MatRIS-MoE and Janus enable efficient exascale training of billion-parameter universal interatomic potentials by addressing second-order derivative computation and communication overheads.
Equiformerv2: Improved equivariant transformer for scaling to higher-degree representations
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Breaking the Training Barrier of Billion-Parameter Universal Machine Learning Interatomic Potentials
MatRIS-MoE and Janus enable efficient exascale training of billion-parameter universal interatomic potentials by addressing second-order derivative computation and communication overheads.