torchgpipe
Pipeline Parallelism Library
A PyTorch-based library for efficient training of large neural networks using pipeline parallelism and automatic recomputation of gradients.
A GPipe implementation in PyTorch
820 stars
33 watching
100 forks
Language: Python
last commit: 8 months ago checkpointingdeep-learninggpipemodel-parallelismparallelismpipeline-parallelismpytorch
Related projects:
Repository | Description | Stars |
---|---|---|
| Trains PyTorch models using a genetic algorithm | 96 |
| A PyTorch framework simplifying neural network training with automated boilerplate code and callback utilities | 572 |
| An efficient library for differentiable optimization built on top of PyTorch. | 554 |
| A C++ implementation of PyTorch tutorials | 1,978 |
| A PyTorch module for dynamic batching and optimized computation on deep neural networks | 221 |
| A JavaScript library that provides GPU-accelerated deep learning capabilities with automatic differentiation and neural network layers. | 1,093 |
| Implementation of a deep learning model for generating high-quality images with improved stability and variation. | 538 |
| A PyTorch model fitting library designed to simplify the process of training deep learning models. | 636 |
| A collection of common machine learning evaluation metrics implemented in PyTorch | 110 |
| A comprehensive Python library with PyTorch extensions for rapid prototyping and machine learning model development. | 1,523 |
| Provides a CUDA backend for the PyTorch deep learning framework | 337 |
| An implementation of cascaded refinement networks for generating photorealistic images from semantic layouts | 65 |
| Provides a flexible and configurable framework for training deep learning models with PyTorch. | 1,196 |
| Enables training of reinforcement learning models with PyTorch on Godot Engine using shared memory | 214 |
| Bindings for PyTorch's tensor library in OCaml for GPU acceleration and automatic differentiation | 416 |