TensorComprehensions
A domain-specific language and C++ library for automatically synthesizing high-performance machine learning kernels.
What is TensorComprehensions?
Tensor Comprehensions is a domain-specific language and C++ library for automatically generating high-performance machine learning kernels. It allows researchers and engineers to express tensor operations at a high level and automatically synthesizes optimized GPU or CPU implementations, eliminating the need for manual kernel programming.
Machine learning researchers and engineers working on performance-critical model components, particularly those needing to deploy custom operations across different hardware platforms.
It dramatically reduces the time and expertise required to create optimized kernels, provides framework-agnostic portability, and achieves near-peak hardware performance through automated autotuning and evolutionary search.
Overview
A domain specific language to express machine learning workloads.
Use Cases
Best For
- Automatically optimizing custom tensor operations for GPU acceleration
- Prototyping new ML layers without writing low-level CUDA code
Related Projects
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
