minibatch for training parallel restart entirely initialize experiment in GPExperiment (likelihood, mean, covar) lib usage documentation (with step-by-step) + coincise API handle nested object (e.g. Product Kernel, Sum Kernel) while (de)serializing extend Plottable whenever possible and implement default plot() pip-tools-multi for requirements test lib is correctly installed, up and running using tests convert notebooks/example_i.ipynb to examples/example_i.py for each i