Bump onnx to get bfloat16 in ops, and some update in ort to support bfloat16 #4791
bump onnx to support bfloat16
4975edcd
sign test code
646b1246
fix ut failures
50728d43
add bfloat type in gradient schema
2e975210
add bfloat16 to gathernd
b6df94dc
add bfloat16 into grad op defs
e2e780f4
Merge branch 'master' into chenta/bfloat16
a176c299
temp disable gpu fusing transformers
b627bc90
bfloat16 support fix
5bec4927
more fix to bfloat
0f4427af
bug ifx
b89da45d
add bfloat16 to transpose matmul
f86e764f
fix sce loss
15ae3101
fix cast opset13 and other missing part of bfloat16
7a02aad1
Revert "temp disable gpu fusing transformers"
461e387d
add SCEloss back
23216add
merge from master
7a57e7a1
fix build break
d04cab89
souptc
requested a review
5 years ago
fix gpu failure due to missing kernel in opset13
661d63d0
add tile opset 13 kernel
8fd56b6d
Revert "fix gpu failure due to missing kernel in opset13"
acbcbba5
Lafi7e
commented
on 2020-08-14
fix comments in pr
fb45afdf
fix cuda break due to opset13
20dfe9c5
fix missing msdomain
8e662d88
add nll loss tests into android build's broken list; disable bfloat16…
2208833e
Lafi7e
approved these changes
on 2020-08-17
souptc
merged
1b1a6a4c
into master 5 years ago
souptc
deleted the chenta/bfloat16_step1 branch 5 years ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub