LaMM pre-trains neural network potentials on about 300 million mixed labeled and unlabeled structures using denoising pseudo-forces and a load-balanced multi-GPU pipeline, improving HME21 fine-tuning accuracy and speed.
Mat- tersim: A deep learning atomistic model across elements, temperatures and pressures, 2024
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
LaMM: Semi-Supervised Pre-Training of Large-Scale Materials Models
LaMM pre-trains neural network potentials on about 300 million mixed labeled and unlabeled structures using denoising pseudo-forces and a load-balanced multi-GPU pipeline, improving HME21 fine-tuning accuracy and speed.