Skip to content

ProduceWordBags Onnx Export Fix #5435

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

Merged
kere-nel merged 5 commits into from
Oct 15, 2020
Merged

ProduceWordBags Onnx Export Fix #5435

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
4ed1b82
fix for issue
Oct 13, 2020
40c9b39
fix documentation
Oct 13, 2020
8571d49
aligning test
Oct 14, 2020
b4ebafd
adding back line
Oct 14, 2020
4497de8
aligning fix
Oct 15, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions src/Microsoft.ML.Transforms/Text/TextCatalog.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -334,7 +334,7 @@ public static CustomStopWordsRemovingEstimator RemoveStopWords(this TransformsCa
=> new CustomStopWordsRemovingEstimator(Contracts.CheckRef(catalog, nameof(catalog)).GetEnvironment(), outputColumnName, inputColumnName, stopwords);

/// <summary>
/// Create a <see cref="WordHashBagEstimator"/>, which maps the column specified in <paramref name="inputColumnName"/>
/// Create a <see cref="WordBagEstimator"/>, which maps the column specified in <paramref name="inputColumnName"/>
/// to a vector of n-gram counts in a new column named <paramref name="outputColumnName"/>.
/// </summary>
/// <remarks>
Expand Down Expand Up @@ -363,7 +363,7 @@ public static WordBagEstimator ProduceWordBags(this TransformsCatalog.TextTransf
outputColumnName, inputColumnName, ngramLength, skipLength, useAllLengths, maximumNgramsCount, weighting);

/// <summary>
/// Create a <see cref="WordHashBagEstimator"/>, which maps the multiple columns specified in <paramref name="inputColumnNames"/>
/// Create a <see cref="WordBagEstimator"/>, which maps the multiple columns specified in <paramref name="inputColumnNames"/>
/// to a vector of n-gram counts in a new column named <paramref name="outputColumnName"/>.
/// </summary>
/// <remarks>
Expand Down
8 changes: 4 additions & 4 deletions src/Microsoft.ML.Transforms/Text/WordTokenizing.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -415,10 +415,10 @@ public void SaveAsOnnx(OnnxContext ctx)
string[] separators = column.SeparatorsArray.Select(c => c.ToString()).ToArray();
tokenizerNode.AddAttribute("separators", separators);

opType = "Squeeze";
var squeezeOutput = ctx.AddIntermediateVariable(_type, column.Name);
var squeezeNode = ctx.CreateNode(opType, intermediateVar, squeezeOutput, ctx.GetNodeName(opType), "");
squeezeNode.AddAttribute("axes", new long[] { 1 });
opType = "Reshape";
var shape = ctx.AddInitializer(new long[] { 1, -1 }, new long[] { 2 }, "Shape");
var reshapeOutput = ctx.AddIntermediateVariable(new VectorDataViewType(TextDataViewType.Instance, 1), column.Name);
var reshapeNode = ctx.CreateNode(opType, new[] { intermediateVar, shape }, new[] { reshapeOutput }, ctx.GetNodeName(opType), "");
}
}
}
Expand Down
21 changes: 12 additions & 9 deletions test/Microsoft.ML.Tests/OnnxConversionTest.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1310,11 +1310,11 @@ public void NgramOnnxConversionTest(
IEstimator<ITransformer>[] pipelines =
{
mlContext.Transforms.Text.TokenizeIntoWords("Tokens", "Text", new[] { ' ' })
.Append(mlContext.Transforms.Conversion.MapValueToKey("Tokens"))
.Append(mlContext.Transforms.Text.ProduceNgrams("NGrams", "Tokens",
ngramLength: ngramLength,
useAllLengths: useAllLength,
weighting: weighting)),
.Append(mlContext.Transforms.Conversion.MapValueToKey("Tokens"))

@harishsk harishsk Oct 14, 2020

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Any particular reason the formatting was changed? Usually there is a small offset to indicate line continuation.

@kere-nel kere-nel Oct 14, 2020

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

No, I noticed that line 1321 was not aligned with the rest. I'll add the offsets back though and just align that line.

@harishsk harishsk Oct 14, 2020

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

nit: It looks like the offset for line 1313 is still wrong. It seems to be left aligned with the previous line.

.Append(mlContext.Transforms.Text.ProduceNgrams("NGrams", "Tokens",
ngramLength: ngramLength,
useAllLengths: useAllLength,
weighting: weighting)),

mlContext.Transforms.Text.TokenizeIntoCharactersAsKeys("Tokens", "Text")
.Append(mlContext.Transforms.Text.ProduceNgrams("NGrams", "Tokens",
Expand All @@ -1323,9 +1323,12 @@ public void NgramOnnxConversionTest(
weighting: weighting)),

mlContext.Transforms.Text.ProduceWordBags("Tokens", "Text",
ngramLength: ngramLength,
useAllLengths: useAllLength,
weighting: weighting)
ngramLength: ngramLength,
useAllLengths: useAllLength,
weighting: weighting),

mlContext.Transforms.Text.TokenizeIntoWords("Tokens0", "Text")
.Append(mlContext.Transforms.Text.ProduceWordBags("Tokens", "Tokens0"))
};

for (int i = 0; i < pipelines.Length; i++)
Expand All @@ -1346,7 +1349,7 @@ public void NgramOnnxConversionTest(
var onnxEstimator = mlContext.Transforms.ApplyOnnxModel(onnxFilePath, gpuDeviceId: _gpuDeviceId, fallbackToCpu: _fallbackToCpu);
var onnxTransformer = onnxEstimator.Fit(dataView);
var onnxResult = onnxTransformer.Transform(dataView);
var columnName = i == pipelines.Length - 1 ? "Tokens" : "NGrams";
var columnName = i >= pipelines.Length - 2 ? "Tokens" : "NGrams";
CompareResults(columnName, columnName, transformedData, onnxResult, 3);

VBuffer<ReadOnlyMemory<char>> mlNetSlots = default;
Expand Down