Different uMLIPs encode chemical space in distinct ways, with high cross-model feature reconstruction errors, and fine-tuning preserves strong pre-training bias in the latent features.
Data-efficient fine- tuning of foundational models for first-principles quality sublimation enthalpies,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
physics.chem-ph 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Comparing the latent features of universal machine-learning interatomic potentials
Different uMLIPs encode chemical space in distinct ways, with high cross-model feature reconstruction errors, and fine-tuning preserves strong pre-training bias in the latent features.