_static/logo_aiaccel.png

Aiaccel Documantation¶

Aiaccel is a toolkit for accelerating machine learning research.

Configuration management

OmegaConf-based config management.

user_guide/config.html

Job management

HPC-oriented job abstraction.

user_guide/config.html

PyTorch/Lightning toolkit

Training toolkit for HPC clusters.

user_guide/torch.html

Hyperparameter optimization

Ready-to-use algorithms/tools.

user_guide/hpo.html

Key Features¶

Research-Oriented: Designed to accelerate your research cycles written in Python
HPC Optimized: Intended to use in high-performance computing (HPC) clusters, including AI Bridging Cloud Infrastructure (ABCI).
Highly Modular: Designed to let you pick up any part of aiaccel for your research project.

Aiaccel is used in …¶

M3L: Multimodal machine listening toolkit (https://github.com/b-sigpro/m3l)
SBSS: Scalable blind source separation toolkit (https://github.com/b-sigpro/sbss)

Acknowledgments¶

Part of this work was developed under a commissioned project of the New Energy and Industrial Technology Development Organization (NEDO).
Part of this software was developed by using ABCI 3.0 provided by AIST and AIST Solutions.
Part of this software was developed by using the TSUBAME4.0 supercomputer at Institute of Science Tokyo.