ChemBERTa-2: Towards Chemical Foundation Models
arXiv:2209.01712
Abstract
Large pretrained models such as GPT-3 have had tremendous impact on modern natural language processing by leveraging self-supervised learning to learn salient representations that can be used to readily finetune on a wide variety of downstream tasks. We investigate the possibility of transferring such advances to molecular machine learning by building a chemical foundation model, ChemBERTa-2, using the language of SMILES. While labeled data for molecular prediction tasks is typically scarce, libraries of SMILES strings are readily available. In this work, we build upon ChemBERTa by optimizing the pretraining process. We compare multi-task and self-supervised pretraining by varying hyperparameters and pretraining dataset size, up to 77M compounds from PubChem. To our knowledge, the 77M set constitutes one of the largest datasets used for molecular pretraining to date. We find that with these pretraining improvements, we are competitive with existing state-of-the-art architectures on the MoleculeNet benchmark suite. We analyze the degree to which improvements in pretraining translate to improvement on downstream tasks.
arXiv commentELLIS Machine Learning for Molecule Discovery Workshop
Cited by in corpus (8)
- polyBERT: A chemical language model to enable fully machine-driven ultrafast polymer informatics
- Materials science in the era of large language models: a perspective
- HANNA: Hard-constraint Neural Network for Consistent Activity Coefficient Prediction
- Artificial Intelligence in Materials Science and Engineering: Current Landscape, Key Challenges, and Future Trajectorie
- Towards Foundation Model for Chemical Reactor Modeling: Meta-Learning with Physics-Informed Adaptation
- A Bayesian Flow Network Framework for Chemistry Tasks
- Tokenization for Molecular Foundation Models
- Machine learning for smell: Ordinal odor strength prediction of molecular perfumery components