API Reference

OmniGenBench is a comprehensive toolkit for genomic foundation models. It provides automated benchmarking and training pipelines, a hub for accessing pre-trained models, datasets, and pipelines, and a flexible framework for building custom models and tasks.

Tip

How to Use This Reference:

• Browse API documentation by functional category below.

• Each section contains detailed classes, functions, usage examples, and extension guidelines.

• Start with Abstract Classes to understand the core interfaces, then explore Auto-Pipelines, CLI Commands, and Online Hubs as needed.

Main API Categories:

• Abstract Classes: Core interfaces for datasets, models, metrics, and tokenizers. Use these as base classes for custom extensions.

• Auto-Pipelines: Automated benchmarking and training workflows, including AutoBench, AutoTrain, and related configuration classes.

• CLI Commands: Command-line tools for running benchmarks, training, and managing resources.

• Online Hubs: Access pre-trained models, datasets, and pipelines via hub utilities.

• Low-level Modules: Detailed and fine-grained modules, for development reference.

Refer to each section below for detailed documentation and usage instructions.