/blocksparse

Efficient GPU kernels for block-sparse matrix multiplication and convolution

Primary LanguageCudaMIT LicenseMIT

Watchers