Table Of Contents
Table Of Contents

Model

Create, load/save, and use MXNet models

mx.model.buckets

Train RNN with bucket support

mx.model.FeedForward.create

Create a MXNet Feedforward neural net model with the specified training

mx.model.init.params

Parameter initialization

mx.model.load

Load model checkpoint from file

mx.model.save

Save model checkpoint into file

predict.MXFeedForwardModel

Predict the outputs given a model and dataset

is.serialized

Check if the model has been serialized into RData-compatiable format

mx.serialize

Serialize MXNet model into RData-compatiable format

mx.unserialize

Unserialize MXNet model from Robject

mx.mlp

Convenience interface for multiple layer perceptron

mx.infer.rnn.one

Inference for one-to-one fusedRNN (CUDA) models

mx.infer.rnn.one.unroll

Inference for one-to-one unroll models

mx.infer.rnn

Inference of RNN model

mx.nd.RNN

Applies recurrent layers to input data. Currently, vanilla RNN, LSTM and GRU are implemented, with both multi-layer and bidirectional support

mx.symbol.RNN

Applies recurrent layers to input data. Currently, vanilla RNN, LSTM and GRU are implemented, with both multi-layer and bidirectional support

rnn.graph

Generate a RNN symbolic model - requires CUDA

rnn.graph.unroll

unroll representation of RNN running on non CUDA device