onnxruntime
Fuse softmax(a + b) in case of simple broadcast
#4937
Merged

Fuse softmax(a + b) in case of simple broadcast #4937

suffiank merged 22 commits into master from sukha/biassoftmax
suffiank
bias softmax kernel
881f4ae9
bias softmax kernel
c2253adc
remove debug comments
3e3808c3
remove debug comment
39a5e541
suffiank suffiank requested a review from edgchen1 edgchen1 5 years ago
suffiank suffiank requested a review 5 years ago
suffiank suffiank added training
windows build doesnt handle unary minus on unsigned type
14fc7207
int64 => int treated as error
a06bb4b9
only support cuda
4a16dd77
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
edgchen1
edgchen1 commented on 2020-09-02
yuslepukhin
yuslepukhin commented on 2020-09-02
yuslepukhin
yuslepukhin commented on 2020-09-02
yuslepukhin
yuslepukhin commented on 2020-09-02
add bias softmax fusion tests
3ccc6316
PR comments
178e2634
more PR comments
22ef2d9c
use MLTypeCallDispatcher
c04a406a
break function into pieces
94a31cbc
add loop unroll and add to list for inference as well
84eab378
merge with master
e37df5d4
use std::min and move operator==
083ade30
revert std::min (doesnt work ci pipeline) and fix int to size_t error
7e38b1e8
edgchen1
edgchen1 commented on 2020-09-17
edgchen1
edgchen1 commented on 2020-09-17
edgchen1
edgchen1 commented on 2020-09-17
pr comments
f9cb3059
fixes for windows ci
2bdc4373
edgchen1
edgchen1 commented on 2020-09-18
edgchen1
edgchen1 commented on 2020-09-18
edgchen1
edgchen1 commented on 2020-09-18
edgchen1
edgchen1 commented on 2020-09-18
edgchen1
edgchen1 commented on 2020-09-18
edgchen1
edgchen1 commented on 2020-09-18
edgchen1
edgchen1 commented on 2020-09-18
edgchen1
edgchen1 commented on 2020-09-18
fix for windows ci
3c93639a
pr comments on consistency
b419a072
p_model_
7b7377df
edgchen1
edgchen1 commented on 2020-09-18
edgchen1
edgchen1 commented on 2020-09-18
fix formatting and add anonymous namespace
c02f8a90
edgchen1
edgchen1 approved these changes on 2020-09-18
suffiank suffiank merged 84589c7e into master 5 years ago
suffiank suffiank deleted the sukha/biassoftmax branch 5 years ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone