Fix MAXIMUM_PROTOBUF size (2GiB, not 2GB)
#6556
Merged
Conversation
This file contains hidden or 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
While saving a model that was just barely under the 2GiB limit (using Optimum), I noticed that it was saved with the external data format, even though it should be able to be . After further investigation, it was related to https://github.com/huggingface/optimum/blob/65a8a94adaf136dd677d28cfc837c0acfe993031/optimum/onnx/graph_transformations.py#L135-L138, where the value of `onnx.MAXIMUM_PROTOBUF` is used to determine whether to use the external data format. This value is hardcoded to 2GB (=2*10^9 bytes), even though it should be 2GiB (=2*1024^3), according to the https://protobuf.dev/programming-guides/proto-limits/#:~:text=form%20must%20be-,%3C2GiB%2C,-as%20that%20is: Signed-off-by: Joshua Lochner <admin@xenova.com>
justinchuby
approved these changes
Nov 26, 2024
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #6556 +/- ##
=======================================
Coverage 57.39% 57.39%
=======================================
Files 507 507
Lines 31528 31528
Branches 3549 3549
=======================================
Hits 18096 18096
Misses 12583 12583
Partials 849 849 ☔ View full report in Codecov by Sentry. |
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Outdated suggestions cannot be applied.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.
Description
While saving a model that was just barely under the 2GiB limit (using Optimum), I noticed that it was saved with the external data format, even though it should be able to be saved in a single file.
After further investigation, it was related to these lines, where the value of
.
onnx.MAXIMUM_PROTOBUFis used to determine whether to use the external data format. This value is hardcoded to 2GB (=2*10^9bytes), even though it should be 2GiB (=2*1024^3), according to the protobuf documentation:Unless there is a specific onnx specification which limits this to 2GB (and not 2GiB), this PR may potentially fix a problem which has existed for a long time.
Motivation and Context
Allows certain models to be saved as a single
.onnxfile without the external data format, which were previously saved separately due to this misconfiguration.