Accelerate was created for PyTorch users who like to write the training loop of PyTorch models but are reluctant to write and maintain the boilerplate code needed to use multi-GPUs/TPU/fp16. It abstracts exactly and only the boilerplate code related to multi-GPUs/TPU/fp16 and leaves the rest of your code unchanged.

PyTorch extension for handling deeply nested sequences of variable length.

Equinox is a comprehensive JAX library that provides a wide range of tools and features not found in core JAX, including neural networks with PyTorch-like syntax, filtered APIs for transformations, PyTree manipulation routines, and advanced features like runtime errors.

Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a set of AI libraries for simplifying ML compute. These are the provided Ray AI libraries:

• Data: Scalable datasets for ML;

• Train: Distributed training;

• Tune: Scalable hyperparameter tuning;

• RLlib: Scalable reinforcement learning;

• Serve: Scalable and programmable serving.

Datasets is a lightweight library providing two main features:

• one-line dataloaders for many public datasets

• efficient data pre-processing

TensorStore is a C++ and Python software library designed for storage and manipulation of large multi-dimensional arrays that:

• Provides advanced, fully composable indexing operations and virtual views.

• Provides a uniform API for reading and writing multiple array formats, including zarr and N5.

• Natively supports multiple storage systems, such as local and network filesystems, Google Cloud Storage, Amazon S3-compatible object stores, HTTP servers, and in-memory storage.

• Offers an asynchronous API to enable high-throughput access even to high-latency remote storage.

• Supports read caching and transactions, with strong atomicity, isolation, consistency, and durability (ACID) guarantees.

• Supports safe, efficient access from multiple processes and machines via optimistic concurrency.

Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a set of AI libraries for simplifying ML compute. These are the provided Ray AI libraries:

• Data: Scalable datasets for ML;

• Train: Distributed training;

• Tune: Scalable hyperparameter tuning;

• RLlib: Scalable reinforcement learning;

• Serve: Scalable and programmable serving.

Melissa is a file-avoiding, adaptive, fault-tolerant and elastic framework, to run large-scale sensitivity analysis or deep-surrogate training on supercomputers. This package builds the API used when instrumenting the clients.

Orbax is a namespace providing common utility libraries for JAX users. Orbax also includes a serialization library for JAX users, enabling the exporting of JAX models to the TensorFlow SavedModel format.

Library implementing Block-GMres with Inexact Breakdown and Deflated Restarting, Breakdown Free Block Conjudate Gradiant, Block General Conjugate Residual and Block General Conjugate Residual with Inner Orthogonalization and with inexact breakdown and deflated restarting.

The Linear Algebra PACKage (LAPACK) is a standard software library for numerical linear algebra. The objective of LAPACK++ is to provide a convenient, performance oriented API for development in the C++ language, that, for the most part, preserves established conventions, while, at the same time, takes advantages of modern C++ features, such as: namespaces, templates, exceptions, etc.

Grace is a 2D plotting tool for the X Window System. It has a Motif-based GUI and a scripting language that includes curve fitting, analysis, and export capabilities.

The Basic Linear Algebra Subprograms (BLAS) have been around for many decades and serve as the de facto standard for performance-portable and numerically robust implementation of essential linear algebra functionality. The objective of BLAS++ is to provide a convenient, performance oriented API for development in the C++ language, that, for the most part, preserves established conventions, while, at the same time, takes advantages of modern C++ features, such as: namespaces, templates, exceptions, etc.

STIR is an object-oriented framework for tomographic image reconstruction, with an emphasis on iterative reconstruction in PET and SPECT. This package includes the C++ core and Python bindings.

MVAPICH (pronounced as “em-vah-pich”) is an open-source MPI software to exploit the novel features and mechanisms of high-performance networking technologies (InfiniBand, iWARP, RDMA over Converged Enhanced Ethernet (RoCE v1 and v2), Slingshot 10, and Rockport Networks) and deliver best performance and scalability to MPI applications. MVAPICH 4.1 has support for the Cray Slingshot 11, Cornelis OPX, and Intel PSM3 interconnects through the OFI libfabric library, and for the UCX communication library.

MVAPICH2 (pronounced as “em-vah-pich 2”) is an open-source MPI software to exploit the novel features and mechanisms of high-performance networking technologies (InfiniBand, iWARP, RDMA over Converged Enhanced Ethernet (RoCE v1 and v2), Slingshot 10, and Rockport Networks) and deliver best performance and scalability to MPI applications.

The etelemetry Python client facilitates communication with the etelemetry server, providing version information and checking for critical bugs in projects. The client allows you to retrieve project details and compare versions to identify and warn about problematic versions.

migas (mee-gahs) is a Python client to facilitate communication with a migas server.

NIPY provides a platform-independent Python environment for the analysis of functional brain imaging data.

The nipype1-workflows repository contains legacy workflows from Nipype 1.x, showcasing nearly a decade of development in neuroimaging data processing and analysis.

The PETPVC toolbox comprises a suite of methods, both classic and more recent approaches, for the purposes of applying PVC to PET data. Eight core PVC techniques are available, and those core methods can be combined to create a total of 22 different PVC techniques.

This package provides utilities for feature analysis, preprocessing and visualization of image quality metrics generated by MRIQC.

Nilearn enables approachable and versatile analyses of brain volumes and surfaces. It provides statistical and machine-learning tools, with instructive documentation & open community.

CiftiLib is a C++ library for CIFTI-2 file reading/writing. It additionally supports CIFTI-1 files, and supports both on-disk and in-memory access. It also provides C++ code for reading and writing generic NIfTI-1 and NIfTI-2 files.

CIFTI (Connectivity Informatics Technology Initiative) standardizes file formats for the storage of connectivity data. These formats are developed by the Human Connectome Project and other interested parties.

See http://www.nitrc.org/projects/cifti/ for more information.