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.
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
[DRAF][one-optimize] Optimize part of the transformer's attention-head #12918
base: master
Are you sure you want to change the base?
Conversation
This draft introduces two new passes to optimize part of the transformer's attention-head: FuseMulWithFullyConnected and FuseStridedSlicesNegAsMul. ONE-DCO-1.0-Signed-off-by: Artem Balyshev <a.balyshev@samsung.com>
75b73de
to
8974450
Compare
|
some comments;
|
faf419e
to
fd5b254
Compare
Done |
@seanshpark, can I split this draft into PRs? |
Yes, please do :) |
input: "output_1" | ||
input: "output_neg" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In the target model, the inputs are swapped. PTAL #12917 (comment)
This draft introduces two new passes to optimize part of the transformer's attention-head: FuseMulWithFullyConnected and FuseStridedSlicesNegAsMul.
for issue: #12917
ONE-DCO-1.0-Signed-off-by: Artem Balyshev a.balyshev@samsung.com