neuralmagic
/

Mistral-Nemo-Instruct-2407-FP8

7900xtx torch._scaled_mm is only supported on CUDA devices with compute capability >= 9.0 or 8.9, or ROCm MI300+

#3 opened about 2 months ago by