PyTorch is a Python package that provides two high-level features: - Tensor computation (like NumPy) with strong GPU acceleration - 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.
| Uploaded | Mon Mar 31 23:41:04 2025 |
| md5 checksum | c38ad8caa8f528103595be9897ccf6e9 |
| arch | x86_64 |
| build | gpu_cuda124_py310h8ea964d_202 |
| build_number | 202 |
| depends | __cuda, _openmp_mutex >=5.1, blas * openblas, cuda-cudart >=12.4.127,<13.0a0, cuda-cupti >=12.4.127,<13.0a0, cuda-nvrtc >=12.4.127,<13.0a0, cuda-nvtx >=12.4.127,<13.0a0, cuda-version >=12.4,<13, cudnn >=9.1,<10.0a0, cudnn >=9.1.1.17,<10.0a0, filelock, fsspec, jinja2, libabseil * cxx17*, libabseil >=20240116.2,<20240116.3.0a0, libcublas >=12.4.5.8,<13.0a0, libcufft >=11.2.1.3,<12.0a0, libcurand >=10.3.5.147,<11.0a0, libcusolver >=11.6.1.9,<12.0a0, libcusparse >=12.3.1.170,<13.0a0, libgcc-ng >=11.2.0, libopenblas >=0.3.21,<1.0a0, libprotobuf >=4.25.3,<4.25.4.0a0, libstdcxx-ng >=11.2.0, libtorch 2.5.1.*, libuv >=1.48.0,<2.0a0, magma >=2.7.1,<3.0a0, nccl >=2.21.5.1,<3.0a0, networkx, nomkl, numpy >=1.21,<3, numpy >=1.24.0,<3.0.0, opentelemetry-api, python >=3.10,<3.11.0a0, setuptools, sleef >=3.5.1,<4.0a0, sympy >=1.13.1,!=1.13.2, triton 3.1.0.*, typing_extensions |
| license | BSD-3-Clause |
| license_family | BSD |
| md5 | c38ad8caa8f528103595be9897ccf6e9 |
| name | pytorch |
| platform | linux |
| sha1 | 7ce07f556397c885d5331c8d6cb7f8f43724c32d |
| sha256 | b749b02172f5f837483e9ed73659dc123a0f8c4a98b227cc13160ce934594328 |
| size | 37214337 |
| subdir | linux-64 |
| timestamp | 1741715485344 |
| version | 2.5.1 |