#machine-learning #tensor

candle-cublaslt

CUBLASLt gemm for the candle ML framework

1 unstable release

0.0.1 Apr 15, 2025

#171 in #tensor

Download history 88/week @ 2025-04-28 41/week @ 2025-05-05 71/week @ 2025-05-12 42/week @ 2025-05-19 44/week @ 2025-05-26 134/week @ 2025-06-02 175/week @ 2025-06-09 89/week @ 2025-06-16 127/week @ 2025-06-23 129/week @ 2025-06-30 104/week @ 2025-07-07 97/week @ 2025-07-14 153/week @ 2025-07-21 65/week @ 2025-07-28 97/week @ 2025-08-04 134/week @ 2025-08-11

457 downloads per month

MIT/Apache

31KB
752 lines

Candle CublasLt Matmul Layer

CublasLt Matmul operation for the Candle ML framework. Allows for bias and Relu/Gelu fusing.

Dependencies

~16–26MB
~411K SLoC