Functional Transforms in PyTorch Using Torch.func

Functional Transforms in PyTorch Using Torch.func

Learn about functional transforms in PyTorch using the torch.func module in this 29-minute conference talk presented by Meta Tech Lead Shagun Sodhani at the Toronto Machine Learning Series. Explore practical use cases where torch.func enhances PyTorch API functionality, including computing per-sample-gradients, vectorizing functions, and creating model ensembles. Gain insights into potential pitfalls and best practices when implementing torch.func in your PyTorch projects.