Fix label_smoothing for sigmoid_cross_entropy_loss, and test for softmax_cross_entropy_loss with label_smoothing.

The sigmoid cross entropy with label smoothing was broken, and worse the tests for both that and the softmax cross entropy with label smoothing were broken.  I've fixed both issues here and added comments walking through the two examples in the tests so as to not inadvertently check in broken tests again.
Change: 127100213

Change: 127093435

If a dataset is private, it will only show up if it is requested via URL, otherwise it will
not appear in the dropdown.
Change: 127090337

Change: 127084359

Now that the file chooser is conditionally shown, cannot use "this.$.[id]" to refer to the element.
Change: 127084144

Change: 127082828

Change: 127081085

Add custom op files build rules for building internal servers including custom ops.
Change: 127077645

Change: 127027579

Change: 127016158

Change: 127009550

Change: 126973281

Change: 126972079

Change: 126965578

Change: 126964234

Change: 126960644

Change: 126959121

Change: 126958138

Change: 126957750

Change: 126954743

Handles errors.AbortedError properly so trainer task does not have to restart/block indefinitely. Without this, currently trainers gets stuck indefinitely on a single ps shard crash.
Change: 126954343

Update documentation for pack and unpack. Add suggestions in split and concat as to when to alternatively use pack and unpack.
Change: 126954039

Change: 126948643

Change: 126948464

threadpool, instead of sequentially. Doing this speeds up training at the
cost of slightly increased memory usage.
Change: 126947298

Change: 126943819

Change: 126941079

Change: 126940987

Change: 126931969

Change: 126930259

Change: 126927117

Change: 126927101

Change: 126926398

Change: 126924005

https://github.com/tensorflow/tensorflow/issues/3177#issuecomment-231151090
https://github.com/tensorflow/tensorflow/issues/3200#issuecomment-231162520

Fixes #3177
Fixes #3200
Change: 126922632

Begin removing feature column inference from linear and dnn estimators. Currently, the fit operation of each of them will infer feature columns from the passed in features. But it only works for dense float inputs.

Also, fixed some lint warnings.
Change: 126921818

Change: 126921726

Add unit test for Composable model.
Remove some dead code.
Change: 126918604

Change: 126916693

where these live in ops libraries.
Change: 126916364