JAX is Autograd and XLA, brought together for high-performance machine learning
research.
With its updated version of Autograd, 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 grad as well as forward-mode differentiation, and the two can be composed
arbitrarily to any order.
What's new is that JAX uses XLA to compile and run your NumPy programs on GPUs
and TPUs. Compilation happens under the hood by default, with library calls
getting just-in-time compiled and executed. But JAX also lets you just-in-time
compile your own Python functions into XLA-optimized kernels using a
one-function API, jit. Compilation and automatic differentiation can be composed
arbitrarily, so you can express sophisticated algorithms and get maximal
performance without leaving Python. You can even program multiple GPUs or TPU
cores at once using pmap, and differentiate through the whole thing.
Dig a little deeper, and you'll see that JAX is really an extensible system for
composable function transformations. Both grad and jit are instances of such
transformations. Others are vmap for automatic vectorization and pmap for
single-program multiple-data (SPMD) parallel programming of multiple
accelerators, with more to come.
NOTE: If this package has multiple flavors (see below), then use one of them instead of the name specified above. NOTE: This is a Python port. Instead of py311-jax listed in the above command, you can pick from the names under the Packages section.
*: Revise USES=python:... version range constraints
and strip all lower bounds of 3.9 or 3.10 because they are implied now.
python:3.9-X -> python:-X
python:3.9+ -> python
python:3.10-X -> python:-X
python:3.10+ -> python
(options such as ,build preserved)
devel/py-setuptools: convert individual consumers to ${PY_SETUPTOOLS}
Currently a no-op, but in the future outputs the correct setuptools
port depending on whether USES_PYTHON=distutils is specified.
With hat: python
PR: 270510, 270358
all: remove explicit versions in USES=python for "3.x+"
The logic in USES=python will automatically convert this to 3.8+ by
itself.
Adjust two ports that only had Python 3.7 mentioned but build fine
on Python 3.8 too.
finance/quickfix: mark BROKEN with PYTHON
libtool: compile: c++ -DHAVE_CONFIG_H -I. -I../.. -I -I. -I.. -I../.. -I../C++
-DLIBICONV_PLUG -DPYTHON_MAJOR_VERSION=3 -Wno-unused-variable
-Wno-maybe-uninitialized -O2 -pipe -DLIBICONV_PLUG -fstack-protector-strong
-fno-strict-aliasing -DLIBICONV_PLUG -Wall -ansi
-Wno-unused-command-line-argument -Wpointer-arith -Wwrite-strings
-Wno-overloaded-virtual -Wno-deprecated-declarations -Wno-deprecated -std=c++0x
-MT _quickfix_la-QuickfixPython.lo -MD -MP -MF
.deps/_quickfix_la-QuickfixPython.Tpo -c QuickfixPython.cpp -fPIC -DPIC -o
.libs/_quickfix_la-QuickfixPython.o
warning: unknown warning option '-Wno-maybe-uninitialized'; did you mean
'-Wno-uninitialized'? [-Wunknown-warning-option]
QuickfixPython.cpp:175:11: fatal error: 'Python.h' file not found
^~~~~~~~~~
1 warning and 1 error generated.
Reviewed by: portmgr, vishwin, yuri
Differential Revision: <https://reviews.freebsd.org/D40568>
Remove WWW entries moved into port Makefiles
Commit b7f05445c00f has added WWW entries to port Makefiles based on
WWW: lines in pkg-descr files.
This commit removes the WWW: lines of moved-over URLs from these
pkg-descr files.
Approved by: portmgr (tcberner)
Add WWW entries to port Makefiles
It has been common practice to have one or more URLs at the end of the
ports' pkg-descr files, one per line and prefixed with "WWW:". These
URLs should point at a project website or other relevant resources.
Access to these URLs required processing of the pkg-descr files, and
they have often become stale over time. If more than one such URL was
present in a pkg-descr file, only the first one was tarnsfered into
the port INDEX, but for many ports only the last line did contain the
port specific URL to further information.
There have been several proposals to make a project URL available as
a macro in the ports' Makefiles, over time.
(Only the first 15 lines of the commit message are shown above )
Add py-jax 0.1.72
JAX is Autograd and XLA, brought together for high-performance machine learning
research.
With its updated version of Autograd, 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 grad as well as forward-mode differentiation, and the two can be composed
arbitrarily to any order.
What's new is that JAX uses XLA to compile and run your NumPy programs on GPUs
and TPUs. Compilation happens under the hood by default, with library calls
getting just-in-time compiled and executed. But JAX also lets you just-in-time
(Only the first 15 lines of the commit message are shown above )
