Install this version:
emerge -a =dev-python/runai-model-streamer-bin-0.16.1
If this version is masked, you can unmask it using the autounmask tool or standard emerge options:
autounmask =dev-python/runai-model-streamer-bin-0.16.1
Or alternatively:
emerge --autounmask-write -a =dev-python/runai-model-streamer-bin-0.16.1
| Version | EAPI | Keywords | Slot |
|---|---|---|---|
| 0.16.1 | 8 | -* ~amd64 | 0 |
# Copyright 1999-2026 Gentoo Authors
# Distributed under the terms of the GNU General Public License v2
EAPI=8
DISTUTILS_USE_PEP517=no
PYTHON_COMPAT=( python3_{12..14} )
DISTUTILS_SINGLE_IMPL=1
inherit distutils-r1
MY_PN=${PN%-bin}
MY_PV=${PV}
MY_WHEEL="${MY_PN//-/_}-${MY_PV}-py3-none-manylinux2014_x86_64.whl"
DESCRIPTION="Run:ai's fast multi-tensor PyTorch model streamer (binary wheel)"
HOMEPAGE="
https://github.com/run-ai/runai-model-streamer
https://pypi.org/project/runai-model-streamer/
"
SRC_URI="
https://files.pythonhosted.org/packages/ee/3d/32656788027d72d6bb54804a1479429ad1ccb36d212ff408c4f513bdd4a5/${MY_WHEEL}
"
S="${WORKDIR}"
LICENSE="Apache-2.0"
SLOT="0"
KEYWORDS="-* ~amd64"
# Upstream releases only manylinux2014_x86_64 wheels and a Bazel-based
# source build. We ship the wheel so vllm has a working
# runai-model-streamer at runtime; cloud backends ([s3,gcs,azure])
# need google-cloud-storage / azure-identity which aren't packaged
# yet, so they're deferred — vllm's local-file model loading still
# works without them.
RDEPEND="
>=sci-ml/pytorch-2.0.0[${PYTHON_SINGLE_USEDEP}]
$(python_gen_cond_dep '
dev-python/humanize[${PYTHON_USEDEP}]
dev-python/numpy[${PYTHON_USEDEP}]
')
"
QA_PREBUILT="usr/lib/python3.*/site-packages/runai_model_streamer/libstreamer/*"
src_unpack() {
mkdir -p "${S}/wheel" || die
cp "${DISTDIR}/${MY_WHEEL}" "${S}/wheel/" || die
}
src_install() {
python_setup
${EPYTHON} -m installer --destdir="${D}" "${S}/wheel/${MY_WHEEL}" || die
}
>=sci-ml/pytorch-2.0.0[${PYTHON_SINGLE_USEDEP}] $(python_gen_cond_dep ' dev-python/humanize[${PYTHON_USEDEP}] dev-python/numpy[${PYTHON_USEDEP}] ')