Where communities thrive

  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
  • Mar 27 12:33
    treo labeled #7319
  • Mar 27 12:33
    treo commented #7319
  • Mar 27 08:15
    AlexDBlack edited #8812
  • Mar 27 08:14
    AlexDBlack labeled #8812
  • Mar 27 08:14
    AlexDBlack labeled #8812
  • Mar 27 08:14
    AlexDBlack labeled #8812
  • Mar 27 08:14
    AlexDBlack opened #8812
  • Mar 26 19:42
    linuxciscoarnaud opened #8811
  • Mar 26 18:26
    sshepel commented #8750
  • Mar 26 13:33
    AlexDBlack closed #8765
  • Mar 26 13:33
    AlexDBlack closed #8715
  • Mar 26 13:33
    AlexDBlack closed #8718
  • Mar 26 13:33
    AlexDBlack closed #8776
  • Mar 26 13:33
    AlexDBlack closed #8704
  • Mar 26 13:33
    AlexDBlack closed #8768
  • Mar 26 13:33
    AlexDBlack closed #8777
  • Mar 26 13:10
    TramsWang opened #8810
  • Mar 26 10:17
    saudet commented #8806
  • Mar 26 09:58
    mheck136 commented #8806
  • Mar 26 09:14
    saudet commented #8750
Bartosz Kruba
Hi! I'm trying to create NeuralNetConfiguration with dl4j-beta6. There seems to be no option for configuring learningRate backpropagation and pretrain. Were those settings removed in newer versions or am i doing something wrong?
Paul Childs
I'd like to verify that models do not uptrain automatically using, for example, vec.writeParagraphVectors(vectorsFile.pv)
This will just overwrite the current .pv file, right?

I have a Keras 2.3.1 h5 CNN fully trained model that I import using DL4J 1.0.0-beta6 version. I intend to use DL4J for evaluating outputs for specific input images. Currently Keras prediction outputs don't match DL4J outputs. The shapes of the inputs are different Keras input shape is (4, 150, 150, 3) whereas DL4J
input shape is [4, 3, 150, 150]. I could not find any sample DL4J code which
performs predictions on a CNN (or any network) trained model imported from
Keras that use images for predictions in Java and hence cannot ascertain if this
is an issue.

1) What is the right shape of input required ini DL4J for the imported Keras
Model that expects image input.

2) If input shapes have to match between DL4J and Keras, I cannot identify a
way to build a "channel last" INDArray from a set of input images.

3) I need to also perform this on Keras imported mixed multi-input and
multi-output networks. What is recommended way to feed inputs to such networks
for predictions.



# Build and train the CNN Model
imgGenerator = ImageDataGenerator(rescale=1./255)
testDir = "/test"

testData = np.array([["cat.1500.jpg"],["cat.1501.jpg"],["dog.1500.jpg"],["dog.1501.jpg"]])
testDf = pd.DataFrame(testData, columns=["file"])
netInput=imgGenerator.flow_from_dataframe( dataframe=testDf, 
    directory=testDir, x_col="file", y_col=None, 
    batch_size=testData.shape[0], seed=42, shuffle=False, class_mode=None, 
    target_size=(150, 150))
netOutput = model.predict(netInput);
print("Prediction: ", netOutput)


Prediction: [[0.01729406]
[0.9373665 ]]

Shape of the batch generated by netInput is (4, 150, 150, 3)


String[] testImgs = new String[]{"cat.1500.jpg","cat.1501.jpg","dog.1500.jpg","dog.1501.jpg"};

INDArray[] ret = new INDArray[testImgs.length];
int count = 0;
for (String testImg: testImgs)
    File f = new File(mInputPath + "/test", testImg);
    NativeImageLoader loader = new NativeImageLoader(150, 150, 3);
    INDArray image = loader.asMatrix(f);
    DataNormalization scalar = new ImagePreProcessingScaler(0, 1);
    ret[count] = image;
// Concatenate the array along dimension 0.
INDArray netInput = Nd4j.concat(0, ret);
INDArray netOutput = model.output(netInput);
System.out.println("Prediction: ", netOutput)

Prediction: [[0.0381],

Shape of netInput is [4, 3, 150, 150]

Ali Akhtar
Which maven dependencies do I need to include in order to be able to use the Word2Vector functionality i.e https://deeplearning4j.org/docs/latest/deeplearning4j-nlp-word2vec ?
On https://github.com/eclipse/deeplearning4j-examples/blob/master/pom.xml I'm not seeing any dependencies but I am seeing a lot of plugins etc, are these plugins necessary for using Word2Vec?
Hi, I'm trying to predict some variables (angle and distance) from different input variables and data. Of I understand correctly this is a multi target regression problem, am I right ? For the implementation should I start with the regression examples and simply add 2 values output instead of a single value output ? Thx ?
Bastien Enjalbert
@AlexDBlack Don't want to reopen this github issue : deeplearning4j/DataVec#516 but what is the current proposal status ? Did it get implemented ?
Ali Akhtar
Which maven dependency should I import in order to be able to access org.deeplearning4j.models.embeddings.loader.WordVectorSerializer ?
Ali Akhtar
oh, its deeplearning4j-nlp
o.n.l.a.o.e.DefaultOpExecutioner - Backend used: [CPU]; OS: [Linux]
How can I get this to use my GPU?
hi ,where to find out the import org.deeplearning4j.examples.download.DownloaderUtility in the example code?
I cannot find this class
I find it under the shared-utilities
i try get an iterator for some rnn computation graph the csv needs to be split in to vectors like: .addInput("myReader",0,2).addInput("myReader",3,8) from RecordReaderMultiDataSetIterator is there anything like this on the class: SequenceRecordReaderDataSetIterator
Raul Feresteanu
How do I get the row and column indices out of INDArray.argMax for a 2D array?
Barney Stinson
any c# programmer here...?
Daniel MÃ¥rtensson
@Barney_Stinson_gitlab Only Java ;)
Raul Feresteanu
From what I understand argMax/argMin seems to return indices along rows/columns but I'd like to know if I can get it to return the row and column indices of the element in question
Raul Feresteanu
Nevermind, I wrote my own helper. Would still be a nice addition to the API
@fortuneNext try Nd4j.setDefaultDataTypes(DataType.DOUBLE,DataType.DOUBLE); put this at the begining of main
I have a question, i tried to import a keras model. the keras model has masking layer plus lstm(64) and dense layers. taking input the padded sequence of one-hot encoded vectors. After importing, i can see it's maskzerolayer then lstm dense layer. Checked each layer weights, the lstm layer weights don't match up, all others match up with python version. The LSTM in dl4j has timestep and features in a swap position compared to python version. The input has been modified accordingly. But it just won't give same result with python verion, any help please?
I run the example code of UIExample,when i go to http://localhost:9000/train, it shows Resource not found,and i search the project ,only two files(log4j ,logback configure file) under the resource folder,where are the needed resource ?
Hi, I'm trying to build regression but have next Exception https://gist.github.com/lemair/2e44f16a3b232019dbf3983db42a22ac
Hello! I'm trying to play around https://github.com/wmeddie/dl4j-gans and to add UIServer server there to observe metrics of the GAN network. But still getting troubles,. First, I got the missing class exceptin, after updating to beta6 I just don't see any localhost:9000 starts for the UI Server, while the calculation is going forward fine
and the second question is how it is designed to use only part of the large jpeg image (2000x3000, 3 chanels) while I instantiate the ImageRecordReader(100, 100, 1). Not clear should I prepare the image by myself in order to use imageRecorderReader correctly: https://stackoverflow.com/questions/60198054/correct-initialization-of-imagerecordreaderlong-height-long-width-long-channe
Hi! I am dl4j begginer, I have a question about scope . Is there some dl4j example for scope of data ?
For example, I teach the A.I below data.
"from 1 to 10 =>legend 1, from 11 to 20 => legend 2, from 21 to 30 => legend 3, from 31 to 40 => legend 4, ... "
Then I can give the scope (a legend) him if a customer input a data.
I am sorry My English is not perfect. I hope you understand it that i want to do.
I have been using computation graph with LSTM layers. But the dependency jars of these have been around 35 jars for linux which sums up to 125 MB. Is there any way to reduce the size by removing the unwanted files from the jar? Most of the jar size where occupied by ".so" files.
Hi, I hope you are all well. I followed the quick start instruction. however, I got an error that the shared utility dependency couldn't resolve. is there any help or recommendation?
if you use eclipse , then you have to import "shared-utilities" project also in your eclipse Package Explorer. I fixed by it.
or you can refer "eclipse/deeplearning4j#8307"
git clone https://github.com/eclipse/deeplearning4j-examples
cd shared-utilities
mvn install
cd ../dl4j-examples
mvn install
Hello. I am trying to train a MultiLayerNetwork using the fit method. I first tried to create an INDArrayDataSetIterator, where I passed the first INDArray as a feature matrix (64 rows and 6 columns) and an INDArray for the label prediction (64 rows and 2 columns).
However, it didn't seem to train correctly as I was constantly getting the same values. I then tried to do it wihtout a dataset iterator but nothing changed.
Any ideas on what i'm doing wrong? I don't think it is related to the network configuration or the learning rate.. Only one parameter of the network output seems to change
Amlesh Sivanantham
hey guys, I'm having some trouble running valgrind to test my java class running some dl4j code. I was following the guide on the old nd4j wiki but the output is not what I expect. They create a script called valgrindJava which essentially wraps the java process into valgrind. I tried running it with my actual arguments, through maven, and standalone without the script. In all cases, it seems that valgrind gives its summary before the java process even runs. Yet in the valgrind logs, I do see the full java command being printed so it is getting it. Hoping someone here understands what the issue could be. Here is the wiki page I was referring to, https://github.com/deeplearning4j/libnd4j/wiki/Debugging-libnd4j
Amlesh Sivanantham
Hey all identified the issue. My java in my path was actually a shell script that then forwarded the call to the actually java binary. Once I passed that in, its correctly being scaned by valgrind
Yuniel Acosta Pérez
I can in the process of transforming my dataset convert a column that contains text into a Vector using Word2Vec. If so, can someone explain to me how
Yuniel Acosta Pérez
I need help please.
@yuniel-acosta You can, see the example in https://deeplearning4j.org/docs/latest/deeplearning4j-nlp-word2vec you the you just need to input the information from your file instead of raw_sentences.txt
Anybody can suggest some good tutorial on how to setup a simple reinforcement learning environment in DL4J? Examples require frameworks, the toy sample is not documented and object involved are not immediate to understand event diving in the code.
new to DL4J
trying to do the serup
getting below exception when i try to run the CSVExample.java in IntelliJ