We are excited to announce the release of PyTorch 1.11 (release notes). This release is composed of over 3,300 commits since 1.10, made by 434 contributors. Along with 1.11, we are releasing beta versions of TorchData and functorch.
Summary:
- TorchData is a new library for common modular data loading primitives for easily constructing flexible and performant data pipelines. View it on GitHub.
- functorch, a library that adds composable function transforms to PyTorch, is now available in beta. View it on GitHub.
- Distributed Data Parallel (DDP) static graph optimizations available in stable.
Introducing TorchData
We are delighted to present the Beta release of TorchData. This is a library of common modular data loading primitives for easily constructing flexible and performant data pipelines. Based on community feedback, we have found that the existing DataLoader bundled too many features together and can be difficult to extend. Moreover, different use cases often have to rewrite the same data loading utilities over and over again. The goal here is to enable composable data loading through Iterable-style and Map-style building blocks called “DataPipes” that work well out of the box with the PyTorch’s DataLoader.
A DataPipe takes in some access function over Python data structures, __iter__ for IterDataPipe and __getitem__ for MapDataPipe, and returns a new access function with a slight transformation applied. You can chain multiple DataPipes together to form a data pipeline that performs all the necessary data transformation.
We have implemented over 50 DataPipes that provide different core functionalities, such as opening files, parsing texts, transforming samples, caching, shuffling, and batching. For users who are interested in connecting to cloud providers (such as Google Drive or AWS S3), the fsspec and iopath DataPipes will allow you to do so. The documentation provides detailed explanations and usage examples of each IterDataPipe and MapDataPipe.
In this release, some of the PyTorch domain libraries have migrated their datasets to use DataPipes. In TorchText, the popular datasets provided by the library are implemented using DataPipes and a section of its SST-2 binary text classification tutorial demonstrates how you can use DataPipes to preprocess data for your model. There also are other prototype implementations of datasets with DataPipes in TorchVision (available in nightly releases) and in TorchRec. You can find more specific examples here.
The documentation for TorchData is now live. It contains a tutorial that covers how to use DataPipes, use them with DataLoader, and implement custom ones. FAQs and future plans related to DataLoader are described in our project’s README file.
Introducing functorch
We’re excited to announce the first beta release of functorch. Heavily inspired by Google JAX, functorch is a library that adds composable function transforms to PyTorch. It aims to provide composable vmap (vectorization) and autodiff transforms that work with PyTorch modules and PyTorch autograd with good eager-mode performance.
Composable function transforms can help with a number of use cases that are tricky to do in PyTorch today:
- computing per-sample-gradients (or other per-sample quantities)
- running ensembles of models on a single machine
- efficiently batching together tasks in the inner-loop of MAML
- efficiently computing Jacobians and Hessians as well as batched ones
Composing vmap (vectorization), vjp (reverse-mode AD), and jvp (forward-mode AD) transforms allows us to effortlessly express the above without designing a separate library for each.
For more details, please see our documentation, tutorials, and installation instructions.
Distributed Training
(Stable) DDP static graph
DDP static graph assumes that your model employs the same set of used/unused parameters in every iteration, so that it can deterministically know states like which hooks will fire, how many times the hooks will fire and gradients computation ready order after the first iteration. Static graph caches these states in the first iteration, and thus it could support features that DDP can not support in previous releases, e.g., support multiple activation checkpoints on the same parameters regardless of whether there are unused parameters or not. The static graph feature also applies performance optimizations when there are unused parameters, e.g., it avoids traversing graphs to search unused parameters every iteration, and enables dynamic bucketing order. These optimizations in the DDP static graph brought 10% QPS gain for some recommendation models.
To enable static graph, just simply set static_graph=True in the DDP API like this:
ddp_model = DistributedDataParallel(model, static_graph=True)
For more details, please see our documentation and tutorials.
Thanks for reading, If you’re interested in these updates and want to join the PyTorch community, we encourage you to join the discussion forums and open GitHub issues. To get the latest news from PyTorch, follow us on Twitter, Medium, YouTube, and LinkedIn.
Cheers!
Team PyTorch