Install this version:
emerge -a =sci-ml/lightning-2.6.5
If this version is masked, you can unmask it using the autounmask tool or standard emerge options:
autounmask =sci-ml/lightning-2.6.5
Or alternatively:
emerge --autounmask-write -a =sci-ml/lightning-2.6.5
| Version | EAPI | Keywords | Slot |
|---|---|---|---|
| 2.6.5 | 8 | ~amd64 | 0 |
# Copyright 1999-2026 Gentoo Authors
# Distributed under the terms of the GNU General Public License v2
EAPI=8
DISTUTILS_USE_PEP517=setuptools
DISTUTILS_SINGLE_IMPL=1
PYTHON_COMPAT=( python3_{12..14} )
inherit distutils-r1 pypi
DESCRIPTION="The deep learning framework to pretrain and finetune AI models"
HOMEPAGE="
https://lightning.ai/
https://github.com/Lightning-AI/pytorch-lightning
https://pypi.org/project/lightning/
"
LICENSE="Apache-2.0"
SLOT="0"
KEYWORDS="~amd64"
# The lightning wheel ships its own self-contained code under the
# lightning/ namespace (lightning.fabric, lightning.pytorch, lightning.data).
# Upstream's Requires-Dist lists pytorch-lightning as a runtime dep, but
# that's a co-installable redundancy: pytorch-lightning ships the same code
# under the legacy pytorch_lightning + lightning_fabric top-level names.
# Downstream consumers in this overlay (sci-ml/pyannote-audio) only use the
# `lightning.*` umbrella import path, so we deliberately omit it. Users
# needing `import pytorch_lightning` can pip-install it separately.
RDEPEND="
sci-ml/pytorch[${PYTHON_SINGLE_USEDEP}]
>=sci-ml/torchmetrics-0.7.1[${PYTHON_SINGLE_USEDEP}]
$(python_gen_cond_dep '
>=dev-python/pyyaml-5.4[${PYTHON_USEDEP}]
>=dev-python/fsspec-2022.5.0[${PYTHON_USEDEP}]
dev-python/aiohttp[${PYTHON_USEDEP}]
dev-python/requests[${PYTHON_USEDEP}]
>=sci-ml/lightning-utilities-0.10.0[${PYTHON_USEDEP}]
>=dev-python/packaging-23.0[${PYTHON_USEDEP}]
>=dev-python/tqdm-4.57.0[${PYTHON_USEDEP}]
>=dev-python/typing-extensions-4.5.0[${PYTHON_USEDEP}]
')
"
# Tests pull a long tail (deepspeed, hydra, jsonargparse, ...).
RESTRICT="test"
src_prepare() {
# Lightning's setup.py reads PACKAGE_NAME from env to choose between the
# `lightning` umbrella and `pytorch-lightning` build artifacts. The sdist
# defaults to building `lightning`, but make it explicit.
export PACKAGE_NAME="lightning"
distutils-r1_src_prepare
}
sci-ml/pytorch[${PYTHON_SINGLE_USEDEP}] >=sci-ml/torchmetrics-0.7.1[${PYTHON_SINGLE_USEDEP}] $(python_gen_cond_dep ' >=dev-python/pyyaml-5.4[${PYTHON_USEDEP}] >=dev-python/fsspec-2022.5.0[${PYTHON_USEDEP}] dev-python/aiohttp[${PYTHON_USEDEP}] dev-python/requests[${PYTHON_USEDEP}] >=sci-ml/lightning-utilities-0.10.0[${PYTHON_USEDEP}] >=dev-python/packaging-23.0[${PYTHON_USEDEP}] >=dev-python/tqdm-4.57.0[${PYTHON_USEDEP}] >=dev-python/typing-extensions-4.5.0[${PYTHON_USEDEP}] ')