Precompiled Python Wheels for CUDA and PyTorch Combinations

A collection of prebuilt Python wheels for specific CUDA and PyTorch versions that are described as awkward to build locally. The dataset was authored by tori29umai and was last updated on May 14, 2026. It includes files for platforms such as Windows x86_64, following a naming convention that encodes package version, CUDA version, PyTorch version, Python tag, ABI tag, and platform.

Use Cases

• Deploying deep learning libraries without local compilation based on the prebuilt wheel files.
• Setting up reproducible development environments based on specific CUDA and PyTorch version combinations mentioned in the description.
• Testing software compatibility across different Python, CUDA, and PyTorch versions based on the structured naming convention.

Strengths

• Files follow a structured naming convention that encodes key dependencies like CUDA version, PyTorch version, and Python version.
• The dataset is hosted on Hugging Face, a platform commonly used by the AI/ML community for sharing resources.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Collection Method

Likely compiled and uploaded by the author for distribution.

Freshness

Last updated 2026-05-14 14:50:10; freshness should be verified.