Skip to content

Move undelegated constants #13606

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
Merged
facebook-github-bot merged 1 commit into from
Aug 27, 2025
Merged

Move undelegated constants #13606

facebook-github-bot merged 1 commit into from
Aug 27, 2025

Conversation

@lucylq
Copy link
Contributor

@lucylq lucylq commented Aug 22, 2025 •
edited
Loading

Currently, the xnnpack + lora + program-data separation flow moves all delegated constants to the .ptd file. This change moves the undelegated constants as well, reducing the pte file size.

Run export_lora.sh from meta-pytorch/executorch-examples#54

-rw-r--r-- 1 lfq users 5994013600 Aug 22 13:32 foundation.ptd
-rw-r--r-- 1 lfq users   27628928 Aug 22 13:32 llama_3_2_1B_lora.pte
-rw-r--r-- 1 lfq users 5994013600 Aug 22 13:30 llama_3_2_1B.ptd
-rw-r--r-- 1 lfq users     317248 Aug 22 13:30 llama_3_2_1B.pte

fp32 Weights: ~6GB
Program: ~300KB
Lora adapter weights: ~27MB

@pytorch-bot
Copy link

pytorch-bot bot commented Aug 22, 2025 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/13606

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 New Failure

As of commit 168d750 with merge base 49bc664 (image):

NEW FAILURE - The following job has failed:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@meta-cla meta-cla bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Aug 22, 2025
@github-actions
Copy link

github-actions bot commented Aug 22, 2025

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

lucylq
lucylq commented Aug 22, 2025
View reviewed changes
exir/program/_program.py Outdated
# there exists an unbacked symint operation.
EdgeToBackendOpsPass(),
RemoveGraphAssertsPass(),
*config.passes,
Copy link
Contributor Author

@lucylq lucylq Aug 22, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

external_constants_pass has to run after SpecPropPass, which populates the tensor names. See if CI passes, not sure if this will break anything.

Copy link
Contributor

@JacobSzwejbka JacobSzwejbka Aug 25, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you do the more minimal change then and just move it after specProp rather then all of them then?

Copy link
Contributor Author

@lucylq lucylq Aug 25, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@JacobSzwejbka, updated. Do EdgeToBackendOpsPass/RemoveGraphAssertsPass have to be relatively late in the passes?

@facebook-github-bot
Copy link
Contributor

facebook-github-bot commented Aug 22, 2025

@lucylq has imported this pull request. If you are a Meta employee, you can view this in D80840528.

@lucylq lucylq requested a review from mergennachin August 22, 2025 20:40
@lucylq lucylq force-pushed the lfq.lora-constants branch from 33c43bf to ffa0a57 Compare August 22, 2025 20:58
@facebook-github-bot
Copy link
Contributor

facebook-github-bot commented Aug 22, 2025

@lucylq has imported this pull request. If you are a Meta employee, you can view this in D80840528.

@lucylq lucylq force-pushed the lfq.lora-constants branch from ffa0a57 to d69aa0f Compare August 22, 2025 21:16
@facebook-github-bot
Copy link
Contributor

facebook-github-bot commented Aug 22, 2025

@lucylq has imported this pull request. If you are a Meta employee, you can view this in D80840528.

@lucylq lucylq force-pushed the lfq.lora-constants branch from d69aa0f to f0f0e9e Compare August 25, 2025 17:54
@lucylq lucylq force-pushed the lfq.lora-constants branch from f0f0e9e to 168d750 Compare August 25, 2025 17:54
@facebook-github-bot
Copy link
Contributor

facebook-github-bot commented Aug 25, 2025

@lucylq has imported this pull request. If you are a Meta employee, you can view this in D80840528.

@facebook-github-bot facebook-github-bot merged commit 8b3d21c into main Aug 27, 2025
104 of 105 checks passed
@facebook-github-bot facebook-github-bot deleted the lfq.lora-constants branch August 27, 2025 00:16
@shewu-quic shewu-quic mentioned this pull request Sep 3, 2025
Merged
cccclai pushed a commit that referenced this pull request Sep 8, 2025
@shewu-quic
Qualcomm AI Engine Direct - Strengthen Unit Test Robustness (#13892)
bf22ab7 
Summary:
  - Fixed the seed for E2E model scripts in unit test
  - Resolved the bug during dump optrace
- The per-channel quant config of the bias for conv op should be derived
by activation and weight.
- Resolve the issue caused by [the
PR](#13606) that prevented the
spec from being correctly updated to the quantized type.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@mergennachin mergennachin mergennachin approved these changes

@JacobSzwejbka JacobSzwejbka Awaiting requested review from JacobSzwejbka JacobSzwejbka is a code owner

@larryliu0820 larryliu0820 Awaiting requested review from larryliu0820 larryliu0820 is a code owner

@jackzhxng jackzhxng Awaiting requested review from jackzhxng jackzhxng is a code owner

Assignees

No one assigned

Labels

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@lucylq @facebook-github-bot @mergennachin @JacobSzwejbka