The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
Pre-built CUDA Wheels: Causal-Conv1d & Flash-Linear-Attention
This dataset is a collection of pre-built Python wheels for ninja, flash-linear-attention, and causal-conv1d specifically compiled for modern NVIDIA GPUs. It is organized to help developers bypass the highly resource-intensive C++ compilation times typically encountered in CI/CD pipelines (like GitHub Actions) and Docker builds.
Dataset Details
Uses
This dataset is open-sourced and available for the general public, researchers, and companies to use. Primary Use Case: Drop-in replacements for pip install during cloud deployments or CI/CD workflows.
Dataset Structure
The dataset contains .whl files categorized by their target environment and hardware architecture.
casual-conv1d-pre-built-wheels/
├──.gitattributes
├── README.md
└── L40S/
│ ├──x_86_64
│ │ ├── torch-2.11-cuda-13.0
│ │ │ ├──.whl
│ │ │ ├──.whl
Note: While ninja and flash-linear-attention are largely environment-agnostic, causal-conv1d is strictly tied to the Python version and Linux architecture it was compiled on.
Dataset Creation
Curation Rationale
Building and compiling C++ code to resolve dependencies like Casual Conv 1D can be expensive and time consuming. This dataset aims to mitigate that cost by providing a wide range of up-to-date pre-built wheels.
Who are the source data producers?
The official github repositories for the libraries this code compiles are:
- https://github.com/Dao-AILab/causal-conv1d
- https://github.com/ninja-build/ninja
- https://github.com/fla-org/flash-linear-attention
Bias, Risks, and Limitations
Compatibility Restrictions: Pre-compiled CUDA wheels are notoriously rigid. If your deployment environment does not closely match the build environment (e.g., you are using Python 3.10 instead of 3.12, or CUDA 12.1 instead of 13.0), pip will likely reject the wheel, or PyTorch will throw a runtime CUDA mismatch error.
Maintenance: While I will try to update this dataset as much as possible, I cannot guarantee that all GPUs, operating systems, PyTorch iterations, and CUDA versions will be actively supported. Users are encouraged to verify the cpXXX (Python version) and cuXXX (CUDA version) tags in the filenames before integrating them into production pipelines.
- Downloads last month
- 99