pytorch/functorch
A PyTorch library providing composable function transforms (vmap, grad, vjp, jvp) for computing gradients, jacobians, and hessians.

Not currently ranked — collecting fresh signals.
star history
Functorch brings JAX-like composable function transforms to PyTorch, enabling efficient per-sample gradients, ensemble computations, batched Jacobians/Hessians, and MAML inner-loop optimization. It integrates with PyTorch autograd for eager-mode performance and includes experimental FX tracing for ahead-of-time compilation of transform results.
Frequently asked
- What is pytorch/functorch?
- A PyTorch library providing composable function transforms (vmap, grad, vjp, jvp) for computing gradients, jacobians, and hessians.
- Is functorch open source?
- Yes — pytorch/functorch is open source, released under the BSD-3-Clause license.
- What language is functorch written in?
- pytorch/functorch is primarily written in Jupyter Notebook.
- How popular is functorch?
- pytorch/functorch has 1.4k stars on GitHub.
- Where can I find functorch?
- pytorch/functorch is on GitHub at https://github.com/pytorch/functorch.