Skip to content

Commit

Permalink
misc/py-xformers: New port: Transformers building blocks, supporting …
Browse files Browse the repository at this point in the history
…a composable construction
  • Loading branch information
yurivict committed Mar 22, 2024
1 parent 471d29f commit 60c128a
Show file tree
Hide file tree
Showing 5 changed files with 44 additions and 0 deletions.
1 change: 1 addition & 0 deletions misc/Makefile
Expand Up @@ -509,6 +509,7 @@
SUBDIR += py-vaderSentiment
SUBDIR += py-wandb
SUBDIR += py-wurlitzer
SUBDIR += py-xformers
SUBDIR += py-xgboost
SUBDIR += pyobd
SUBDIR += pytorch
Expand Down
25 changes: 25 additions & 0 deletions misc/py-xformers/Makefile
@@ -0,0 +1,25 @@
PORTNAME= xformers
DISTVERSION= 0.0.25
CATEGORIES= misc python # machine-learning
MASTER_SITES= PYPI
PKGNAMEPREFIX= ${PYTHON_PKGNAMEPREFIX}

MAINTAINER= yuri@FreeBSD.org
COMMENT= Transformers building blocks, supporting a composable construction
WWW= https://facebookresearch.github.io/xformers/

LICENSE= BSD3CLAUSE
LICENSE_FILE= ${WRKSRC}/LICENSE

BUILD_DEPENDS= pybind11>0:devel/pybind11 \
${PYTHON_PKGNAMEPREFIX}pytorch>=2.1:misc/py-pytorch@${PY_FLAVOR}
RUN_DEPENDS= ${PYTHON_PKGNAMEPREFIX}pytorch>=2.1:misc/py-pytorch@${PY_FLAVOR} \
${PYNUMPY}

USES= compiler:c++17-lang localbase python
USE_PYTHON= distutils autoplist pytest # tests fail, see https://github.com/facebookresearch/xformers/issues/1003

post-install:
@${STRIP_CMD} ${STAGEDIR}${PYTHONPREFIX_SITELIBDIR}/xformers/_C.so

.include <bsd.port.mk>
3 changes: 3 additions & 0 deletions misc/py-xformers/distinfo
@@ -0,0 +1,3 @@
TIMESTAMP = 1711070989
SHA256 (xformers-0.0.25.tar.gz) = 63f74d96c82d6b9bf3daf38f53cf173a29370ee6bd1dfde15836d0d598b6f82f
SIZE (xformers-0.0.25.tar.gz) = 4080980
11 changes: 11 additions & 0 deletions misc/py-xformers/pkg-descr
@@ -0,0 +1,11 @@
xFormers is a Toolbox to Accelerate Research on Transformers.

xFormers is:
* Customizable building blocks: Independent/customizable building blocks that
can be used without boilerplate code. The components are domain-agnostic and
xFormers is used by researchers in vision, NLP and more.
* Research first: xFormers contains bleeding-edge components, that are not yet
available in mainstream libraries like PyTorch.
* Built with efficiency in mind: Because speed of iteration matters, components
are as fast and memory-efficient as possible. xFormers contains its own CUDA
kernels, but dispatches to other libraries when relevant.
4 changes: 4 additions & 0 deletions misc/py-xformers/pkg-plist
@@ -0,0 +1,4 @@
%%PYTHON_SITELIBDIR%%/xformers/__pycache__/version%%PYTHON_EXT_SUFFIX%%.opt-1.pyc
%%PYTHON_SITELIBDIR%%/xformers/__pycache__/version%%PYTHON_EXT_SUFFIX%%.pyc
%%PYTHON_SITELIBDIR%%/xformers/cpp_lib.json
%%PYTHON_SITELIBDIR%%/xformers/version.py

0 comments on commit 60c128a

Please sign in to comment.