CUDA BFloat16 Refactor #10085
Use BFLoat16 instead of nv_bfloat16 for CUDA
2b3f8f7d
fix BFloat16 rocm
8815c589
remove gemm ut from rocm
5467e8e4
remove reducesum ut from rocm
2cfec118
add biasgelu
2b8868ee
add support for more ops
7dc37da4
Merge branch 'master' into weicwang/bfloat16
d071935a
fix build error
a3c394af
op md update
882cb356
Merge branch 'master' into weicwang/bfloat16
8c3240dc
Lafi7e
merged
44e2db93
into master 4 years ago
Lafi7e
deleted the weicwang/bfloat16 branch 4 years ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub