Charles Weill

@fangkuann: Yes it does! You can see how we use it [here](https://github.com/tensorflow/adanet/blob/master/adanet/core/estimator_distributed_test_runner.py#L316). Note this is the old `contrib` version of GBDT, and there is a new one called `tf.estimator.BoostedTreesEstimator` which...

@fangkuann Thanks for sharing that paper, I wasn't aware of it, and will share it with our team. As for the Core GBDT in AdaNet: your sample code looks fine....

@cxchen100 In your TensorBoard you'll see three kinds of curves: `.*subnetwork.*`, `.*ensemble.*`, and a couple like `/eval`. The `.*ensemble.*` models are the only candidates that will be considered for serving....

Do you have a colab you can share with some sample data?

Could you please reproduce the error in one of the tutorial notebooks, and share the Colab? That will be most helpful to future users who have this problem.

The AdaNet *is* the best performing model. You can export the model using any of the `Estimator#export*` methods for serving, or simply call `Estimator#predict` to get its predictions.

@minimaxir Thank you for filing this feature request. We'll take a look.

@svjack: You can visualize the learned architecture in TensorBoard by looking at the Graph tab.

We're working on a new Keras API for AdaNet, but don't have a timeline on it yet. Keep an eye out for related commits over the next few months.

The timeline will look roughly like so (subject to change): 1. Support `tf.keras.Layers` in subnetworks. 2. Support `tf.keras.Sequential` and [Keras functional API](https://www.tensorflow.org/guide/keras#functional_api) models in `adanet.AutoEnsembleEstimator`. 2. Implement `adanet.keras.Model` and `adanet.keras.AutoEnsemble`...