-
Notifications
You must be signed in to change notification settings - Fork 2.6k
[GPU] Fix accuracy issues for sdpa_micro #30698
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
[GPU] Fix accuracy issues for sdpa_micro #30698
Conversation
if (MSK_D2 == 1) { | ||
// Define mask dimensions for single Query dimension | ||
uint mask_m = MSK_D1; // num_heads | ||
uint mask_n = MSK_D3; // sequence_length |
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.
why num_heads? masks are usually [q, seqlen]..
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.
Yes, you're right. I fixed codes. please review it again. thanks.
c127bc8
to
03475fe
Compare
src/plugins/intel_gpu/src/kernel_selector/cl_kernels/sdpa_micro.cl
Outdated
Show resolved
Hide resolved
@ahnyoung-paul please add some test case for this scenario |
I'm making unit test for this case. I will update it soon. |
### Details: - *Fix attn mask tile loading issue when attn mask shape has single query dimension [batch, num_heads, 1, sequence_length of k]* ### Tickets: - *166691*
Details:
Tickets: