API Reference

This section provides a detailed introduction to the core API interfaces in FluxVLA, enabling users to gain an in-depth understanding of the framework’s API design and usage. These APIs represent the key abstractions exposed to users, designed to simplify the complex data flows involved in VLA model training, inference, and deployment, allowing users to focus on high-level interfaces without concerning themselves with underlying implementation details.

This API documentation is organized around four primary axes: training, evaluation, inference, and configuration. It is recommended to begin with API Overview before proceeding to the relevant topic-specific sections.

The following entry points provide quick navigation:

• General conventions and parameter specifications: General Conventions

• Training and evaluation script parameters: Training and Evaluation Script Interfaces

• Model interfaces and extension points: Model Interfaces

• Dataset interfaces and data flow: Dataset Interfaces

• Inference and deployment interfaces: Inference Interfaces

• Configuration field reference: Configuration Schema