Using Dataset with EvaluationLogger #5532
Description
I have a use case where I'm using the EvaluationLogger to track an eval, and I'd like to pass a Dataset when instantiating the logger that I can then add rows to as the eval progresses, as I don't have access to them up front. However, the resulting Evaluation gets whatever the first version of the Dataset was, regardless of whether I add rows during the course of the eval. It would be good to be able to construct a dataset in this way as the eval progresses.
One related issue that is easier to overcome is that the Dataset can't be instantiated empty - it would also be useful to define a Dataset before any of the samples are accessible.
An example of what I'm trying to describe can be found here