keras th ordering

dl4j.scala

  val model = KerasModelImport.importKerasSequentialModelAndWeights(
    "/opt/devel/src/hannibal/hannibal-python/ecoa/keras4j.json",
    "/opt/devel/src/hannibal/hannibal-python/ecoa/keras4j.h5")


  val (trackLength, vectorDim, groupSize) = (30, 2, 8)

  val o = Array.fill(trackLength * vectorDim * groupSize)(1.0)

  val ones = Nd4j.create(o, Array(1, groupSize, vectorDim, trackLength))

  val oneOut = model.output(ones)
// [0.41, 0.40, 0.19]

firstLayerWeights.txt

In dl4j:
scala> println(res8.getRow(0))
[[[-0.01, -0.06, 0.01, 0.07, -0.01, 0.09, 0.08, 0.05, -0.02, -0.00],
  [-0.07, 0.05, 0.02, -0.07, 0.07, 0.00, 0.01, 0.07, -0.03, 0.08]],

 [[0.05, 0.07, -0.00, 0.01, 0.01, -0.04, -0.04, 0.08, 0.07, -0.07],
  [-0.03, -0.01, 0.07, 0.08, -0.01, 0.04, -0.00, 0.02, -0.05, 0.05]],

 [[0.01, 0.02, -0.05, 0.02, -0.07, -0.07, -0.06, -0.00, 0.03, 0.05],
  [-0.06, 0.06, -0.00, 0.02, -0.04, -0.00, 0.07, -0.03, -0.03, -0.06]],

 [[0.08, -0.08, -0.08, -0.01, -0.02, 0.02, 0.01, 0.02, 0.00, -0.06],
  [0.08, -0.03, -0.08, -0.03, 0.08, 0.02, 0.02, 0.04, 0.07, -0.02]],

 [[0.06, -0.07, 0.04, 0.01, 0.06, -0.05, 0.07, 0.06, 0.02, -0.05],
  [-0.01, 0.03, -0.02, 0.05, -0.02, -0.05, -0.07, 0.03, 0.08, -0.05]],

 [[-0.02, 0.07, 0.05, 0.04, -0.08, -0.05, -0.03, 0.02, -0.04, -0.08],
  [-0.03, -0.02, 0.01, -0.03, 0.03, -0.01, 0.05, 0.03, 0.00, -0.05]],

 [[-0.07, 0.01, -0.06, -0.05, -0.00, 0.04, -0.02, -0.03, -0.07, 0.01],
  [-0.04, 0.07, -0.04, 0.06, 0.08, -0.03, -0.04, -0.02, 0.04, -0.02]],

 [[0.08, 0.04, -0.08, -0.04, 0.03, 0.07, 0.07, -0.04, -0.03, 0.07],
  [-0.05, -0.01, 0.04, -0.08, 0.06, 0.07, -0.02, -0.02, -0.05, -0.01]]]

In Python:
  print model.get_weights()[0][0]
[[[ 0.07676533 -0.03117377  0.06891505  0.00826494  0.00290956  0.07187851
   -0.07164567  0.0169886   0.05354632 -0.06551674]
  [-0.00348628 -0.01874308  0.05135714  0.08106387  0.08658282 -0.0125316
    0.07233612  0.00676258 -0.0583616  -0.00871787]]

 [[ 0.05433306 -0.05441435  0.01570591 -0.00498693  0.04137021 -0.00683805
    0.08082487  0.07156506 -0.00727309 -0.03086448]
  [-0.07242365  0.06931256  0.07738896 -0.03929505 -0.03519952  0.01477922
    0.00650515 -0.00138164  0.06599145  0.0548986 ]]

 [[-0.06366576 -0.02515704 -0.03476964  0.0700172  -0.00052914 -0.03933529
    0.02269914 -0.00237731  0.06332166 -0.05861359]
  [ 0.05349948  0.02544831 -0.00453325 -0.05656359 -0.06786122 -0.07119821
    0.02357902 -0.04714607  0.01953493  0.00579332]]

 [[-0.01521546  0.07284176  0.04143168  0.02168714  0.02437495  0.08211306
   -0.02770139 -0.08141261 -0.02879204  0.07727191]
  [-0.06016762  0.00331838  0.01770175  0.01325131  0.01809193 -0.02122333
   -0.01375933 -0.07907184 -0.07925276  0.08340394]]

 [[-0.04790336  0.08006929  0.03018796 -0.07247118 -0.04933546 -0.02051881
    0.04873452 -0.01753233  0.03327365 -0.01034204]
  [-0.04579724  0.01649789  0.05555908  0.0725324  -0.04999334  0.06409916
    0.01181215  0.04371393 -0.06629309  0.05552076]]

 [[-0.05160485  0.00228648  0.02856116  0.04772033 -0.01463272  0.02620524
   -0.02530901  0.01186486 -0.01923612 -0.03276508]
  [-0.07548475 -0.04069661  0.02406996 -0.0337285  -0.05363468 -0.07712238
    0.03820712  0.04714371  0.06748833 -0.02      ]]

 [[-0.01733848  0.03871317 -0.01713949 -0.03550589 -0.02539484  0.08202767
    0.06316736 -0.03533741  0.06995845 -0.04261048]
  [ 0.00951921 -0.06653412 -0.03266657 -0.01623273  0.03671551 -0.00367464
   -0.04966186 -0.05797498  0.00909566 -0.06771586]]

 [[-0.00626809 -0.05195602 -0.01752971 -0.02099497  0.0686373   0.06118456
   -0.07832465  0.03741774 -0.01404193 -0.04903556]
  [ 0.06766865 -0.02603962 -0.03748363  0.07432315  0.06585176  0.03124297
   -0.03805119 -0.0758684   0.04481697  0.08070454]]]

keras-model.py

from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Activation, Flatten
from keras.layers.convolutional import Convolution2D, Convolution3D
from keras.optimizers import RMSprop, Adadelta

groupSize = 8
vectorDim = 2
trackLength = 30

model = Sequential()
model.add(Convolution2D(32, 2, 10, border_mode='same', dim_ordering='th', input_shape=(groupSize, vectorDim, trackLength)))
model.add(Activation('relu'))

model.add(Convolution2D(16, 2, 5, border_mode='same', dim_ordering='th'))
model.add(Activation('relu'))
model.add(Flatten())

model.add(Dense(128))

model.add(Dense(3))
model.add(Activation('softmax'))

optim = RMSprop()
# optim = Adadelta()
model.compile(loss='categorical_crossentropy', optimizer=optim)

model.predict_proba(np.ones((1, groupSize, vectorDim, trackLength)))
#array([[ 0.45079258,  0.24883677,  0.30037063]], dtype=float32)

model_json = model.to_json()
with open("keras4j.json", "w") as json_file:
    json_file.write(model_json)
# serialize weights to HDF5
model.save_weights("keras4j.h5")
print("Saved model to disk")

lastDenseLayerWeights.txt

In dl4j:
Map(W -> [[0.14, 0.12, -0.16],
 [-0.04, 0.15, 0.02],
 [-0.16, -0.06, -0.07],
 [-0.13, -0.14, -0.07],
 [0.19, 0.09, 0.09],
 [-0.11, -0.13, 0.12],
 [-0.00, 0.05, -0.06],
 [-0.19, 0.17, -0.11],
...

In Python:
[[  1.38035819e-01   1.24384061e-01  -1.55227304e-01]
 [ -3.71695161e-02   1.51023373e-01   2.41503268e-02]
 [ -1.59801364e-01  -5.85760027e-02  -6.58198148e-02]
 [ -1.26131564e-01  -1.43218338e-01  -7.05740899e-02]
 [  1.93560138e-01   8.63352865e-02   9.10301656e-02]

warn.log

17/05/18 16:56:22 WARN reflections.Reflections: could not create Vfs.Dir from url. ignoring the exception and continuing
org.reflections.ReflectionsException: could not create Vfs.Dir from url, no matching UrlType was found [file:/opt/devel/src/expeditionary-warfare/exw-app/target/test-classes]
either use fromURL(final URL url, final List<UrlType> urlTypes) or use the static setDefaultURLTypes(final List<UrlType> urlTypes) or addDefaultURLTypes(UrlType urlType) with your specialized UrlType.
	at org.reflections.vfs.Vfs.fromURL(Vfs.java:109)
	at org.reflections.vfs.Vfs.fromURL(Vfs.java:91)
	at org.reflections.Reflections.scan(Reflections.java:237)
	at org.reflections.Reflections.scan(Reflections.java:204)
	at org.reflections.Reflections.<init>(Reflections.java:129)
	at org.deeplearning4j.nn.conf.NeuralNetConfiguration.registerSubtypes(NeuralNetConfiguration.java:431)
	at org.deeplearning4j.nn.conf.NeuralNetConfiguration.configureMapper(NeuralNetConfiguration.java:386)
	at org.deeplearning4j.nn.conf.NeuralNetConfiguration.initMapper(NeuralNetConfiguration.java:376)
	at org.deeplearning4j.nn.conf.NeuralNetConfiguration.<clinit>(NeuralNetConfiguration.java:123)
	at org.deeplearning4j.nn.conf.NeuralNetConfiguration$Builder.build(NeuralNetConfiguration.java:1019)
	at org.deeplearning4j.nn.conf.NeuralNetConfiguration$ListBuilder.build(NeuralNetConfiguration.java:269)
	at org.deeplearning4j.nn.modelimport.keras.KerasSequentialModel.getMultiLayerConfiguration(KerasSequentialModel.java:203)
	at org.deeplearning4j.nn.modelimport.keras.KerasSequentialModel.getMultiLayerNetwork(KerasSequentialModel.java:224)
	at org.deeplearning4j.nn.modelimport.keras.KerasSequentialModel.getMultiLayerNetwork(KerasSequentialModel.java:213)
	at org.deeplearning4j.nn.modelimport.keras.KerasModelImport.importKerasSequentialModelAndWeights(KerasModelImport.java:236)
	at $line12.$read$$iw$$iw$$iw$$iw$.<init>(<console>:21)
	at $line12.$read$$iw$$iw$$iw$$iw$.<clinit>(<console>)
	at $line12.$eval$.$print$lzycompute(<console>:7)