Raising of the recursions limit for onnx model loading. (Issue #5585) #5796
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
…m in the OnnxTransformer (as the default value in the Google.Protobuf). Otherwise some models cannot be loaded (ex. TF2 Efficentdet).
Codecov Report
@@ Coverage Diff @@
## main #5796 +/- ##
=======================================
Coverage 68.32% 68.32%
=======================================
Files 1131 1131
Lines 241291 241291
Branches 25053 25053
=======================================
+ Hits 164863 164872 +9
+ Misses 69923 69918 -5
+ Partials 6505 6501 -4
Flags with carried forward coverage won't be shown. Click here to find out more.
|
* updated arcade to the latest version * updated eng/common correctly * Fixed benchmark test.
* Use dotnet certificate * Update 3.1 SDK Co-authored-by: Prashanth Govindarajan <[email protected]> Co-authored-by: Michael Sharp <[email protected]>
* arm testing * initial commit with build working on arm64 * windows changes * build fixes for arm/arm64 with cross compilation * cross build instructions added * renamed arm to Arm. Changed TargetArchitecture to default to OS architecture * fixed some formatting * fixed capitilization * fixed Arm Capitilization * Fix cross-compilation if statement * building on apple silicon * removed non build related files * Changes from PR comments. Removal of FastTreeNative flag. * Changes from pr comments. * Fixes from PR comments. * Changed how we are excluding files.
* fixed onnx temp model deleting * random file path fixed * updates from pr * Changes from PR comments. * Changed how auto ml caches. * PR fixes. * Update src/Microsoft.ML.AutoML/API/ExperimentSettings.cs Co-authored-by: Eric Erhardt <[email protected]> * Tensorflow fixes from PR comments * fixed filepath issues Co-authored-by: Eric Erhardt <[email protected]>
|
I've just merged the latest 4 official main branch's commits to resolve the conflict on OnnxUtils.cs and make more easy the integration. |
|
I just resolved the conflict. It looks good to me. I think eventually we will need to expose that value so it can be changed as needed, but setting our default to match google is a good idea. I'll go ahead and approve/merge when the tests pass. Thanks so much for submitting this @darth-vader-lg |
|
Hello @michaelgsharp |
michaelgsharp
approved these changes
May 27, 2021
ghost
locked as resolved and limited conversation to collaborators
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
Raised the limit of recursions (from 10 to 100) in the creation of the CodedInputStream in the OnnxTransformer (as the default value in the Google.Protobuf). Otherwise some models cannot be loaded (ex. TF2 Efficientdet).
Fixes #5585
I have already tested it on all TF2 ModelZoo models converted to onnx and on my custom object detection models. It's all right now.
saved_model.onnx.zip