Created
March 22, 2017 10:38
-
-
Save bikashg/132bd63011636d07719fbd6fd97ef534 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
ComputationGraphConfiguration configuration = new NeuralNetConfiguration.Builder() | |
.weightInit(WeightInit.XAVIER) | |
.learningRate(0.25) | |
.updater(Updater.RMSPROP) | |
.optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).iterations(1) // Nb. of iterations = 1 | |
.seed(123) | |
.graphBuilder() | |
//These are the names of the two inputs to the computation graph. | |
.addInputs("nlPhrase", "owlAxiom") | |
.setInputTypes(InputType.recurrent(vocab.size()), InputType.recurrent(vocab.size())) // The size of the vocab. | |
//The inputs to the encoder will have size = minibatch x featuresize x timesteps | |
//Note that the network only knows of the feature vector size. It does not know how many time steps unless it sees an instance of the data | |
.addLayer("encoder", new GravesLSTM.Builder().nIn(vocab.size()).nOut(128).activation(Activation.SOFTSIGN).build(),"nlPhrase") // For now, the LSTM cell in the encoder has 128 nodes; input timesteps will contain as many as vocab.size() elements. | |
.addLayer("decoder", new GravesLSTM.Builder().nIn(vocab.size()).nOut(128).activation(Activation.SOFTSIGN).build(), "owlAxiom") // For now, the LSTM cell in the decoder also has 128 nodes; input timesteps will contain as many as vocab.size() nodes. | |
// nIn for RnnOutputLayer is the number of hidden layers in the decoder's last LSTM cell | |
.addLayer("output", new RnnOutputLayer.Builder().nIn(128).nOut(vocab.size()).activation(Activation.SOFTMAX).lossFunction(LossFunctions.LossFunction.MCXENT).build(), "decoder") | |
.setOutputs("output") | |
.pretrain(false).backprop(true) | |
.build(); | |
ComputationGraph net = new ComputationGraph(configuration); |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment