Fixed bugs in OptionalColumnTransform and added bool support #4815
+352
−196
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
- Cleaned up column comparison functionality on OnnxConversionTest - Fixed bugs in OptionalColumnTransform's onnx export and added support for boolean initializers
ganik
reviewed
Feb 10, 2020
| @@ -139,16 +149,17 @@ public OnnxNode CreateNode(string opType, string input, string output, string na | |||
| public abstract string AddInitializer(float value, string name = null, bool makeUniqueName = true); | |||
|
|
|||
| /// <summary> | |||
| /// Call this function can declare a global long | |||
| /// Call this function to declare a global float | |||
There was a problem hiding this comment.
float [](start = 51, length = 5)
long? #Resolved
ganik
reviewed
Feb 10, 2020
| /// <param name="name">A string used as a seed to generate this initializer's name in the ONNX graph.</param> | ||
| /// <param name="makeUniqueName">Whether a unique name should be picked for this initializer.</param> | ||
| /// <returns>The initializer's ONNX name</returns> | ||
| public abstract string AddInitializer(IEnumerable<double> values, IEnumerable<long> dims, string name = null, bool makeUniqueName = true); | ||
|
|
||
| /// <summary> | ||
| /// Call this function can declare a global string tensor | ||
| /// Call this function to declare a global double tensor |
There was a problem hiding this comment.
double [](start = 51, length = 6)
ulong? #Resolved
Lynx1820
reviewed
Feb 10, 2020
| /// <param name="name">A string used as a seed to generate this initializer's name in the ONNX graph.</param> | ||
| /// <param name="makeUniqueName">Whether a unique name should be picked for this initializer.</param> | ||
| /// <returns>The initializer's ONNX name</returns> | ||
| public abstract string AddInitializer(IEnumerable<float> values, IEnumerable<long> dims, string name = null, bool makeUniqueName = true); | ||
|
|
||
| /// <summary> | ||
| /// Call this function can declare a global long tensor | ||
| /// Call this function to declare a global long tensor |
There was a problem hiding this comment.
This conflicts with the type param description: "Use this for adding array initializers of integer types smaller than Int32" #Resolved
| /// </summary> | ||
| /// <param name="value">The long number which is going to be added into the ONNX graph</param> | ||
| /// <param name="value">The float number which is going to be added</param> | ||
| /// <param name="type">The type of integer to be added, e.g. typeof(short). Use this for all integer types smaller than Int32</param> |
There was a problem hiding this comment.
Is the type inclusive of Int32s? #Resolved
Lynx1820
approved these changes
Feb 10, 2020
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.
OptionalColumnTransform didn't have support for boolean initializers. When I went to add that I saw that the TensorProto data was being set incorrectly for a few other data types. I tried to fix that and I had to redo the AddInitializer functionality in OnnxContext and related plumbing in OnnxUtils.cs.
Then when I went to add tests for all the data types that are now supported in OptionalColumnTransform, I realized I needed to simplify the column comparison in OnnxConversionTest. In the process I also added support SByte and Byte in OnnxUtils - CreateNamedValue and fixed up the corresponding tests.
In summary: