This sample is calling Fit() with a dataview generated from one sample: https://github.com/dotnet/machinelearning/blob/master/docs/samples/Microsoft.ML.Samples/Dynamic/TensorFlow/TextClassification.cs
The pipeline contains a pre-trained TensorFlow model, so there is not need (and is misleading) to call Fit with data.
In other places where no dataview is required, we generate an empty one.
Related to #3795
This sample is calling Fit() with a dataview generated from one sample: https://github.com/dotnet/machinelearning/blob/master/docs/samples/Microsoft.ML.Samples/Dynamic/TensorFlow/TextClassification.cs
The pipeline contains a pre-trained TensorFlow model, so there is not need (and is misleading) to call Fit with data.
In other places where no dataview is required, we generate an empty one.
Related to #3795