional one-hot encodings.technical 75%
• Crystalite preserves the simplicity and efficiency of a standard Transformer while making it better matched to crystalline materials structure.technical 80%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card