Pierrick Bruneau

**Is your feature request related to a problem? Please describe.** Referring to the Keras+Tensorflow example at https://github.com/horovod/horovod/blob/master/examples/tensorflow2/tensorflow2_keras_mnist.py, I'm considering distinct training and validation datasets by expanding l42 as follows: ```...

Estimator.create_predictor triggers RuntimeError with DeepAREstimator

3

## Description When trying to adapt the custom callback as described at https://ts.gluon.ai/master/tutorials/mxnet_models/trainer_callbacks.html to `DeepAREstimator`, I get a `RuntimeError`. I guess this is not exactly a bug, as there must...

## Description Depending on the execution context (CPU or GPU) as determined by where a given trained model is deserialized, the result from the `self.rnn.unroll` in `DeepARPredictionNetwork.sampling_decoder` (`mx.model.deepar`) gives a...

Blocked at implementing torch port of mx/distributions/mixture.py

2

## Description I am currently trying to port gluonts.mx.distributions.mixture.py to torch, aiming at using the ported MixtureDistributionOutput with NormalOutput components and DeepAR. I came up with the following implementation: https://gist.github.com/pbruneau/3c9d62f694c50ead8da7adf50014d13a...