Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Multi head attention #22143

Closed
wants to merge 83 commits into from
Closed

Multi head attention #22143

wants to merge 83 commits into from

Conversation

xhcao
Copy link
Contributor

@xhcao xhcao commented Sep 19, 2024

Description

Motivation and Context

fs-eire and others added 30 commits August 28, 2024 14:53
@fs-eire
[WIP] WebGPU EP initial commit
4037bd4
@fs-eire
update C-API
9c36250
@fs-eire
fix build break
3a0756d
@fs-eire
add an empty symbols.txt file
5199e98
@fs-eire
fix an error in doc
1c68dbd
@fs-eire
remove string_join.h in favor of absl::StrJoin
7db03de
@fs-eire
fix DLL copy
6a373c2
@fs-eire
update doc: require --skip_tests
ee42bba
@fs-eire
Merge remote-tracking branch 'origin/main' into fs-eire/webgpu-ep
5fac202
@fs-eire
update dawn version
3f46e5c
@fs-eire
disable Tint tests
9f61279
@fs-eire
fix one build break in Linux
6bb6335
@fs-eire
remove unused variables
d839dbc
@guschmue @fs-eire
make webgpu build on linux and known to most tools (microsoft#21937)
b70943d 
Co-authored-by: Yulong Wang <[email protected]>
@fs-eire
Merge remote-tracking branch 'origin/main' into fs-eire/webgpu-ep
c33ac2e
@fs-eire
revert type of ShaderVariable::rank_ to int
8437267
@fs-eire
output Impl() for variables
3caf032
@fs-eire
code formatting
84494c4
@fs-eire
better format of Uniform
aa70163
@fs-eire
revise document
d772db7
@fs-eire
more build fix for linux
6ef3dad
@fs-eire
apply formatter
a56f6c3
@fs-eire
simple test runner
12cd79d
@fs-eire
Program macros update - allow extend
14c8966
@fs-eire
fix BucketCacheManager
4fff35f
@fs-eire
add a method to get logger from ComputeContext
4fd8ad1
@fs-eire
add verbose log for cache key
3bd92ad
@fs-eire
revise suite test
6a1bbfe
@fs-eire
device lost handler
947aee1
@fs-eire
add '-a' and '-t' to test runner
99b2578
fs-eire and others added 26 commits September 6, 2024 00:01
@fs-eire
simplify cache hint interface.
cd961c3
@fs-eire
revise expand
ddc2fbb
@fs-eire
revise unary
e8be835
@fs-eire
Elu/Relu/LeakyRelu/ThresholdedRelu/Gelu
bd7d592
@fs-eire
Merge remote-tracking branch 'origin/main' into fs-eire/webgpu-ep
eecac18
@fs-eire
remove unused field in class Gelu
601e50f
@fs-eire
remove out-of-dated comments
8f36da2
@fs-eire
Clip
72ebd85
@fs-eire
fix rank in shader helper
a3244ae
@fs-eire
fix shader variable
5a2ae8c
@fs-eire
move components number from variable to program
aa54ff8
@fs-eire
mark components in cache key
969384d
@qjia7 @fs-eire
Add FastGelu op (microsoft#21991)
6b82486 
### Description
<!-- Describe your changes. -->



### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->

---------

Co-authored-by: Yulong Wang <[email protected]>
@fs-eire
use 'set/add' as prefix for some functions
2b3e7c2
@fs-eire
remove unnecessary cache hint for FastGelu
ef0d53b
@fs-eire
revise unary - expose consts in header
c4ca47f
@fs-eire
use path for header file
8806d57
@fs-eire
a few revises to the code (microsoft#22047)
0568e2b
@fs-eire
use OrtMutex
b7a9c0e
@fs-eire
Merge remote-tracking branch 'origin/main' into fs-eire/webgpu-ep
f65ade9
@axinging @fs-eire
[webgpu-native] Add transpose op (microsoft#21986)
d4a963d 
Co-authored-by: Yulong Wang <[email protected]>
@fs-eire
PushErrorScope and PopErrorScope
8b61532
@fs-eire
placeholder for setting proc table
dce0f18
@fs-eire
Revert "placeholder for setting proc table"
8978d89 
This reverts commit dce0f18.
@fs-eire
allow setting "ValidationMode"
43ccaf4
@xhcao
webgpu: support MultiHeadAttention operator
409ac5c
@xhcao
Copy link
Contributor Author

xhcao commented Sep 19, 2024

Although I think there are some unreasonable codes in JS MultiHeadAttention operator, I still kept the code in webgpu EP nearly the same as JS EP. Let's adjust and optimize the code in future.
@fs-eire @qjia7

@xhcao
Copy link
Contributor Author

xhcao commented Sep 19, 2024

Make a mistake here, close it. see #22144

@xhcao
Copy link
Contributor Author

xhcao commented Sep 19, 2024

Close

@xhcao xhcao closed this Sep 19, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
5 participants
@xhcao @fs-eire @guschmue @qjia7 @axinging