deb_control_files:
- control
- md5sums
- shlibs
- triggers
deb_fields:
Architecture: amd64
Conflicts: libtorch-cuda-1.13, libtorch-cuda-2.0, libtorch-cuda-2.1, libtorch1.13,
libtorch2.0, libtorch2.1, libtorch2.4
Depends: libblas3 | libblas.so.3, libc6 (>= 2.38), libcpp-httplib0.16 (>= 0.16.3+ds),
libcpuinfo0 (>= 0.0~git20230113.6481e8b~), libcublas12, libcublaslt12, libcuda1
(>= 384) | libcuda.so.1 (>= 384), libcudart12, libcufft11, libcurand10, libcusolver11,
libcusparse12, libdnnl3 (>= 3.5.3), libfmt9 (>= 9.1.0+ds1), libgcc-s1 (>= 3.4),
libgloo-cuda-0 (>= 0.0~git20231202.5354032), libgomp1 (>= 4.9), liblapack3 | liblapack.so.3,
libnccl2 (>= 2.22.3), libnuma1 (>= 2.0.11), libnvrtc12, libnvtoolsext1 (>= 5.0),
libonnx1t64 (>= 1.16.2), libprotobuf32t64 (>= 3.21.12), libpthreadpool0 (>= 0.0~git20210507.1787867~),
libsleef3 (>= 3.6.1-1~), libstdc++6 (>= 12), libtensorpipe-cuda-0 (>= 0.0~git20220513.bb1473a),
libxnnpack0 (>= 0.0~git20240821.87ee0b4), nvidia-cudnn (>= 8.9.7.29~)
Description: |-
Tensors and Dynamic neural networks in Python (Shared Objects)
PyTorch is a Python package that provides two high-level features:
.
(1) Tensor computation (like NumPy) with strong GPU acceleration
(2) Deep neural networks built on a tape-based autograd system
.
You can reuse your favorite Python packages such as NumPy, SciPy and Cython
to extend PyTorch when needed.
.
This is the CUDA version of PyTorch (Shared Objects).
Homepage: https://pytorch.org/
Installed-Size: '533462'
Maintainer: Debian Deep Learning Team <debian-ai@lists.debian.org>
Multi-Arch: same
Package: libtorch-cuda-2.4
Priority: optional
Recommends: libopenblas0 | libblis3 | libatlas3-base | libmkl-rt | libblas3
Replaces: libtorch-cuda-1.13, libtorch-cuda-2.0, libtorch-cuda-2.1, libtorch1.13,
libtorch2.0, libtorch2.1, libtorch2.4
Section: contrib/libs
Source: pytorch-cuda
Version: 2.4.1-4
srcpkg_name: pytorch-cuda
srcpkg_version: 2.4.1-4