TenSure

Sparse Tensor Compiler Fuzzer

TenSure (Mahathevan et al., 2026) is a modular fuzzing framework designed to uncover correctness bugs in Sparse Tensor Compilers (STCs). It generates randomized sparse tensor kernels, executes them across compiler backends, and detects behavioral inconsistencies. TenSure is backend-agnostic and supports dynamic loading of multiple STC implementations.

More details can be found in the GitHub repository.

References

2026

  1. TENSURE: Fuzzing Sparse Tensor Compilers
    Kabilan Mahathevan, Yining Zhang, Muhammad Ali Gulzar, and Kirshanthan Sundararajah
    In Fuzzing Workshop (FUZZING) 2026, Network and Distributed System Security (NDSS). More Information can be found here , 2026