rocAL documentation

The ROCm Augmentation Library (rocAL) is a Python library that provides a way to customize audio, video, and image pipelines for different datasets and models, improving the throughput and performance of deep learning applications. rocAL is optimized for loading and pre-processing data for deep learning applications, with support for multiple data formats and augmentations.

The rocAL public repository is located at ROCm/rocAL.

Installation

• rocAL prerequisites

• Installing rocAL with the package installer

• Building and installing rocAL from source

Conceptual

• rocAL Operators

How to

• Create and run the rocAL pipeline

• Run PyTorch training with rocAL

• Run TensorFlow training with rocAL

• Run JAX training with rocAL

Examples

• rocAL framework integration examples

Reference

• rocAL pipelines

• rocAL RNNT dataloading

• rocAL Python API overview

• rocAL Python reference

The C++ API for contributors:

• rocAL C++ API overview

• rocAL C++ reference

To contribute to the documentation refer to Contributing to ROCm Docs.

You can find licensing information on the Licensing page.