I am following the basic tutorial https://docs.gpytorch.ai/en/stable/examples/01_Exact_GPs/Simple_GP_Regression.html to train a Gaussian Process for the following data
train_x is a tensor of [4058, 12]
train_y is a tensor of [4058, 140]
I get an error calculating the loss
loss = -mll(output, train_y)
saying that output (model(train_x)) and train_y don't have the same dimension. Given this, I tried MultitaskGPModel https://docs.gpytorch.ai/en/stable/examples/03_Multitask_Exact_GPs/Multitask_GP_Regression.html getting a very similar error
RuntimeError: The size of tensor a (568120) must match the size of tensor b (8116) at non-singleton dimension 0
Apparently MultitaskGPModel requires the same total number of entries to be equal. Is there a way to train a multiple-entry GP?