Created by susaneraly on 3/27/16.

Created by susaneraly on 3/27/16.

Train AlphaGo Zero model on dummy data. To run a full AGZ system with DL4J, check out https://github.com/maxpumperla/ScalphaGoZero for a complete ex

Issue Summary: The labels in the test data are all the same. So even though net trains properly it gives terrible eval results. And this is confusing

Animal Classification Example classification of photos from 4 different animals (bear, duck, deer, turtle). References: - U.S. Fish and Wildlife Se

@author wangfeng

@author wangfeng

@author WangFeng

This example is intended to be a simple CSV classifier that seperates the training data from the test data for the classification of animals. It would

Basic DataVec example for preprocessing operations on some simple CSV data. If you just want to load CSV data and pass it on for learning take a look

This example trains a RNN. WHen trained we only have to put the first character of LEARNSTRING to the RNN, and it will recite the following chars @au

A simple example of how to attach Deeplearning4j's training UI to a network To change the UI port (usually not necessary) - set the org.deeplearning4

This example demonstrates how to run inference on a fine tuned BERT model in SameDiff, where the fine tuning happens outside in Tensorflow <p> Details

Read a csv file. Fit and plot the data using Deeplearning4J. @author Robert Altena

@author Adam Gibson

This example is a version of the basic CSV example, but adds the following: (a) Meta data tracking - i.e., where data for each example comes from (b)

This example is a version of the basic CSV example, but adds the following: (a) Meta data tracking - i.e., where data for each example comes from (b)

Basic DataVec example for preprocessing operations on some simple CSV data. If you just want to load CSV data and pass it on for learning take a look

Basic DataVec example for preprocessing operations on some simple CSV data. If you just want to load CSV data and pass it on for learning take a look

Read a csv file. Fit and plot the data using Deeplearning4J. @author Robert Altena

@author rubenfiszel (ruben.fiszel@epfl.ch) on 8/11/16. Main example for Cartpole DQN

Example: training an embedding using the center loss model, on MNIST The motivation is to use the class labels to learn embeddings that have the follo

A simple DataSetIterator for use in the LSTMCharModellingExample. Given a text file and a few options, generate feature vectors and labels for trainin

A simple DataSetIterator for use in the LSTMCharModellingExample. Given a text file and a few options, generate feature vectors and labels for trainin

This example is almost identical to the LSTMCharModellingExample, except that it utilizes the ComputationGraph architecture instead of MultiLayerNetwo

Sequence Classification Example Using a LSTM Recurrent Neural Network This example learns how to classify univariate time series as belonging to one

This is a runnable used to asynchronously and periodically save a copy of the specified parameters vector, while Spark training is proceeding. This is

This is an example of how to implement a custom activation function that does not take any learnable parameters Custom activation functions of this ca

This is an example of how to implement a custom activation function that does not take any learnable parameters Custom activation functions of this ca

This is an example that illustrates how to instantiate and use a custom activation function. The example is identical to the one in org.deeplearning4j

This is an example that illustrates how to instantiate and use a custom activation function. The example is identical to the one in org.deeplearning4j

Layer configuration class for the custom layer example @author Alex Black

Layer configuration class for the custom layer example @author Alex Black

Custom layer example. This example shows the use and some basic testing for a custom layer implementation. For more details, see the CustomLayerExampl

Layer (implementation) class for the custom layer example @author Alex Black

Layer (implementation) class for the custom layer example @author Alex Black

Custom layer example. This example shows the use and some basic testing for a custom layer implementation. For more details, see the README.md file @

A basic custom listener that records the values of z and out, for comparison or printing later.

This example shows how to use a custom listener, and is based on the {@link MNISTCNN}.<br><br> <p> We use a basic custom listener that records the val

@author susaneraly

Created by susaneraly on 11/9/16. This is an example that illustrates how to instantiate and use a custom loss function. The example is identical to t

Created by susaneraly on 11/8/16.

This is an example that illustrates how to define and instantiate a custom loss function. The example uses the quickstart SumModel example as a basis

NOTE: THIS IS ONLY RELEVANT TO THE 1.0.0-beta6 RELEASE Custom ops are those defined in C++ (libnd4j) that haven't yet been mapped to Java convenie

Custom Reduction example for operations on some simple CSV data that involve a custom reduction. @author François Garillot

Created by susaneraly on 3/27/16. This is class to generate pairs of random numbers given a maximum number of digits This class can also be used as a

Created by susaneraly on 3/27/16. This is class to generate a multidataset from the AdditionRNN problem Features of the multidataset - encoder in

Provides input blocks for dual residual or convolutional neural networks for Go move prediction. @author Max Pumperla

Common data utility functions. @author fvaleri

Common data utility functions. @author fvaleri

Attention layer for DeepMoji network architecture, following the implementation here: https://github.com/bfelbo/DeepMoji @author Max Pumperla

This Spark script downloads and preprocesses the patent files. Is is part 1 of 2 of this example. See PatentExampleReadme.md for more details. NOTE:

Given a base url and a zipped file name downloads contents to a specified directory under ~/dl4j-examples-data Will check md5 sum of downloaded file <

Given a base url and a zipped file name downloads contents to a specified directory under ~/dl4j-examples-data Will check md5 sum of downloaded file <

Given a base url and a zipped file name downloads contents to a specified directory under ~/dl4j-examples-data Will check md5 sum of downloaded file <

Given a base url and a zipped file name downloads contents to a specified directory under ~/dl4j-examples-data Will check md5 sum of downloaded file <

Given a base url and a zipped file name downloads contents to a specified directory under ~/dl4j-examples-data Will check md5 sum of downloaded file <

Define and load an AlphaGo Zero dual ResNet architecture into DL4J. The dual residual architecture is the strongest of the architectures tested by De

Early stopping example on a subset of MNIST Idea: given a small subset of MNIST (1000 examples + 500 test set), conduct training and get the parameter

Early stopping example on a subset of MNIST Idea: given a small subset of MNIST (1000 examples + 500 test set), conduct training and get the parameter

@author susaneraly on 3/1/17. IMPORTANT: 1. The forward pass on VGG16 is time consuming. Refer to "FeaturizedPreSave" and "FitFromFeaturized" for how

@author susaneraly on 3/9/17. We use the transfer learning API to construct a new model based of org.deeplearning4j.transferlearning.vgg16 We will ho

<p> This is a seq2seq encoder-decoder LSTM model made according to Google's paper <a href="https://arxiv.org/abs/1506.05869">A Neural Conversational M

{@link org.datavec.api.split.InputSplit} and its implementations are utility classes for defining and managing a catalog of loadable locations (paths/

{@link org.datavec.api.split.CollectionInputSplit} is a basic implementation of {@link org.datavec.api.split.InputSplit} that's useful when we want to

{@link org.datavec.api.split.NumberedFileInputSplit} is useful when you have a sequence of numbered files, following a common pattern. For example, if

{@link org.datavec.api.split.TransformSplit} is an InputSplit implementation that maps the URIs of a given {@link org.datavec.api.split.BaseInputSplit

{@link org.datavec.api.split.BaseInputSplit} and its implementation provides a {@link org.datavec.api.split.BaseInputSplit#sample(PathFilter, double..

@author Robert Altena

This example: how to implement a custom DL4J layer using SameDiff. The layer itself is a standard dense (fully connected) layer with an activation fu

This is a simple example introducing the basic concepts of SameDiff: - The SameDiff class - Variables - Functions - Getting arrays - Executing forward

This example: how to implement a simple custom DL4J lambda layer using SameDiff. The lambda layer (see L2NormalizeLamdaLayer) implements "out = in /

This example shows how to implement a simple linear regression graph with a mean-squared error loss function. Specifically, we will implement: - outp

This example: how to implement a simple custom DL4J lambda vertex using SameDiff. Lamda vertices are graph vertices that do not have any parameters,

This example: shows some more ways to create variables @author Alex Black

Instrumented test, which will execute on an Android device. @see <a href="http://d.android.com/tools/testing">Testing documentation</a>

The TransferLearningHelper class allows users to "featurize" a dataset at specific intermediate vertices/layers of a model This example demonstrates h

Important: 1. Either run "EditAtBottleneckOthersFrozen" first or save a model named "MyComputationGraph.zip" based on org.deeplearning4j.transferlearn

@author susaneraly on 3/10/17. Important: Run the class "FeaturizePreSave" before attempting to run this. The outputs at the boundary of the frozen a