HIPS/autograd
Automatic differentiation library that computes gradients of NumPy code using reverse-mode backpropagation.
Collecting fresh signals — velocity needs a few days of history.
collecting data…
star history
Autograd can automatically differentiate native Python and NumPy code, supporting loops, conditionals, recursion, and closures. It implements reverse-mode differentiation (backpropagation) for efficient gradient computation of scalar-valued functions with array arguments, as well as forward-mode differentiation. The library was developed at Harvard and directly contributed to the development of JAX, serving as foundational ML tooling for gradient-based optimization.