News
PyTorch 1.0 Goes Live
Last week Facebook announced the 1.0 release of PyTorch, it's Python-first open source deep learning platform.
Facebook announced that 1.0 was coming back in May, stating that the release would offer artificial intelligence (AI) developers "a fast, seamless path from research prototyping to production deployment for a broad range of AI projects" via a "hybrid front end that seamlessly transitions between imperative and declarative execution modes," with the 1.0 version finally offering "performance at production-scale."
Previous versions have been available on GitHub for a while, but according to the PyTorch Web site, 1.0 is a vast improvement for developers. As part of this, the company introduced Torch.jit into PyTorch 1.0, a new just-in-time (JIT) compiler for models (reviewed in detail here) that essentially allows you much more flexibility on how to export and use the code you create an PyTorch.
The 1.0 version also offers a new hybrid front end for "seamless transitions between eager mode and graph mode to provide both flexibility and speed" in C++ runtimes.
As with previous versions, if desired, PyTorch can be set up in the cloud via partnerships with machine learning services from Amazon, Google and Microsoft.
The Github page for PyTorch is here. The PyTorch Web site is here.
About the Author
Becky Nagel serves as vice president of AI for 1105 Media specializing in developing media, events and training for companies around AI and generative AI technology. She also regularly writes and reports on AI news, and is the founding editor of PureAI.com. She's the author of "ChatGPT Prompt 101 Guide for Business Users" and other popular AI resources with a real-world business perspective. She regularly speaks, writes and develops content around AI, generative AI and other business tech. Find her on X/Twitter @beckynagel.