GLACIER combines graph, SMILES, and descriptor encoders with Finsler fusion and contrastive distillation to produce an efficient multimodal model for molecular property prediction.
On an asymmetrical metric in the four-space of general rela- tivity.Phys
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
GLACIER: A Multimodal Student-Teacher Foundation Model for Molecular Property Prediction
GLACIER combines graph, SMILES, and descriptor encoders with Finsler fusion and contrastive distillation to produce an efficient multimodal model for molecular property prediction.