BERTology of Molecular Property Prediction

2026-03-13 · Mohammad Mostafanejad, Paul Saxe, T. Daniel Crawford arxiv

Chemical language models (CLMs) have emerged as promising competitors to popular classical machine learning models for molecular property prediction (MPP) tasks. However, an increasing number of studies have reported inconsistent and contradictory results for the performance of CLMs across various MPP benchmark tasks. In this study, we conduct and analyze hundreds of meticulously controlled experiments to systematically investigate the effects of various factors, such as dataset size, model size, and standardization, on the pre-training and fine-tuning performance of CLMs for MPP. In the absence of well-established scaling laws for encoder-only masked language models, our aim is to provide comprehensive numerical evidence and a deeper understanding of the underlying mechanisms affecting the performance of CLMs for MPP tasks, some of which appear to be entirely overlooked in the literature.

📄 PDF Abstract BibTeX arXiv:2603.13627

Code (0)

등록된 구현이 없습니다.

Tasks

Molecular Property Prediction

Property-Aware Relation Networks for Few-Shot Molecular Property Prediction

2021-07-16 · NeurIPS 2021 12 · Yaqing Wang, Abulikemu Abuduweili, Quanming Yao, Dejing Dou

Molecular property prediction plays a fundamental role in drug discovery to identify candidate molecules with target properties. However, molecular property prediction is essentially a few-shot problem which makes it har…

Drug Discovery Graph Learning Meta-Learning Molecular Property Prediction+3

Pre-training Transformers for Molecular Property Prediction Using Reaction Prediction

2022-07-06 · Johan Broberg, Maria Bånkestad, Erik Ylipää

Molecular property prediction is essential in chemistry, especially for drug discovery applications. However, available molecular property data is often limited, encouraging the transfer of information from related data.…

Drug Discovery Molecular Property Prediction molecular representation Prediction+3

Few-Shot Graph Learning for Molecular Property Prediction

2021-02-16 · Zhichun Guo, Chuxu Zhang, Wenhao Yu, John Herr 외

The recent success of graph neural networks has significantly boosted molecular property prediction, advancing activities such as drug discovery. The existing deep neural network methods usually require large training da…

Attribute Drug Discovery Graph Learning Graph Neural Network+7

Contrastive Dual-Interaction Graph Neural Network for Molecular Property Prediction

2024-05-04 · Zexing Zhao, Guangsi Shi, Xiaopeng Wu, Ruohua Ren 외

Molecular property prediction is a key component of AI-driven drug discovery and molecular characterization learning. Despite recent advances, existing methods still face challenges such as limited ability to generalize,…

Drug Discovery Graph Neural Network Molecular Property Prediction Prediction+1

Graph neural networks for the prediction of molecular structure-property relationships

2022-07-25 · Jan G. Rittig, Qinghe Gao, Manuel Dahmen, Alexander Mitsos 외

Molecular property prediction is of crucial importance in many disciplines such as drug discovery, molecular biology, or material and process design. The frequently employed quantitative structure-property/activity relat…

Drug Discovery Molecular Property Prediction Prediction Property Prediction