Type annotations and runtime checking for:

1. shape and dtype of JAX arrays;
2. PyTrees.

For example:

from jaxtyping import Array, Float, PyTree

# Accepts floating-point 2D arrays with matching dimensions
def matrix_multiply(x: Float[Array, "dim1 dim2"],
                    y: Float[Array, "dim2 dim3"]
                  ) -> Float[Array, "dim1 dim3"]:
    ...

def accepts_pytree_of_ints(x: PyTree[int]):
    ...

def accepts_pytree_of_arrays(x: PyTree[Float[Array, "batch c1 c2"]]):
    ...

pip install jaxtyping

Also install your favourite runtime type-checking package. The two most popular are typeguard (which exhaustively checks every argument) and beartype (which checks random pieces of arguments).

Full API reference

FAQ (static type checking, flake8, etc.)

Neural networks: Equinox.

Numerical differential equation solvers: Diffrax.

Computer vision models: Eqxvision.

SymPy<->JAX conversion; train symbolic expressions via gradient descent: sympy2jax.

Shape annotations + runtime type checking is inspired by TorchTyping.

The concise syntax is partially inspired by etils.array_types.

This is not an official Google product.