v0.2.3

• Adding the MSE and sMAPE error measures
• Fix returning tensors from the GPU back to the CPU
• Fix repeated use of a dataset by properly deepcopying it
• Add console output for training
• Fix the LBFGS optimizer
• Add the plot_losses function to the Model class to plot losses/errors after train separately

v0.2.2

• Allow mean functions for MOGPs
• Add the Matérn kernel
• Fix a bug in the get_gram_matrix function

v0.2.0

Migration to PyTorch from using GPflow and TensorFlow. Many kernels have been observed with speed improvements of up to 10x.

v0.1.4

Release 0.1.4