Adding cuda kernel (optimized for sm80) for block-wise 4b quantized float 16 GEMM. #18619
Azure Pipelines / orttraining-amd-gpu-ci-pipeline (Linux_Test_ubuntu)
failed
Feb 29, 2024 in 1h 2m 54s
Linux_Test_ubuntu failed
Annotations
Check failure on line 1303 in Build log
azure-pipelines / orttraining-amd-gpu-ci-pipeline (Linux_Test_ubuntu)
Build log #L1303
Bash exited with code '1'.
Loading