New CV classes

Added GridSearchCV and RandomizedSearchCV which are based on the scikit-learn equivalents

v1.7.19

Removed tensorboard for simplicity

Fixed NOI

v1.7.18

Switching to pytest

1.7.17

First time using github actions to do releases.

Switched to pytest for testing.

score() for full sklearn compatibility

v1.7.11

Created using Colaboratory

More stable deep objectives

v1.7.10

Created using Colaboratory

Rejigged Deep API

Previously:
dcca=DCCA(optimizer=optimizer)
dcca=DeepWrapper(dcca).fit()

Now:
dcca=DCCA()
dcca=DeepWrapper(dcca,optimizer=optimizer).fit()

So don't actually need to use DeepWrapper its just a helper class for training in standardised way.

Simpler deep models

v1.7.3

Bump version

v1.7.1

Changed defaults for regularisation r and eps .

Stable DCCA hopefully

Now requires pytorch 1.9 which has the new pytorch.linalg which has stable gradients even when some eigenvalues are the same!