* gnu/packages/machine-learning.scm (llama-cpp)[arguments]<#:configure-flags>:
Use new names from the package update.
Change-Id: I14095b5a594efbcfc7abfdd50fd2930fa7878394
This also updates the qnnpack-pytorch package.
* gnu/packages/machine-learning.scm (%python-pytorch-version): Set to 2.4.0.
(%python-pytorch-src): Adjust hash accordingly.
[source]: Remove obsolete file deletions in snippet.
(python-pytorch)[arguments]<#:phases>: Disable telemetry and set Python
install directory. Remove obsolete USE_FFMPEG, USE_OPENCV and
USE_PYTORCH_QNNPACK environment variables.
[native-inputs]: Remove python-typing-extensions.
[inputs]: Add brotli, cpp-httplib and zlib. Remove qnnpack. Use oneapi-dnnl,
qnnpack-pytorch and xnnpack for all systems.
[propagated-inputs]: Add onnx and python-typing-extensions. Use cpuinfo for
all systems.
[supported-systems]: New field.
(python-pytorch-for-r-torch)[inputs]: Add qnnpack.
* gnu/packages/patches/python-pytorch-fix-codegen.patch: Adjust patch.
* gnu/packages/patches/python-pytorch-runpath.patch: Adjust patch.
* gnu/packages/patches/python-pytorch-system-libraries.patch: Adjust patch.
* gnu/packages/patches/python-pytorch-without-kineto.patch: Adjust patch.
* gnu/packages/machine-learning.scm (gloo): Update to commit 81925d1.
[arguments]: Add -DCMAKE_CXX_STANDARD=17 to #:configure-flags.
[supported-systems]: New field.
* gnu/packages/machine-learning.scm (onnx-optimizer): Update to 0.3.19.
* gnu/packages/patches/onnx-optimizer-system-library.patch: Adjust patch.
Co-authored-by: Andy Tai <atai@atai.org>
* gnu/packages/machine-learning.scm (onnx): Update to 1.16.2.
[source]: Remove patch.
[build-system]: Move to pyproject-build-system.
[arguments]<#:tests?>: Disable for systems where python-nbval is not
supported.
<#:phases>: Remove 'relax-requirements' phase. Pass "-DONNX_BUILD_TESTS" to
CMake depending on tests?.
[native-inputs]: Add python-fb-re2 and python-parameterized-next. Remove
python-nbval on unsupported systems.
* gnu/packages/patches/onnx-shared-libraries.patch: Adjust patch.
* gnu/packages/patches/onnx-skip-model-downloads.patch: Adjust patch.
* gnu/packages/patches/onnx-use-system-googletest.patch: Remove file.
* gnu/local.mk (dist_patch_DATA): Remove it.
Co-authored-by: Andy Tai <atai@atai.org>
More tests fail when building the shared library, so they are disabled. The
scripts to generate the autogenerated files run all commands at once, so they
are converted to Makefiles first.
* gnu/packages/machine-learning.scm (xnnpack): Update to commit 08f1489.
[source]: Remove patch. Adjust snippet.
[arguments]<#:configure-flags>: Add "-DXNNPACK_BUILD_TESTS=OFF".
<#:tests?>: Disable tests.
<#:phases>: Adjust 'fix-cmake' phase. Convert code generation scripts to
Makefiles and run them. Run different missing scripts.
* gnu/packages/patches/xnnpack-remove-broken-tests.patch: Remove file.
* gnu/packages/patches/xnnpack-system-libraries.patch: Remove file.
* gnu/local.mk (dist_patch_DATA): Remove them.
Until now users would have to cargo cult or inspect the private
%default-modules variable of (guix build-systems gnu) to discover which
modules to include when extending the used modules via the #:modules argument.
The renaming was automated via the command:
$ git grep -l %gnu-build-system-modules
| xargs sed 's/%gnu-build-system-modules/%default-gnu-imported-modules/' -i
* guix/build-system/gnu.scm (%gnu-build-system-modules): Rename to...
(%default-gnu-imported-modules): ... this.
(%default-modules): Rename to...
(%default-gnu-modules): ... this. Export.
(dist-package, gnu-build, gnu-cross-build): Adjust accordingly.
Change-Id: Idef307fff13cb76f3182d782b26e1cd3a5c757ee
* gnu/packages/machine-learning.scm (python-pyro-ppl): Update to 1.9.1.
[arguments]: Remove #:test-flags and skip all tests in #:tests?.
Change-Id: If90024d904bb830df2868423a5947ecaf45b46fc
* gnu/packages/machine-learning.scm (lama-cpp): Update to commit a5735e with
pkg-config support.
[arguments](configure-flags): Add cmake configure flag
for shared library build and adjust arguments to make openblas
found by cmake.
(phases) 'install-python-scripts: Remove references to deleted scripts
and add new ones upsteeam.
Change-Id: I7c4bc219a22aa9a949e811b340c7cf745b176d14
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* gnu/packages/machine-learning.scm (%python-pytorch-for-r-torch-version): New
variable.
(%python-pytorch-for-r-torch-src): New variable.
(qnnpack-pytorch-for-r-torch): New variable.
This is an internal fork of QNNPACK in the PyTorch source tree.
* gnu/packages/machine-learning.scm (%python-pytorch-version): New variable.
(%python-pytorch-src): New variable.
(qnnpack-pytorch): New variable.
The tests which fail to build if building the shared library are removed by a
patch. Autogenerated files are removed by a snippet and generated in a
separate build phase.
* gnu/packages/machine-learning.scm (xnnpack): Update to commit 51a9875.
[source]: Add patch for tests. Add snippet to remove autogenerated files.
[arguments]: New modules and phase arguments.
[inputs]: Add clog and cpuinfo.
[native-inputs]: New field.
* gnu/packages/patches/xnnpack-remove-broken-tests.patch: Add file.
* gnu/local.mk (dist_patch_DATA): Add it.
* gnu/packages/patches/xnnpack-system-libraries.patch: Update patch.
* gnu/packages/machine-learning.scm (foxi): New variable.
* gnu/packages/patches/foxi-fix-build.patch: New file.
* gnu/local.mk (dist_patch_DATA): Register it.
* gnu/packages/machine-learning.scm (fbgemm): New variable.
* gnu/packages/patches/fbgemm-use-system-libraries.patch: New file.
* gnu/local.mk (dist_patch_DATA): Register it.
