A 1.5-billion-molecule pretrained transformer family, NovoMolGen, sets new state-of-the-art results in de novo and goal-directed molecule generation, and shows pretraining loss correlates only weakly with downstream generation quality.
Molecular Fingerprints for Robust and Efficient ML-Driven Molecular Generation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
We propose a novel molecular fingerprint-based variational autoencoder applied for molecular generation on real-world drug molecules. We define more suitable and pharma-relevant baseline metrics and tests, focusing on the generation of diverse, drug-like, novel small molecules and scaffolds. When we apply these molecular generation metrics to our novel model, we observe a substantial improvement in chemical synthetic accessibility ($\Delta\bar{{SAS}}$ = -0.83) and in computational efficiency up to 5.9x in comparison to an existing state-of-the-art SMILES-based architecture.
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
NovoMolGen: Rethinking Molecular Language Model Pretraining
A 1.5-billion-molecule pretrained transformer family, NovoMolGen, sets new state-of-the-art results in de novo and goal-directed molecule generation, and shows pretraining loss correlates only weakly with downstream generation quality.