-
Notifications
You must be signed in to change notification settings - Fork 685
Fix quantized linear -> quantized fully connected replacement pass + add quantized fully connected per_tensor #6976
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
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/6976
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 75f1476 with merge base cd306d3 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This pull request was exported from Phabricator. Differential Revision: D66208417 |
@pytorchbot label "topic: not user facing" |
…add quantized fully connected per_tensor (#6976) Summary: As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector. Shaves 8.5k cyles from WW stage 1 model. Differential Revision: D66208417
b0eb57f
to
c04e166
Compare
This pull request was exported from Phabricator. Differential Revision: D66208417 |
…add quantized fully connected per_tensor (#6976) Summary: As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector. Shaves 8.5k cyles from WW stage 1 model. Differential Revision: D66208417
c04e166
to
42003c3
Compare
This pull request was exported from Phabricator. Differential Revision: D66208417 |
…add quantized fully connected per_tensor (#6976) Summary: As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector. Shaves 8.5k cyles from WW stage 1 model. Differential Revision: D66208417
42003c3
to
93a02cc
Compare
This pull request was exported from Phabricator. Differential Revision: D66208417 |
…add quantized fully connected per_tensor (#6976) Summary: As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector. Shaves 8.5k cyles from WW stage 1 model. Differential Revision: D66208417
93a02cc
to
c9992eb
Compare
This pull request was exported from Phabricator. Differential Revision: D66208417 |
…add quantized fully connected per_tensor (#6976) Summary: As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector. Shaves 8.5k cyles from WW stage 1 model. Differential Revision: D66208417
c9992eb
to
3829c66
Compare
This pull request was exported from Phabricator. Differential Revision: D66208417 |
unrelated lint error, please rebase and try |
…add quantized fully connected per_tensor (#6976) Summary: As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector. Shaves 8.5k cyles from WW stage 1 model. Reviewed By: zonglinpeng Differential Revision: D66208417
3829c66
to
27f0b82
Compare
This pull request was exported from Phabricator. Differential Revision: D66208417 |
27f0b82
to
b951ccf
Compare
…add quantized fully connected per_tensor (#6976) Summary: As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector. Shaves 8.5k cyles from WW stage 1 model. Reviewed By: zonglinpeng Differential Revision: D66208417
This pull request was exported from Phabricator. Differential Revision: D66208417 |
…add quantized fully connected per_tensor (#6976) Summary: As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector. Shaves 8.5k cyles from WW stage 1 model. Reviewed By: zonglinpeng Differential Revision: D66208417
b951ccf
to
3c8360f
Compare
This pull request was exported from Phabricator. Differential Revision: D66208417 |
…add quantized fully connected per_tensor (#6976) Summary: As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector. Shaves 8.5k cyles from WW stage 1 model. Reviewed By: zonglinpeng Differential Revision: D66208417
3c8360f
to
75f1476
Compare
This pull request was exported from Phabricator. Differential Revision: D66208417 |
Summary:
As titled. This allows removing the outer loop unrolling in cases where the input to linear is a vector.
Shaves ~10k cyles from WW stage 1 model.
Differential Revision: D66208417