Training a single GPyTorch model on a batch of data

[scfeltman]

I am trying to train a single model on a batch of data, much like a typical deep learning model with training and test splits of different observations. I've looked through the documentation page and tutorials and have found that each model requires an X and Y input during initialization, and the model cannot be trained on any other set of data. This makes sense, considering initial mean and covariance matrices, but I was wondering if there's any way to train a universal model on a batch of data? I've tried stacking the data on top of each other so multiple observations per X coordinate, but these results are inadequate. Has anyone used GPyTorch in this way before? Thanks in advance.