{"added":"2026-07-05T21:22:17-04:00","apiVersion":1,"behindSince":null,"buildDepends":["devel/py-setuptools","devel/py-wheel"],"categories":["math","python"],"comment":"Differentiate, compile, and transform Numpy code","compare":null,"dependentCount":null,"description":"JAX is Autograd and XLA, brought together for high-performance machine learning\nresearch.\n\nWith its updated version of Autograd, JAX can automatically differentiate native\nPython and NumPy functions. It can differentiate through loops, branches,\nrecursion, and closures, and it can take derivatives of derivatives of\nderivatives. It supports reverse-mode differentiation (a.k.a. backpropagation)\nvia grad as well as forward-mode differentiation, and the two can be composed\narbitrarily to any order.\n\nWhat's new is that JAX uses XLA to compile and run your NumPy programs on GPUs\nand TPUs. Compilation happens under the hood by default, with library calls\ngetting just-in-time compiled and executed. But JAX also lets you just-in-time\ncompile your own Python functions into XLA-optimized kernels using a\none-function API, jit. Compilation and automatic differentiation can be composed\narbitrarily, so you can express sophisticated algorithms and get maximal\nperformance without leaving Python. You can even program multiple GPUs or TPU\ncores at once using pmap, and differentiate through the whole thing.\n\nDig a little deeper, and you'll see that JAX is really an extensible system for\ncomposable function transformations. Both grad and jit are instances of such\ntransformations. Others are vmap for automatic vectorization and pmap for\nsingle-program multiple-data (SPMD) parallel programming of multiple\naccelerators, with more to come.","flavors":[],"html":"https://daemonports.org/midnightbsd/math/py-jax/","kind":"port","lastUpdate":"2026-07-05T21:22:17-04:00","license":["Apache-2.0"],"maintainer":"ports@MidnightBSD.org","name":"${PYTHON_PKGNAMEPREFIX}jax","newest":true,"newestVersion":"0.4.38","noVulnData":true,"origin":"math/py-jax","otherSystems":[{"maintainer":"sunpoet@FreeBSD.org","newest":true,"origin":"math/py-jax","path":"/math/py-jax/","system":"FreeBSD","version":"0.4.38_2"},{"maintainer":"sunpoet@FreeBSD.org","newest":true,"origin":"math/py-jax","path":"/dragonfly/math/py-jax/","system":"DragonFly","version":"0.4.38"},{"maintainer":"sunpoet@FreeBSD.org","newest":true,"noVulnData":true,"origin":"math/py-jax","path":"","system":"HardenedBSD","version":"0.4.38_2"},{"maintainer":"sunpoet@FreeBSD.org","newest":true,"noVulnData":true,"origin":"math/py-jax","path":"","system":"GhostBSD","version":"0.4.38_2"}],"packages":null,"pkgName":"${PYTHON_PKGNAMEPREFIX}jax-0.4.38_2","related":null,"repository":"mports","requiredBy":[],"requiredByCount":0,"runDepends":["math/py-ml-dtypes","math/py-numpy","math/py-opt-einsum","science/py-scipy"],"status":null,"system":"midnightbsd","systemLabel":"MidnightBSD","version":"0.4.38_2","versions":[{"commit":"8af06753b2eaf0c58214302c2a501feadc09477a","date":"2026-07-05","version":"0.4.38"}],"vulnerabilities":null,"www":"https://jax.readthedocs.io/en/latest/"}