Early Stopping Optimization

[ramosv]

Some of the algorithms and models we have could take advantage or early stopping during training. I can help with this task if someone is interested to take it on.

[ElyasYassin]

Suggestion:
Can we get the option to disable early stopping? There's no parameter for it in GNN embedding.
https://arxiv.org/abs/1705.08741