Introduce FP8 row-based quantization#194

Merged

dzmitry-huba merged 3 commits intomainfrom

huba/fp8_allreduce

May 15, 2025

Commits on May 15, 2025

Introduce FP8 row-based quantization
dzmitry-huba
committed
Address lint errors and make tests runnable when CUDA is enabled
dzmitry-huba
committed
Replace missing hardcoded FP8 type and ensure test is not running if Triton is not available
dzmitry-huba
committed