Shuffle input cursor reader failed with an exception in ml.net 1.4.0

[Spocher]

• Windows 10 x64
• NET Framework 4.7.2, ML.Net 1.4.0

I create project for multiclass classification, some code:
`var trainingDataView = mlContext.Data.LoadFromEnumerable(allTrainData);

var trainingPipeline = mlContext.Transforms.Conversion.MapValueToKey(inputColumnName: "IdCategory", outputColumnName: "Label")
.Append(mlContext.Transforms.Text.FeaturizeText(inputColumnName: "HeaderProduct", outputColumnName: "HeaderProductFeaturized"))
.Append(mlContext.Transforms.Text.FeaturizeText(inputColumnName: "DescriptionProduct", outputColumnName: "DescriptionProductFeaturized"))
.Append(mlContext.Transforms.Concatenate("Features", "HeaderProductFeaturized", "DescriptionProductFeaturized"))
.Append(mlContext.MulticlassClassification.Trainers.SdcaMaximumEntropy())
.Append(mlContext.Transforms.Conversion.MapKeyToValue("PredictedLabel"));

var trainedModel = trainingPipeline.Fit(trainingDataView);`

I try train model and have Exception:
Shuffle input cursor reader failed with an exception, ( in Microsoft.ML.Transforms.RowShufflingTransformer.Cursor.MoveNextCore()
in Microsoft.ML.Data.RootCursorBase.MoveNext()
in Microsoft.ML.Trainers.TrainingCursorBase.MoveNext()
in Microsoft.ML.Trainers.SdcaTrainerBase3.TrainCore(IChannel ch, RoleMappedData data, LinearModelParameters predictor, Int32 weightSetCount) in Microsoft.ML.Trainers.StochasticTrainerBase2.TrainModelCore(TrainContext context)
in Microsoft.ML.Trainers.TrainerEstimatorBase2.TrainTransformer(IDataView trainSet, IDataView validationSet, IPredictor initPredictor) in Microsoft.ML.Data.EstimatorChain1.Fit(IDataView input)

I downgraded the version ml.net to 1.3.1 an all worked fine.
a little later, I upgraded the version to 1.5.0-preview and all worked fine.

[frank-dong-ms-zz]

Can you please provide a repro project and input data so we can look further?

I have also checked the release notes, seems nothing relate to this issue I can find.

[frank-dong-ms-zz]

We didn't hear back for couple months now, so I will close the issue for now, @Spocher feel free to reopen this issue if you can provide more details on this issue, thanks.

[ScubaAddict1]

I had the same issue, fixed it by adding
_trainingDataView = _mlContext.Data.Cache(_trainingDataView);
Not sure if its ideal as am using a asp.net core (.net 5) website with ML 1.6 and loading the data from SQL.