Added enhancements to the Gluon package, a high-level interface designed to be easy to use while keeping most of the flexibility of low level API. Gluon supports both imperative and symbolic programming, making it easy to train complex models imperatively with minimal impact on performance. Neural networks (and other machine learning models) can be defined and trained with gluon.nn and gluon.rnn packages.
Added new loss functions - SigmoidBinaryCrossEntropyLoss, CTCLoss, HuberLoss, HingeLoss, SquaredHingeLoss, LogisticLoss, TripletLoss.
gluon.Trainer now allows reading and setting learning rate with trainer.learning_rate property.
Added API HybridBlock.export for exporting gluon models to MXNet format.
Added Convolutional recurrent networks gluon.contrib package.
- Convolutional recurrent network cells for RNN, LSTM
,
- and GRU.
Added VariationalDropout.
- VariationalDropoutCell

New Features - Autograd

Added enhancements to autograd package, which enables automatic differentiation of NDArray operations.
autograd.Function allows defining both forward and backward computation for custom operators. See documentation for examples.
Added mx.autograd.grad and experimental second order gradient support (most operators don't support second order gradient yet).
Autograd now supports cross-device graphs. Use x.copyto(mx.gpu(i)) and x.copyto(mx.cpu()) to do computation on multiple devices.

...

Page tree