add load from buffer #10162
Merged
add load from buffer #10162
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
added 3 commits
November 30, 2021 07:33
added 4 commits
January 3, 2022 03:19
…o jingf/add-LoadFromBuffer
|
TODO: still need to add tests |
martinb35
requested changes
Jan 5, 2022
added 2 commits
January 5, 2022 11:22
martinb35
reviewed
Jan 5, 2022
| @@ -104,22 +104,42 @@ namespace ROOT_NS.AI.MachineLearning | |||
| //! Loads an ONNX model from a stream asynchronously. | |||
| [remote_async] | |||
| static Windows.Foundation.IAsyncOperation<LearningModel> LoadFromStreamAsync(Windows.Storage.Streams.IRandomAccessStreamReference modelStream); | |||
| //! Loads an ONNX model from a buffer asynchronously. | |||
There was a problem hiding this comment.
curious what you think about grouping all of these into one (MachineLearningContract, 6) block instead of having four of them?
There was a problem hiding this comment.
Yeah technically I don't have problem with that, just thought it might make more sense to keep them together within their own group
There was a problem hiding this comment.
Remember that this needs an associated change in the OS repo as well.
martinb35
approved these changes
Jan 7, 2022
smk2007
approved these changes
Jan 10, 2022
fdwr
reviewed
Jan 19, 2022
ancienticybeing
pushed a commit
that referenced
this pull request
Feb 17, 2022
This reverts commit 5cd57bb.
ancienticybeing
pushed a commit
that referenced
this pull request
Mar 8, 2022
lavanyax
pushed a commit
to intel/onnxruntime
that referenced
this pull request
Mar 29, 2022
This reverts commit 5cd57bb.
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Outdated suggestions cannot be applied.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: Add LoadFromBuffer API for LearningModel
Motivation and Context
Currently, to load a model for inferencing, developers are given 2 options:
However, neither LoadFromFilePath nor LoadFromStream method work well in VTL1 scenario. LoadFromFilePath would fail because VTL1 restricts file access to ensure a more secure trust level than the normal VTL0. LoadFromStream loads extra WinRT dependencies that are either not IUM signed or are not VTL1 registered COM components; on the other hand, LoadFromStream also makes several unnecessary model copies throughout the method call, heavily impacting performance when loading large models.
The proposed LoadFromBuffer method solves the above issues by not accessing the file system, avoiding loading unnecessary winrt dependencies, and making no extra model copies, making Windows Machine Learning one step closer to VTL1 ready