Ecosystem
Tools

CrypTen is a framework for Privacy Preserving ML. Its goal is to make secure computing techniques accessible to ML practitioners.

TorchQuantum is a quantum classical simulation framework based on PyTorch. It supports statevector, density matrix simulation and pulse simulation on different hardware platforms such as CPUs and GPUs.

PyTorch/TorchScript compiler for NVIDIA GPUs using TensorRT

Train PyTorch models with Differential Privacy

Basic Utilities for PyTorch Natural Language Processing (NLP).

Data-centric declarative deep learning framework

DeepSpeed is a deep learning optimization library that makes distributed training easy, efficient, and effective.

Poutyne is a Keras-like framework for PyTorch and handles much of the boilerplating code needed to train neural networks.

Flower - A Friendly Federated Learning Framework

A lightweight declarative PyTorch wrapper for context switching between devices, distributed modes, mixed-precision, and PyTorch extensions.

SimulAI is basically a toolkit with pipelines for physics-informed machine learning.

torchdistill is a coding-free framework built on PyTorch for reproducible deep learning and knowledge distillation studies.

PyTorch Geometric Temporal is a temporal (dynamic) extension library for PyTorch Geometric.

depyf is a tool to help users understand and adapt to PyTorch compiler torch.compile.

AdaptDL is a resource-adaptive deep learning training and scheduling framework.

TorchOpt is a PyTorch-based library for efficient differentiable optimization.

Colossal-AI is a Unified Deep Learning System for Big Model Era

AllenNLP is an open-source research library built on PyTorch for designing and evaluating deep learning models for NLP.

PySyft is a Python library for encrypted, privacy preserving deep learning.

🚀 A simple way to train and use PyTorch models with multi-GPU, TPU, mixed-precision

An open source hyperparameter optimization framework to automate hyperparameter search.

Renate is a library providing tools for re-training pytorch models over time as new data becomes available.

ParlAI is a unified platform for sharing, training, and evaluating dialog models across many tasks.

A toolbox for adversarial robustness research. It contains modules for generating adversarial examples and defending against attacks.

Deep Graph Library (DGL) is a Python package built for easy implementation of graph neural network model family, on top of PyTorch and other frameworks.

A modular framework for vision & language multimodal research from Facebook AI Research (FAIR).

FairScale is a PyTorch extension library for high performance and large scale training on one or multiple machines/nodes.

Ray is a fast and simple framework for building and running distributed applications.

baal (bayesian active learning) aims to implement active learning using metrics of uncertainty derived from approximations of bayesian posteriors in neural networks.

Minimalist Neural Machine Translation toolkit for educational purposes

Kornia is a differentiable computer vision library that consists of a set of routines and differentiable modules to solve generic CV problems.

A library for state-of-the-art self-supervised learning

PennyLane is a library for quantum ML, automatic differentiation, and optimization of hybrid quantum-classical computations.

higher is a library which facilitates the implementation of arbitrarily complex gradient-based meta-learning algorithms and nested optimisation loops with near-vanilla PyTorch.

The Unified Machine Learning Framework

Stable Baselines3 (SB3) is a set of reliable implementations of reinforcement learning algorithms in PyTorch.

Flair is a very simple framework for state-of-the-art natural language processing (NLP).

ML Prediction, Planning and Simulation for Self-Driving built on PyTorch.

Flexible and powerful tensor operations for readable and reliable code.

Fast and extensible image augmentation library for different CV tasks like classification, segmentation, object detection and pose estimation.

Catalyst helps you write compact, but full-featured deep learning and reinforcement learning pipelines with a few lines of code.

Machine learning metrics for distributed, scalable PyTorch applications.

A Python toolbox for data mining on Partially-Observed Time Series (POTS) and helps engineers focus more on the core problems in rather than missing parts in their data.

State-of-the-art Natural Language Processing for PyTorch.

Detectron2 is FAIR's next-generation platform for object detection and segmentation.

PyTorch Lightning is a Keras-like ML library for PyTorch. It leaves core training and validation logic to you and automates the rest.

PyTorch Geometric is a library for deep learning on irregular input data such as graphs, point clouds, and manifolds.

Pipeline Abstractions for Deep Learning in PyTorch

fastai is a library that simplifies training fast and accurate neural nets using modern best practices.

BoTorch is a library for Bayesian Optimization. It provides a modular, extensible interface for composing Bayesian optimization primitives.

A complete and open-sourced solution for injecting domain-specific knowledge into pre-trained LLM.

Determined is a platform that helps deep learning teams train models more quickly, easily share GPU resources, and effectively collaborate.

Horovod is a distributed training library for deep learning frameworks. Horovod aims to make distributed DL fast and easy to use.

pomegranate is a library of probabilistic models that is built in a modular manner and treats all models as the probability distributions that they are.

Lightly is a computer vision framework for self-supervised learning.

MONAI provides domain-optimized foundational capabilities for developing healthcare imaging training workflows.

Hummingbird compiles trained ML models into tensor computation for faster inference.

PyKale is a PyTorch library for multimodal learning and transfer learning with deep learning and dimensionality reduction on graphs, images, texts, and videos.

PFRL is a deep reinforcement learning library that implements various state-of-the-art deep reinforcement algorithms in Python using PyTorch.

FuseMedML is a python framework accelerating ML based discovery in the medical field by encouraging code reuse

skorch is a high-level library for PyTorch that provides full scikit-learn compatibility.

A runtime fault injection tool for PyTorch.

PyPose is a robotics-oriented, PyTorch-based library that combines deep perceptual models with physics-based optimization techniques, so that users can focus on their novel applications.

Avalanche: an End-to-End Library for Continual Learning

Ignite is a high-level library for training neural networks in PyTorch. It helps with writing compact, but full-featured training loops.

Forte is a toolkit for building NLP pipelines featuring composable components, convenient data interfaces, and cross-task interaction.

Diffusers provides pretrained diffusion models across multiple modalities, such as vision and audio, and serves as a modular toolbox for inference and training of diffusion models.

A PyTorch framework for deep learning on point clouds.

Datasets, transforms, and models for geospatial data

A deep learning library for video understanding research. Hosts various video-focused models, datasets, training pipelines and more.

Substra is a federated learning Python library to run federated learning experiments at scale on real distributed data.

Captum (“comprehension” in Latin) is an open source, extensible library for model interpretability built on PyTorch.

Framework for reproducible classification of Alzheimer's Disease

TorchDrift is a data and concept drift library for PyTorch. It lets you monitor your PyTorch models to see if they operate within spec.

A Python package for improving PyTorch performance on Intel platforms

Polyaxon is a platform for building, training, and monitoring large-scale deep learning applications.

octoml-profile is a python library and cloud service designed to provide a simple experience for assessing and optimizing the performance of PyTorch models.

Pyro is a universal probabilistic programming language (PPL) written in Python and supported by PyTorch on the backend.

USB is a Pytorch-based Python package for Semi-Supervised Learning (SSL). It is easy-to-use/extend, affordable to small groups, and comprehensive for developing and evaluating SSL algorithms.

Intel® Neural Compressor provides unified APIs for network compression technologies for faster inference

ClearML is a full system ML / DL experiment manager, versioning and ML-Ops solution.

PyTorch3D provides efficient, reusable components for 3D Computer Vision research with PyTorch.

GPyTorch is a Gaussian process library implemented using PyTorch, designed for creating scalable, flexible Gaussian process models.

A powerful and flexible machine learning platform for drug discovery.

A unified ensemble framework for PyTorch to improve the performance and robustness of your deep learning model.

An open source framework for deep learning on satellite and aerial imagery.

ONNX Runtime is a cross-platform inferencing and training accelerator.

pystiche is a framework for Neural Style Transfer (NST) built upon PyTorch.

TensorLy is a high level API for tensor methods and deep tensorized neural networks in Python that aims to make tensor learning simple.

RoMa is a standalone library to handle rotation representations with PyTorch (rotation matrices, quaternions, rotation vectors, etc). It aims for robustness, ease-of-use, and efficiency.

library of algorithms to speed up neural network training

OpenMMLab covers a wide range of computer vision research topics including classification, detection, segmentation, and super-resolution.

TIAToolbox provides an easy-to-use API where researchers can use, adapt and create models for CPath.

A PyTorch-based knowledge distillation toolkit for natural language processing

NeMo: a toolkit for conversational AI.

TorchIO is a set of tools to efficiently read, preprocess, sample, augment, and write 3D medical images in deep learning applications written in PyTorch.

The PopTorch interface library is a simple wrapper for running PyTorch programs directly on Graphcore IPUs.

A framework for elegantly configuring complex applications.

The easiest way to use deep metric learning in your application. Modular, flexible, and extensible.

A generalizable application framework for segmentation, regression, and classification using PyTorch