jax README

The project's README from github.com/jax-ml/jax. As one markdown file: README.md.

logo

Transformable numerical computing at scale

Continuous integration PyPI version

Transformations | Scaling | Install guide | Change logs | Reference docs

What is JAX?

JAX is a Python library for accelerator-oriented array computation and program transformation, designed for high-performance numerical computing and large-scale machine learning.

JAX can automatically differentiate native Python and NumPy functions. It can differentiate through loops, branches, recursion, and closures, and it can take derivatives of derivatives of derivatives. It supports reverse-mode differentiation (a.k.a. backpropagation) via jax.grad as well as forward-mode differentiation, and the two can be composed arbitrarily to any order.

JAX uses XLA to compile and scale your NumPy programs on TPUs, GPUs, and other hardware accelerators. You can compile your own pure functions with jax.jit. Compilation and automatic differentiation can be composed arbitrarily.

Dig a little deeper, and you'll see that JAX is really an extensible system for composable function transformations at scale.

This is a research project, not an official Google product. Expect sharp edges. Please help by trying it out, reporting bugs, and letting us know what you think!

import jax
import jax.numpy as jnp

def predict(params, inputs):
  for W, b in params:
    outputs = jnp.dot(inputs, W) + b
    inputs = jnp.tanh(outputs)  # inputs to the next layer
  return outputs                # no activation on last layer

def loss(params, inputs, targets):
  preds = predict(params, inputs)
  return jnp.sum((preds - targets)**2)

grad_loss = jax.jit(jax.grad(loss))  # compiled gradient evaluation function
perex_grads = jax.jit(jax.vmap(grad_loss, in_axes=(None, 0, 0)))  # fast per-example grads

Contents

Transformations

At its core, JAX is an extensible system for transforming numerical functions. Here are three: jax.grad, jax.jit, and jax.vmap.

Automatic differentiation with grad

Use jax.grad to efficiently compute reverse-mode gradients:

import jax
import jax.numpy as jnp

def tanh(x):
  y = jnp.exp(-2.0 * x)
  return (1.0 - y) / (1.0 + y)

grad_tanh = jax.grad(tanh)
print(grad_tanh(1.0))
# prints 0.4199743

You can differentiate to any order with grad:

print(jax.grad(jax.grad(jax.grad(tanh)))(1.0))
# prints 0.62162673

You're free to use differentiation with Python control flow:

def abs_val(x):
  if x > 0:
    return x
  else:
    return -x

abs_val_grad = jax.grad(abs_val)
print(abs_val_grad(1.0))   # prints 1.0
print(abs_val_grad(-1.0))  # prints -1.0 (abs_val is re-evaluated)

See the JAX Autodiff Cookbook and the reference docs on automatic differentiation for more.

Compilation with jit

Use XLA to compile your functions end-to-end with jit, used either as an @jit decorator or as a higher-order function.

import jax
import jax.numpy as jnp

def slow_f(x):
  # Element-wise ops see a large benefit from fusion
  return x * x + x * 2.0

x = jnp.ones((5000, 5000))
fast_f = jax.jit(slow_f)
%timeit -n10 -r3 fast_f(x)
%timeit -n10 -r3 slow_f(x)

Using jax.jit constrains the kind of Python control flow the function can use; see the tutorial on Control Flow and Logical Operators with JIT for more.

Auto-vectorization with vmap

vmap maps a function along array axes. But instead of just looping over function applications, it pushes the loop down onto the function’s primitive operations, e.g. turning matrix-vector multiplies into matrix-matrix multiplies for better performance.

Using vmap can save you from having to carry around batch dimensions in your code:

import jax
import jax.numpy as jnp

def l1_distance(x, y):
  assert x.ndim == y.ndim == 1  # only works on 1D inputs
  return jnp.sum(jnp.abs(x - y))

def pairwise_distances(dist1D, xs):
  return jax.vmap(jax.vmap(dist1D, (0, None)), (None, 0))(xs, xs)

xs = jax.random.normal(jax.random.key(0), (100, 3))
dists = pairwise_distances(l1_distance, xs)
dists.shape  # (100, 100)

By composing jax.vmap with jax.grad and jax.jit, we can get efficient Jacobian matrices, or per-example gradients:

per_example_grads = jax.jit(jax.vmap(jax.grad(loss), in_axes=(None, 0, 0)))

Scaling

To scale your computations across thousands of devices, you can use any composition of these:

Mode View? Explicit sharding? Explicit Collectives?
Auto Global ❌ ❌
Explicit Global ✅ ❌
Manual Per-device ✅ ✅ 
from jax.sharding import set_mesh, AxisType, PartitionSpec as P
mesh = jax.make_mesh((8,), ('data',), axis_types=(AxisType.Explicit,))
set_mesh(mesh)

# parameters are sharded for FSDP:
for W, b in params:
  print(f'{jax.typeof(W)}')  # f32[512@data,512]
  print(f'{jax.typeof(b)}')  # f32[512]

# shard data for batch parallelism:
inputs, targets = jax.device_put((inputs, targets), P('data'))

# evaluate gradients, automatically parallelized!
gradfun = jax.jit(jax.grad(loss))
param_grads = gradfun(params, (inputs, targets))

See the tutorial and advanced guides for more.

Gotchas and sharp bits

See the Gotchas Notebook.

Installation

Supported platforms

Linux x86_64 Linux aarch64 Mac aarch64 Windows x86_64 Windows WSL2 x86_64
CPU yes yes yes yes yes
NVIDIA GPU yes yes n/a no experimental
Google TPU yes n/a n/a n/a n/a
AMD GPU yes no n/a no experimental
Apple GPU n/a no experimental n/a n/a
Intel GPU experimental n/a n/a no no

Instructions

Platform Instructions
CPU pip install -U jax
NVIDIA GPU pip install -U "jax[cuda13]"
Google TPU pip install -U "jax[tpu]"
AMD GPU (Linux) pip install -U "jax[rocm7-local]"
Intel GPU Follow Intel's instructions.

See the documentation for information on alternative installation strategies. These include compiling from source, installing with Docker, using other versions of CUDA, a community-supported conda build, and answers to some frequently-asked questions.

Citing JAX

To cite this repository:

@software{jax2018github,
  author = {James Bradbury and Roy Frostig and Peter Hawkins and Matthew James Johnson and Yash Katariya and Chris Leary and Dougal Maclaurin and George Necula and Adam Paszke and Jake Vander{P}las and Skye Wanderman-{M}ilne and Qiao Zhang},
  title = {{JAX}: composable transformations of {P}ython+{N}um{P}y programs},
  url = {http://github.com/jax-ml/jax},
  version = {0.3.13},
  year = {2018},
}

In the above bibtex entry, names are in alphabetical order, the version number is intended to be that from jax/version.py, and the year corresponds to the project's open-source release.

A nascent version of JAX, supporting only automatic differentiation and compilation to XLA, was described in a paper that appeared at SysML 2018. We're currently working on covering JAX's ideas and capabilities in a more comprehensive and up-to-date paper.

Reference documentation

For details about the JAX API, see the reference documentation.

For getting started as a JAX developer, see the developer documentation.

jax (jax-ml/jax)

Searching Database…

36.5K ·GitHub
Jax
What is this?JAX is a Python library for accelerator-oriented array computation and program transformation, built for high-performance numerical computing and large-scale machine learning.
Is JAX a library I import into my own Python code, or a separate program?It is a library you import, not a standalone app: you write ordinary Python and NumPy functions, and JAX transforms them — differentiating, compiling and vectorizing — while your code stays plain Python.
What should I watch out for when writing JAX code?jax.jit restricts the Python control flow a compiled function can use, and the README keeps a Gotchas Notebook for sharp bits. The project also bills itself as research, not an official Google product — expect rough edges.
What can it actually do to my code?Three composable transformations: jax.grad computes reverse-mode gradients (backprop), jax.jit compiles functions end-to-end with XLA, and jax.vmap auto-vectorizes by pushing loops down into matrix operations. Compose them freely — vmap over grad yields efficient per-example gradients.
Can it differentiate regular Python code, not just formulas?Yes — jax.grad differentiates native Python and NumPy functions through loops, branches, recursion and closures, and you can take derivatives of derivatives to any order, mixing forward- and reverse-mode arbitrarily.
How does it scale to many devices?Three composable styles: compiler-driven auto-parallelism (you program a single global machine), explicit sharding you can inspect with jax.typeof, and manual per-device programming with explicit collectives. The README targets scaling across thousands of devices.
Is JAX an official Google product?No — the README calls it a research project with sharp edges and asks users to try it, report bugs and share feedback. The core ideas (autodiff plus XLA compilation) were first described in a SysML 2018 paper.
What are similar projects?tilelang, cupy, pytorch, turboquant_plus, vision_transformer, DeepSpeed, numba, TornadoVM
Which topics is it in?tpu, wsl2, jax, scientific-computing, amd, numpy, google, compiler, apple-silicon, nvidia, cuda, performance
Creator
jax-ml
Size
191.7 MB
Stars
36.5K
Updated
CommitOct 17Releasejax-v0.11.2Sep 17
Age
7y
Commits · 90 Days