This repository provides a curated list of resources for chemical pre-trained models, serving as a comprehensive survey for researchers and practitioners interested in molecular representation learning. It aims to systematically cover pre-training strategies, applications, open-source models, and datasets in this rapidly evolving field.

How It Works

The repository categorizes pre-training approaches into general strategies, knowledge-enriched methods, hard negative mining, and tuning techniques. It highlights various model architectures and their applications in areas like drug discovery and property prediction, offering a structured overview of the landscape.

Quick Start & Requirements

This repository is a curated list of papers and models, not a runnable codebase. Specific model requirements and installation instructions would be found by following the links provided for each individual model.

Highlighted Details

• Comprehensive coverage of pre-training strategies including contrastive learning, masked modeling, and knowledge graph integration.
• Extensive list of open-sourced molecular pre-trained models with details on architecture, pre-training datasets, and parameter counts.
• Curated list of relevant datasets for molecular graph representation learning.
• Includes links to associated papers, code repositories, and demos for many listed models.

Maintenance & Community

The repository is maintained by junxia97, with contributions welcomed via GitHub issues and pull requests. The last update was on 2023-6-17, indicating active, though potentially infrequent, maintenance.

Licensing & Compatibility

The repository itself is a list of resources and does not have a specific license. The licenses of the individual models and papers linked within would vary and should be checked separately.

Limitations & Caveats

As a survey, this repository does not provide a unified API or framework. Users must consult individual model resources for specific implementation details, dependencies, and potential compatibility issues. The rapid pace of research means some information may become outdated.