April 29, 2019
Stochastic Weight Averaging in PyTorch
In this blogpost we describe the recently proposed Stochastic Weight Averaging (SWA) technique [1, 2], and its new implementation in torchcontrib. SWA is a simple procedure that improves generalization in deep learning over Stochastic Gradient Descent (SGD) at no additional cost, and can be used as a drop-in replacement for any other optimizer in PyTorch. SWA has a wide range of applications and features:
May 02, 2018
The road to 1.0: production ready PyTorch
We would like to give you a preview of the roadmap for PyTorch 1.0 , the next release of PyTorch. Over the last year, we’ve had 0.2, 0.3 and 0.4 transform PyTorch from a [Torch+Chainer]-like interface into something cleaner, adding double-backwards, numpy-like functions, advanced indexing and removing Variable boilerplate. At this time, we’re confident that the API is in a reasonable and stable state to confidently release a 1.0.
April 22, 2018
PyTorch 0.4.0 Migration Guide
Welcome to the migration guide for PyTorch 0.4.0. In this release we introduced many exciting new features and critical bug fixes, with the goal of providing users a better and cleaner interface. In this guide, we will cover the most important changes in migrating existing code from previous versions:
