Embedding state constraints into Operator Inference yields reduced-order models that remain stable and physically consistent when extrapolating over 200% beyond the training regime for char combustion.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2025 2verdicts
UNVERDICTED 2representative citing papers
Proposes low-rank orthogonalization and derives low-rank Muon and MSGD variants that outperform standard Muon on GPT-2 and LLaMA pretraining while providing iteration complexity bounds.
citing papers explorer
-
Physically consistent predictive reduced-order modeling by enhancing Operator Inference with state constraints
Embedding state constraints into Operator Inference yields reduced-order models that remain stable and physically consistent when extrapolating over 200% beyond the training regime for char combustion.
-
Low-rank Orthogonalization for Large-scale Matrix Optimization with Applications to Foundation Model Training
Proposes low-rank orthogonalization and derives low-rank Muon and MSGD variants that outperform standard Muon on GPT-2 and LLaMA pretraining while providing iteration complexity bounds.