onnxruntime
ZCode FastFormers changes
#5827
Merged

ZCode FastFormers changes #5827

ykim362
ykim362 Add FBGEMM submodule
6ea91f12
ykim362 Add fbgemm based per-channel quantization
7461ea3c
ykim362 Add missing logic for pre-layernorm transformer model fusion
641b2dbd
ykim362 add support for structured pruning architecture -fastformers
33ab92e7
ykim362 Fix windows build
81f22d1d
ykim362 Add a default behavior when head_size is not present for the backward…
1f9234db
ykim362 Remove FBGEMM and default to tensor-wise quantization, column-wise qu…
6accff94
ykim362 Fixed some unit test errors
55eb2e18
ykim362 Fix windows compile error and unit test errors
828d6213
ykim362 Merge branch 'master' into youki/pr-revise
30686f8c
ykim362 delete the option removed from the upstream
49a86943
ykim362
ykim362 ykim362 marked this pull request as ready for review 5 years ago
ykim362 ykim362 requested a review 5 years ago
yufenglee
yufenglee commented on 2020-12-15
yufenglee
yufenglee commented on 2020-12-15
yufenglee
yufenglee commented on 2020-12-15
yufenglee
yufenglee commented on 2020-12-15
yufenglee
yufenglee commented on 2020-12-15
ykim362 Addresses review comments and fixes a merge error
4cf7ee4b
yufenglee
yufenglee commented on 2020-12-17
yufenglee
yufenglee commented on 2020-12-17
yufenglee yufenglee requested a review from wangyems wangyems 5 years ago
yufenglee
wangyems
wangyems commented on 2020-12-17
wangyems
wangyems commented on 2020-12-17
wangyems
wangyems commented on 2020-12-17
wangyems
wangyems commented on 2020-12-17
ykim362
wangyems
wangyems commented on 2021-01-25
ykim362 Remove commented out code
38abb2eb
ykim362 Merge branch 'master' into youki/pr-revise
77fa341a
yufenglee
yufenglee commented on 2021-01-26
yufenglee
yufenglee commented on 2021-01-26
yufenglee
yufenglee commented on 2021-01-26
yufenglee add non-zero zp support
335e3514
yufenglee Merge branch 'master' into yufeng/matmul_non_zp
bb515c93
yufenglee support A and B scale with any dimensions
8b395219
yufenglee fix build breaks
49f6edab
yufenglee fix warning in MSVC
745c6371
zhanghuanrong Fix bug for not checking original float value names when treat it as …
264531b9
ykim362 Merge commit 'a647da3e1acb5ecb750c19df6b1b85af4572237d' into youki/pr…
35c10d8a
ykim362 Clean up head size
e3dc7f0d
ykim362 Clean up python tools
7f8f57fd
ykim362 Enable per column quantization
1d40fa8d
ykim362 Merge branch 'yufeng/matmul_non_zp' into youki/percol-test
ae81e733
ykim362 Merge branch 'zhalei/checking_origin_fpvalue_before_dequantize' into …
e8d11aa7
yufenglee fix quant weight cleanup bug
bc9e8491
ykim362 Merge branch 'yufeng/quant_tool_weight' into youki/percol-test
b624e617
ykim362 A few code clean up
21c86619
ykim362 Merge branch 'master' into youki/percol-test
6b85a9e3
ykim362 Some code clean-up
e9003026
ykim362 Merge branch 'master' into youki/fastformers-pr
71b44d46
yufenglee
azure-pipelines
yufenglee
yufenglee
azure-pipelines
azure-pipelines
yufenglee
yufenglee commented on 2021-05-17
ykim362 Some code clean-up
c1f128c2
ykim362 Merge branch 'master' into youki/fastformers-pr
03785f13
ykim362 Change option name
950b9c3a
ykim362 update default value
2895a273
yufenglee
yufenglee commented on 2021-05-17
yufenglee
yufenglee commented on 2021-05-17
yufenglee
yufenglee commented on 2021-05-17
yufenglee
yufenglee commented on 2021-05-17
yufenglee
yufenglee commented on 2021-05-17
ykim362 Rename option and parameter names
d4494600
ykim362 Missing argument name change
b29d03d5
yufenglee
yufenglee
yufenglee
azure-pipelines
azure-pipelines
azure-pipelines
yufenglee yufenglee added release:1.8
ykim362 Add tests for quantization options for attention and matmul
5232d33c
yufenglee
yufenglee approved these changes on 2021-05-17
yufenglee
yufenglee
yufenglee
azure-pipelines
azure-pipelines
azure-pipelines
yufenglee yufenglee merged e9057d2e into master 4 years ago
yufenglee
ykim362
xzhu1900 xzhu1900 removed release:1.8

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone