githubnemo

Sorry for not getting back to you sooner. There's still one unaddressed review comment (`watchDirectories` vs. `watchDirectory`). Do you still want to address this? It's OK if you don't, I...

Just so that I understand the intent correctly, you propose something that enables the user to easily use pretrained models or transforming/embedding models in sklearn pipelines? Generally I like this...

> When do you see a user change the validation data in the middle of an epoch? Specifically, a user would need to change the validation set after the `y_pred`...

> Any additional thoughts to get this off the ground? I wonder if it would be best to add a 'def to_device' type function in utils.py that checks the device...

> > We already have this addition in one of the open PRs > > Which PR are you talking about? #557 introduces a `to_device` utility function. > > *...

https://github.com/tartley/colorama seems to be most popular at the moment.

@karmus89 Can you elaborate which problem you thought `max_epochs` was solving for you when you found out that it didn't? I think that perspective would add value to the discussion....

I can verify that building with maven works with the freshly cloned civcraft repository of @vemacs. Built with: ``` $ JAVA_HOME=/usr/lib/jvm/java-7-openjdk-amd64/jre/ mvn clean install ``` Great work, saved me much...

@vemacs I don't know what The Maven Way™ is but having a ready to go server structure cannot be bad, can it? If not with maven, how about a (shell)...

Hey! Thanks for the suggestion. Indeed this is something that is not symmetrical with `.fit()`. [There are](https://github.com/scikit-learn/scikit-learn/blob/cb836be0ff8347ccb0ab722760df68d07485101e/examples/miscellaneous/plot_metadata_routing.py#L182) sklearn classifiers and [transformers](https://github.com/scikit-learn/scikit-learn/blob/cb836be0ff8347ccb0ab722760df68d07485101e/sklearn/compose/_target.py#L281) that support additional parameters so I don't see an...