-
Notifications
You must be signed in to change notification settings - Fork 54
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
[skip ci] ViT TTNN Tech Report #12800
Merged
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mbahnasTT
requested review from
mywoodstock,
davorchap,
tt-aho,
TT-BrianLiu and
bbradelTT
September 18, 2024 01:23
mbahnasTT
force-pushed
the
mbahnas/vit_tech_report
branch
from
September 18, 2024 03:25
9552a04
to
aa0e0a1
Compare
mbahnasTT
force-pushed
the
mbahnas/vit_tech_report
branch
from
September 18, 2024 17:05
3ba1089
to
666080c
Compare
bbradelTT
reviewed
Sep 18, 2024
tech_reports/ViT-TTNN/vit.md
Outdated
|
||
` seqL × head_count × head_size` | ||
|
||
This step aggregates the outputs from the different heads into a single vector representation for each position in the sequence. The followin step is the Linear OP to calculate the self output, which is the output of the self multi-head attention module. |
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.
followin -> following
bbradelTT
reviewed
Sep 18, 2024
tech_reports/ViT-TTNN/vit.md
Outdated
``` | ||
|
||
#### 4.4.1 Q,K,V Generation using the Fused Linear OP | ||
The encoder input is matrix-mutiplied by the Q,K,V weights to generate the individual Query, Key, Value tensors. In the TT-NN implementation, the input is multipled by the pre-fused weights to generate the merged 3 tensors that will be split in a following step. The fused linear operation obective is to maximize the utilization by increasing the workload that is computed simultaneously on the Tensic core grid. |
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.
fix: multipled, obective
mbahnasTT
force-pushed
the
mbahnas/vit_tech_report
branch
from
September 22, 2024 20:18
3a6811d
to
c37acc1
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Ticket
Link to Github Issue
Problem description
Provide context for the problem.
What's changed
Describe the approach used to solve the problem.
Summarize the changes made and its impact.
Checklist