CPU info:
    CPU Model Name: Intel(R) Xeon(R) CPU E5-2690 v3 @ 2.60GHz
    Hardware threads: 12
    Total Memory: 57691188 kB
-------------------------------------------------------------------
=== Running /home/ubuntu/workspace/build/gpu/release/bin/cntk configFile=/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/cntk_sequence.cntk currentDirectory=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData RunDir=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu DataDir=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData ConfigDir=/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining OutputDir=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu DeviceId=0 timestamping=true
CNTK 2.3.1+ (HEAD b7b3e4, Jan 17 2018 02:42:45) at 2018/01/17 06:13:45

/home/ubuntu/workspace/build/gpu/release/bin/cntk  configFile=/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/cntk_sequence.cntk  currentDirectory=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData  RunDir=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu  DataDir=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData  ConfigDir=/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining  OutputDir=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu  DeviceId=0  timestamping=true
Changed current directory to /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData
01/17/2018 06:13:45: -------------------------------------------------------------------
01/17/2018 06:13:45: Build info: 

01/17/2018 06:13:45: 		Built time: Jan 17 2018 02:36:21
01/17/2018 06:13:45: 		Last modified date: Wed Jan 17 02:34:37 2018
01/17/2018 06:13:45: 		Build type: release
01/17/2018 06:13:45: 		Build target: GPU
01/17/2018 06:13:45: 		With ASGD: yes
01/17/2018 06:13:45: 		Math lib: mkl
01/17/2018 06:13:45: 		CUDA version: 9.0.0
01/17/2018 06:13:45: 		CUDNN version: 7.0.4
01/17/2018 06:13:45: 		Build Branch: HEAD
01/17/2018 06:13:45: 		Build SHA1: b7b3e4fb3ff0f69024ce19a19b8f2780fb63078b
01/17/2018 06:13:45: 		MPI distribution: Open MPI
01/17/2018 06:13:45: 		MPI version: 1.10.7
01/17/2018 06:13:45: -------------------------------------------------------------------
01/17/2018 06:13:45: -------------------------------------------------------------------
01/17/2018 06:13:45: GPU info:

01/17/2018 06:13:45: 		Device[0]: cores = 3072; computeCapability = 5.2; type = "Tesla M60"; total memory = 8123 MB; free memory = 8112 MB
01/17/2018 06:13:45: -------------------------------------------------------------------

Configuration After Processing and Variable Resolution:

configparameters: cntk_sequence.cntk:addLayer2=[    
    action = "edit"
    currLayer = 1
    newLayer = 2
    currModel = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre1/cntkSpeech"
    newModel  = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre2/cntkSpeech.0"
    editPath  = "/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/add_layer.mel"
]

configparameters: cntk_sequence.cntk:AddLayer3=[    
    action = "edit"
    currLayer = 2
    newLayer = 3
    currModel = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre2/cntkSpeech"
    newModel  = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.0"
    editPath  = "/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/add_layer.mel"
]

configparameters: cntk_sequence.cntk:command=dptPre1:addLayer2:dptPre2:addLayer3:speechTrain:replaceCriterionNode:sequenceTrain
configparameters: cntk_sequence.cntk:ConfigDir=/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining
configparameters: cntk_sequence.cntk:currentDirectory=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData
configparameters: cntk_sequence.cntk:DataDir=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData
configparameters: cntk_sequence.cntk:deviceId=0
configparameters: cntk_sequence.cntk:dptPre1=[
    action = "train"
    modelPath = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre1/cntkSpeech"
    NDLNetworkBuilder = [
        networkDescription = "/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/dnn_1layer.txt"
    ]
]

configparameters: cntk_sequence.cntk:dptPre2=[
    action = "train"
    modelPath = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre2/cntkSpeech"
    NDLNetworkBuilder = [
        networkDescription = "/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/dnn_1layer.txt"
    ]
]

configparameters: cntk_sequence.cntk:globalInvStdPath=GlobalStats/var.363
configparameters: cntk_sequence.cntk:globalMeanPath=GlobalStats/mean.363
configparameters: cntk_sequence.cntk:globalPriorPath=GlobalStats/prior.132
configparameters: cntk_sequence.cntk:ndlMacros=/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/macros.txt
configparameters: cntk_sequence.cntk:OutputDir=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu
configparameters: cntk_sequence.cntk:precision=float
configparameters: cntk_sequence.cntk:reader=[
    readerType = "HTKMLFReader"
    readMethod = "blockRandomize"
    miniBatchMode = "partial"
    randomize = "auto"
    verbosity = 0
    features = [
        dim = 363
        type = "real"
        scpFile = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.scp"
    ]
    labels = [
        mlfFile = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.mlf"
        labelMappingFile = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/state.list"
        labelDim = 132
        labelType = "category"
    ]
]

configparameters: cntk_sequence.cntk:replaceCriterionNode=[
    action = "edit"
    currModel = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech"
    newModel  = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.sequence.0"
    editPath  = "/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/replace_ce_with_sequence_criterion.mel"
]

configparameters: cntk_sequence.cntk:RunDir=/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu
configparameters: cntk_sequence.cntk:sequenceTrain=[
    action = "train"
    modelPath = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.sequence"
    traceLevel = 1
    NDLNetworkBuilder = [
        networkDescription = "/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/nonexistentfile.txt"
    ]
    SGD = [
        epochSize = 81920
        minibatchSize = 10
        learningRatesPerSample = 0.000002
        momentumPerSample = 0.999589
        dropoutRate = 0.0
        maxEpochs = 3
        hsmoothingWeight = 0.95
        frameDropThresh = 1e-10
        numMBsToShowResult = 10
        gradientClippingWithTruncation = true
        clippingThresholdPerSample = 1.0
    ]
    reader = [
        readerType = "HTKMLFReader"
        readMethod = "blockRandomize"
        useMersenneTwisterRand=true      
        frameMode = false
        nbruttsineachrecurrentiter = 1
        miniBatchMode = "partial"
        randomize = "auto"
        verbosity = 0
        features = [
            dim = 363
            type = "real"
            scpFile = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.scp"
        ]
        labels = [
            mlfFile = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.mlf"
            labelMappingFile = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/state.list"
            labelDim = 132
            labelType = "category"
        ]
        hmms = [
            phoneFile  = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/model.overalltying"
            transpFile = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/model.transprob"
        ]
        lattices = [
            denlatTocFile = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/*.lats.toc"
        ]
    ]
]

configparameters: cntk_sequence.cntk:SGD=[
    epochSize = 81920
    minibatchSize = 256
    learningRatesPerMB = 0.8
    numMBsToShowResult = 10
    momentumPerMB = 0.9
    dropoutRate = 0.0
    maxEpochs = 2
]

configparameters: cntk_sequence.cntk:speechTrain=[
    action = "train"
    modelPath = "/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech"
    traceLevel = 1
    NDLNetworkBuilder = [
        networkDescription = "/home/ubuntu/workspace/Tests/EndToEndTests/Speech/DNN/SequenceTraining/dnn.txt"
    ]
    SGD = [
        epochSize = 81920
        minibatchSize = 256:512
        learningRatesPerMB = 0.8:1.6
        numMBsToShowResult = 10
        momentumPerSample = 0.999589
        dropoutRate = 0.0
        maxEpochs = 4
        gradUpdateType = "none"
        normWithAveMultiplier = true
        clippingThresholdPerSample = 1#INF
    ]
]

configparameters: cntk_sequence.cntk:timestamping=true
configparameters: cntk_sequence.cntk:traceLevel=1
configparameters: cntk_sequence.cntk:truncated=false
01/17/2018 06:13:45: Commands: dptPre1 addLayer2 dptPre2 addLayer3 speechTrain replaceCriterionNode sequenceTrain
01/17/2018 06:13:45: precision = "float"

01/17/2018 06:13:45: ##############################################################################
01/17/2018 06:13:45: #                                                                            #
01/17/2018 06:13:45: # dptPre1 command (train action)                                             #
01/17/2018 06:13:45: #                                                                            #
01/17/2018 06:13:45: ##############################################################################

01/17/2018 06:13:45: 
Creating virgin network.
NDLBuilder Using GPU 0
SetUniformRandomValue (GPU): creating curand object with seed 1, sizeof(ElemType)==4
reading script file /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.scp ... 948 entries
total 132 state names in state list /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/state.list
htkmlfreader: reading MLF file /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.mlf ... total 948 entries
...............................................................................................feature set 0: 252734 frames in 948 out of 948 utterances
label set 0: 129 classes
minibatchutterancesource: 948 utterances grouped into 3 chunks, av. chunk size: 316.0 utterances, 84244.7 frames
01/17/2018 06:13:45: 
Model has 19 nodes. Using GPU 0.

01/17/2018 06:13:45: Training criterion:   ce = CrossEntropyWithSoftmax
01/17/2018 06:13:45: Evaluation criterion: err = ClassificationError


Allocating matrices for forward and/or backward propagation.

Gradient Memory Aliasing: 2 are aliased.
	OL.t (gradient) reuses OL.z (gradient)

Memory Sharing: Out of 29 matrices, 12 are shared as 3, and 17 are not shared.

Here are the ones that share memory:
	{ HL1.t : [512 x *]
	  HL1.t : [512 x *] (gradient)
	  HL1.y : [512 x 1 x *] (gradient)
	  OL.z : [132 x 1 x *] }
	{ HL1.W : [512 x 363] (gradient)
	  HL1.z : [512 x 1 x *]
	  HL1.z : [512 x 1 x *] (gradient)
	  OL.t : [132 x 1 x *]
	  OL.t : [132 x 1 x *] (gradient)
	  OL.z : [132 x 1 x *] (gradient) }
	{ HL1.b : [512 x 1] (gradient)
	  HL1.y : [512 x 1 x *] }

Here are the ones that don't share memory:
	{scaledLogLikelihood : [132 x 1 x *]}
	{featNorm : [363 x *]}
	{ce : [1] (gradient)}
	{logPrior : [132 x 1]}
	{OL.b : [132 x 1] (gradient)}
	{err : [1]}
	{ce : [1]}
	{OL.W : [132 x 512] (gradient)}
	{globalPrior : [132 x 1]}
	{HL1.W : [512 x 363]}
	{HL1.b : [512 x 1]}
	{OL.W : [132 x 512]}
	{OL.b : [132 x 1]}
	{globalInvStd : [363 x 1]}
	{globalMean : [363 x 1]}
	{labels : [132 x *]}
	{features : [363 x *]}


01/17/2018 06:13:45: Training 254084 parameters in 4 out of 4 parameter tensors and 10 nodes with gradient:

01/17/2018 06:13:45: 	Node 'HL1.W' (LearnableParameter operation) : [512 x 363]
01/17/2018 06:13:45: 	Node 'HL1.b' (LearnableParameter operation) : [512 x 1]
01/17/2018 06:13:45: 	Node 'OL.W' (LearnableParameter operation) : [132 x 512]
01/17/2018 06:13:45: 	Node 'OL.b' (LearnableParameter operation) : [132 x 1]

01/17/2018 06:13:45: No PreCompute nodes found, or all already computed. Skipping pre-computation step.

01/17/2018 06:13:45: Starting Epoch 1: learning rate per sample = 0.003125  effective momentum = 0.900000  momentum as time constant = 2429.8 samples
minibatchiterator: epoch 0: frames [0..81920] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 10.0 ms

01/17/2018 06:13:45: Starting minibatch loop.
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[   1-  10, 3.12%]: ce = 3.74183846 * 2560; err = 0.80195313 * 2560; time = 0.2218s; samplesPerSecond = 11539.7
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  11-  20, 6.25%]: ce = 2.91124763 * 2560; err = 0.70898438 * 2560; time = 0.0069s; samplesPerSecond = 372271.6
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  21-  30, 9.38%]: ce = 2.58015900 * 2560; err = 0.66640625 * 2560; time = 0.0072s; samplesPerSecond = 356456.6
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  31-  40, 12.50%]: ce = 2.27427139 * 2560; err = 0.58750000 * 2560; time = 0.0067s; samplesPerSecond = 382614.9
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  41-  50, 15.62%]: ce = 2.05503540 * 2560; err = 0.56093750 * 2560; time = 0.0066s; samplesPerSecond = 387087.0
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  51-  60, 18.75%]: ce = 1.91055145 * 2560; err = 0.52812500 * 2560; time = 0.0066s; samplesPerSecond = 386642.7
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  61-  70, 21.88%]: ce = 1.81562805 * 2560; err = 0.51171875 * 2560; time = 0.0065s; samplesPerSecond = 391162.2
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  71-  80, 25.00%]: ce = 1.68803253 * 2560; err = 0.48476562 * 2560; time = 0.0070s; samplesPerSecond = 367162.0
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  81-  90, 28.12%]: ce = 1.57382050 * 2560; err = 0.45429687 * 2560; time = 0.0068s; samplesPerSecond = 377809.6
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  91- 100, 31.25%]: ce = 1.62090302 * 2560; err = 0.47304687 * 2560; time = 0.0070s; samplesPerSecond = 367668.2
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 101- 110, 34.38%]: ce = 1.59272614 * 2560; err = 0.47500000 * 2560; time = 0.0065s; samplesPerSecond = 393555.5
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 111- 120, 37.50%]: ce = 1.51520386 * 2560; err = 0.44531250 * 2560; time = 0.0067s; samplesPerSecond = 380284.6
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 121- 130, 40.62%]: ce = 1.49181976 * 2560; err = 0.45039062 * 2560; time = 0.0066s; samplesPerSecond = 386923.2
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 131- 140, 43.75%]: ce = 1.53703613 * 2560; err = 0.44804688 * 2560; time = 0.0065s; samplesPerSecond = 394040.1
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 141- 150, 46.88%]: ce = 1.43095398 * 2560; err = 0.41640625 * 2560; time = 0.0068s; samplesPerSecond = 377926.7
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 151- 160, 50.00%]: ce = 1.41503601 * 2560; err = 0.40078125 * 2560; time = 0.0065s; samplesPerSecond = 394398.3
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 161- 170, 53.12%]: ce = 1.38912964 * 2560; err = 0.41132812 * 2560; time = 0.0069s; samplesPerSecond = 373504.5
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 171- 180, 56.25%]: ce = 1.41208496 * 2560; err = 0.42226562 * 2560; time = 0.0065s; samplesPerSecond = 396825.4
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 181- 190, 59.38%]: ce = 1.39966125 * 2560; err = 0.40664062 * 2560; time = 0.0068s; samplesPerSecond = 376581.3
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 191- 200, 62.50%]: ce = 1.42728271 * 2560; err = 0.42617187 * 2560; time = 0.0067s; samplesPerSecond = 381952.7
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 201- 210, 65.62%]: ce = 1.41336060 * 2560; err = 0.42304687 * 2560; time = 0.0065s; samplesPerSecond = 393924.9
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 211- 220, 68.75%]: ce = 1.33200073 * 2560; err = 0.39960937 * 2560; time = 0.0067s; samplesPerSecond = 379951.6
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 221- 230, 71.88%]: ce = 1.28576965 * 2560; err = 0.38671875 * 2560; time = 0.0065s; samplesPerSecond = 392373.2
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 231- 240, 75.00%]: ce = 1.34133301 * 2560; err = 0.40937500 * 2560; time = 0.0068s; samplesPerSecond = 378496.7
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 241- 250, 78.12%]: ce = 1.32666321 * 2560; err = 0.39609375 * 2560; time = 0.0065s; samplesPerSecond = 392987.6
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 251- 260, 81.25%]: ce = 1.21424866 * 2560; err = 0.37226562 * 2560; time = 0.0070s; samplesPerSecond = 367399.1
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 261- 270, 84.38%]: ce = 1.23750610 * 2560; err = 0.37382813 * 2560; time = 0.0065s; samplesPerSecond = 393840.1
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 271- 280, 87.50%]: ce = 1.29965820 * 2560; err = 0.39062500 * 2560; time = 0.0065s; samplesPerSecond = 391940.7
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 281- 290, 90.62%]: ce = 1.21221924 * 2560; err = 0.37382813 * 2560; time = 0.0068s; samplesPerSecond = 377988.1
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 291- 300, 93.75%]: ce = 1.20538635 * 2560; err = 0.36757812 * 2560; time = 0.0067s; samplesPerSecond = 384217.1
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 301- 310, 96.88%]: ce = 1.23562927 * 2560; err = 0.37187500 * 2560; time = 0.0069s; samplesPerSecond = 371439.8
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 311- 320, 100.00%]: ce = 1.25470886 * 2560; err = 0.37812500 * 2560; time = 0.0065s; samplesPerSecond = 394276.8
01/17/2018 06:13:46: Finished Epoch[ 1 of 2]: [Training] ce = 1.62940331 * 81920; err = 0.46009521 * 81920; totalSamplesSeen = 81920; learningRatePerSample = 0.003125; epochTime=0.543902s
01/17/2018 06:13:46: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre1/cntkSpeech.1'

01/17/2018 06:13:46: Starting Epoch 2: learning rate per sample = 0.003125  effective momentum = 0.900000  momentum as time constant = 2429.8 samples
minibatchiterator: epoch 1: frames [81920..163840] (first utterance at frame 81920), data subset 0 of 1, with 1 datapasses

01/17/2018 06:13:46: Starting minibatch loop.
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[   1-  10, 3.12%]: ce = 1.23162079 * 2560; err = 0.38125000 * 2560; time = 0.0078s; samplesPerSecond = 329786.4
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[  11-  20, 6.25%]: ce = 1.20301991 * 2560; err = 0.37226562 * 2560; time = 0.0069s; samplesPerSecond = 370375.7
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[  21-  30, 9.38%]: ce = 1.28580151 * 2560; err = 0.37851563 * 2560; time = 0.0069s; samplesPerSecond = 371660.9
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[  31-  40, 12.50%]: ce = 1.23043137 * 2560; err = 0.37460938 * 2560; time = 0.0070s; samplesPerSecond = 366851.5
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[  41-  50, 15.62%]: ce = 1.18316193 * 2560; err = 0.35429688 * 2560; time = 0.0070s; samplesPerSecond = 363141.2
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[  51-  60, 18.75%]: ce = 1.27994614 * 2560; err = 0.37812500 * 2560; time = 0.0067s; samplesPerSecond = 384303.6
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[  61-  70, 21.88%]: ce = 1.22171936 * 2560; err = 0.37070313 * 2560; time = 0.0069s; samplesPerSecond = 372352.8
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[  71-  80, 25.00%]: ce = 1.17933273 * 2560; err = 0.36250000 * 2560; time = 0.0068s; samplesPerSecond = 376194.0
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[  81-  90, 28.12%]: ce = 1.23844833 * 2560; err = 0.36289063 * 2560; time = 0.0071s; samplesPerSecond = 362842.6
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[  91- 100, 31.25%]: ce = 1.18221588 * 2560; err = 0.37460938 * 2560; time = 0.0069s; samplesPerSecond = 372385.3
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 101- 110, 34.38%]: ce = 1.19557495 * 2560; err = 0.36093750 * 2560; time = 0.0073s; samplesPerSecond = 349426.0
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 111- 120, 37.50%]: ce = 1.18080750 * 2560; err = 0.35078125 * 2560; time = 0.0067s; samplesPerSecond = 379737.4
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 121- 130, 40.62%]: ce = 1.16538544 * 2560; err = 0.35820313 * 2560; time = 0.0066s; samplesPerSecond = 385298.5
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 131- 140, 43.75%]: ce = 1.13251953 * 2560; err = 0.35039063 * 2560; time = 0.0070s; samplesPerSecond = 366127.5
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 141- 150, 46.88%]: ce = 1.09806366 * 2560; err = 0.32539062 * 2560; time = 0.0067s; samplesPerSecond = 380154.7
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 151- 160, 50.00%]: ce = 1.10407715 * 2560; err = 0.33984375 * 2560; time = 0.0069s; samplesPerSecond = 370681.4
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 161- 170, 53.12%]: ce = 1.20419312 * 2560; err = 0.36054687 * 2560; time = 0.0066s; samplesPerSecond = 387403.3
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 171- 180, 56.25%]: ce = 1.17373505 * 2560; err = 0.35781250 * 2560; time = 0.0070s; samplesPerSecond = 363651.9
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 181- 190, 59.38%]: ce = 1.12243347 * 2560; err = 0.34609375 * 2560; time = 0.0067s; samplesPerSecond = 382072.4
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 191- 200, 62.50%]: ce = 1.12005615 * 2560; err = 0.35625000 * 2560; time = 0.0069s; samplesPerSecond = 372661.8
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 201- 210, 65.62%]: ce = 1.10305176 * 2560; err = 0.33046875 * 2560; time = 0.0067s; samplesPerSecond = 381423.5
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 211- 220, 68.75%]: ce = 1.13120422 * 2560; err = 0.34257813 * 2560; time = 0.0066s; samplesPerSecond = 389602.5
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 221- 230, 71.88%]: ce = 1.14404602 * 2560; err = 0.35390625 * 2560; time = 0.0073s; samplesPerSecond = 350339.4
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 231- 240, 75.00%]: ce = 1.28562622 * 2560; err = 0.39414063 * 2560; time = 0.0066s; samplesPerSecond = 388986.8
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 241- 250, 78.12%]: ce = 1.17830811 * 2560; err = 0.35585937 * 2560; time = 0.0068s; samplesPerSecond = 375086.1
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 251- 260, 81.25%]: ce = 1.12961731 * 2560; err = 0.35820313 * 2560; time = 0.0065s; samplesPerSecond = 393525.3
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 261- 270, 84.38%]: ce = 1.13842163 * 2560; err = 0.34843750 * 2560; time = 0.0068s; samplesPerSecond = 376100.0
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 271- 280, 87.50%]: ce = 1.14543152 * 2560; err = 0.34648438 * 2560; time = 0.0065s; samplesPerSecond = 392259.0
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 281- 290, 90.62%]: ce = 1.06640625 * 2560; err = 0.33203125 * 2560; time = 0.0065s; samplesPerSecond = 395727.4
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 291- 300, 93.75%]: ce = 1.10130005 * 2560; err = 0.33593750 * 2560; time = 0.0067s; samplesPerSecond = 382637.8
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 301- 310, 96.88%]: ce = 1.08510742 * 2560; err = 0.33750000 * 2560; time = 0.0065s; samplesPerSecond = 392343.2
01/17/2018 06:13:46:  Epoch[ 2 of 2]-Minibatch[ 311- 320, 100.00%]: ce = 1.06571045 * 2560; err = 0.33515625 * 2560; time = 0.0067s; samplesPerSecond = 384927.7
01/17/2018 06:13:46: Finished Epoch[ 2 of 2]: [Training] ce = 1.16583672 * 81920; err = 0.35583496 * 81920; totalSamplesSeen = 163840; learningRatePerSample = 0.003125; epochTime=0.222587s
01/17/2018 06:13:46: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre1/cntkSpeech'

01/17/2018 06:13:46: Action "train" complete.


01/17/2018 06:13:46: ##############################################################################
01/17/2018 06:13:46: #                                                                            #
01/17/2018 06:13:46: # addLayer2 command (edit action)                                            #
01/17/2018 06:13:46: #                                                                            #
01/17/2018 06:13:46: ##############################################################################


01/17/2018 06:13:46: Action "edit" complete.


01/17/2018 06:13:46: ##############################################################################
01/17/2018 06:13:46: #                                                                            #
01/17/2018 06:13:46: # dptPre2 command (train action)                                             #
01/17/2018 06:13:46: #                                                                            #
01/17/2018 06:13:46: ##############################################################################

01/17/2018 06:13:46: 
Starting from checkpoint. Loading network from '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre2/cntkSpeech.0'.
NDLBuilder Using GPU 0
reading script file /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.scp ... 948 entries
total 132 state names in state list /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/state.list
htkmlfreader: reading MLF file /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.mlf ... total 948 entries
...............................................................................................feature set 0: 252734 frames in 948 out of 948 utterances
label set 0: 129 classes
minibatchutterancesource: 948 utterances grouped into 3 chunks, av. chunk size: 316.0 utterances, 84244.7 frames
01/17/2018 06:13:46: 
Model has 24 nodes. Using GPU 0.

01/17/2018 06:13:46: Training criterion:   ce = CrossEntropyWithSoftmax
01/17/2018 06:13:46: Evaluation criterion: err = ClassificationError

01/17/2018 06:13:46: Training 516740 parameters in 6 out of 6 parameter tensors and 15 nodes with gradient:

01/17/2018 06:13:46: 	Node 'HL1.W' (LearnableParameter operation) : [512 x 363]
01/17/2018 06:13:46: 	Node 'HL1.b' (LearnableParameter operation) : [512 x 1]
01/17/2018 06:13:46: 	Node 'HL2.W' (LearnableParameter operation) : [512 x 512]
01/17/2018 06:13:46: 	Node 'HL2.b' (LearnableParameter operation) : [512 x 1]
01/17/2018 06:13:46: 	Node 'OL.W' (LearnableParameter operation) : [132 x 512]
01/17/2018 06:13:46: 	Node 'OL.b' (LearnableParameter operation) : [132 x 1]

01/17/2018 06:13:46: No PreCompute nodes found, or all already computed. Skipping pre-computation step.

01/17/2018 06:13:46: Starting Epoch 1: learning rate per sample = 0.003125  effective momentum = 0.900000  momentum as time constant = 2429.8 samples
minibatchiterator: epoch 0: frames [0..81920] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 10.0 ms

01/17/2018 06:13:46: Starting minibatch loop.
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[   1-  10, 3.12%]: ce = 4.61674881 * 2560; err = 0.80742187 * 2560; time = 0.0110s; samplesPerSecond = 233730.2
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  11-  20, 6.25%]: ce = 2.86666870 * 2560; err = 0.70507812 * 2560; time = 0.0085s; samplesPerSecond = 301375.0
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  21-  30, 9.38%]: ce = 2.29427795 * 2560; err = 0.59960938 * 2560; time = 0.0086s; samplesPerSecond = 297477.3
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  31-  40, 12.50%]: ce = 1.96351089 * 2560; err = 0.52851563 * 2560; time = 0.0085s; samplesPerSecond = 301321.8
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  41-  50, 15.62%]: ce = 1.74446564 * 2560; err = 0.48007813 * 2560; time = 0.0087s; samplesPerSecond = 293507.3
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  51-  60, 18.75%]: ce = 1.62170563 * 2560; err = 0.45546875 * 2560; time = 0.0085s; samplesPerSecond = 301336.0
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  61-  70, 21.88%]: ce = 1.57501984 * 2560; err = 0.45546875 * 2560; time = 0.0086s; samplesPerSecond = 296835.7
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  71-  80, 25.00%]: ce = 1.47702789 * 2560; err = 0.42773438 * 2560; time = 0.0085s; samplesPerSecond = 301730.2
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  81-  90, 28.12%]: ce = 1.38880768 * 2560; err = 0.40156250 * 2560; time = 0.0087s; samplesPerSecond = 295254.0
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[  91- 100, 31.25%]: ce = 1.42063293 * 2560; err = 0.42773438 * 2560; time = 0.0085s; samplesPerSecond = 299758.8
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 101- 110, 34.38%]: ce = 1.41058807 * 2560; err = 0.43789062 * 2560; time = 0.0089s; samplesPerSecond = 289118.5
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 111- 120, 37.50%]: ce = 1.38001099 * 2560; err = 0.41445312 * 2560; time = 0.0086s; samplesPerSecond = 298090.4
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 121- 130, 40.62%]: ce = 1.34645538 * 2560; err = 0.41250000 * 2560; time = 0.0085s; samplesPerSecond = 302121.9
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 131- 140, 43.75%]: ce = 1.38398743 * 2560; err = 0.40195313 * 2560; time = 0.0087s; samplesPerSecond = 294374.7
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 141- 150, 46.88%]: ce = 1.32409363 * 2560; err = 0.38984375 * 2560; time = 0.0084s; samplesPerSecond = 303080.5
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 151- 160, 50.00%]: ce = 1.31575928 * 2560; err = 0.39414063 * 2560; time = 0.0091s; samplesPerSecond = 280717.1
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 161- 170, 53.12%]: ce = 1.25869446 * 2560; err = 0.37148437 * 2560; time = 0.0085s; samplesPerSecond = 302346.7
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 171- 180, 56.25%]: ce = 1.27994385 * 2560; err = 0.38398437 * 2560; time = 0.0087s; samplesPerSecond = 294107.5
01/17/2018 06:13:46:  Epoch[ 1 of 2]-Minibatch[ 181- 190, 59.38%]: ce = 1.29792175 * 2560; err = 0.39335938 * 2560; time = 0.0085s; samplesPerSecond = 299495.8
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 191- 200, 62.50%]: ce = 1.28697815 * 2560; err = 0.39843750 * 2560; time = 0.0085s; samplesPerSecond = 301052.5
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 201- 210, 65.62%]: ce = 1.26717834 * 2560; err = 0.38593750 * 2560; time = 0.0086s; samplesPerSecond = 298601.5
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 211- 220, 68.75%]: ce = 1.21615295 * 2560; err = 0.36718750 * 2560; time = 0.0085s; samplesPerSecond = 302496.8
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 221- 230, 71.88%]: ce = 1.21445923 * 2560; err = 0.37031250 * 2560; time = 0.0087s; samplesPerSecond = 295366.4
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 231- 240, 75.00%]: ce = 1.25004578 * 2560; err = 0.38085938 * 2560; time = 0.0085s; samplesPerSecond = 302243.2
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 241- 250, 78.12%]: ce = 1.22538452 * 2560; err = 0.37656250 * 2560; time = 0.0086s; samplesPerSecond = 297975.9
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 251- 260, 81.25%]: ce = 1.15360413 * 2560; err = 0.34843750 * 2560; time = 0.0084s; samplesPerSecond = 303331.9
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 261- 270, 84.38%]: ce = 1.16656189 * 2560; err = 0.35312500 * 2560; time = 0.0086s; samplesPerSecond = 297716.0
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 271- 280, 87.50%]: ce = 1.22569275 * 2560; err = 0.36640625 * 2560; time = 0.0085s; samplesPerSecond = 302511.1
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 281- 290, 90.62%]: ce = 1.16463623 * 2560; err = 0.36054687 * 2560; time = 0.0088s; samplesPerSecond = 291028.1
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 291- 300, 93.75%]: ce = 1.16964111 * 2560; err = 0.35351562 * 2560; time = 0.0086s; samplesPerSecond = 297965.5
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 301- 310, 96.88%]: ce = 1.16557617 * 2560; err = 0.35351562 * 2560; time = 0.0085s; samplesPerSecond = 302286.0
01/17/2018 06:13:47:  Epoch[ 1 of 2]-Minibatch[ 311- 320, 100.00%]: ce = 1.17247925 * 2560; err = 0.35156250 * 2560; time = 0.0088s; samplesPerSecond = 291914.2
01/17/2018 06:13:47: Finished Epoch[ 1 of 2]: [Training] ce = 1.52014723 * 81920; err = 0.42670898 * 81920; totalSamplesSeen = 81920; learningRatePerSample = 0.003125; epochTime=0.394854s
01/17/2018 06:13:47: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre2/cntkSpeech.1'

01/17/2018 06:13:47: Starting Epoch 2: learning rate per sample = 0.003125  effective momentum = 0.900000  momentum as time constant = 2429.8 samples
minibatchiterator: epoch 1: frames [81920..163840] (first utterance at frame 81920), data subset 0 of 1, with 1 datapasses

01/17/2018 06:13:47: Starting minibatch loop.
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[   1-  10, 3.12%]: ce = 1.14981880 * 2560; err = 0.35156250 * 2560; time = 0.0097s; samplesPerSecond = 262680.0
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[  11-  20, 6.25%]: ce = 1.17322617 * 2560; err = 0.36015625 * 2560; time = 0.0092s; samplesPerSecond = 279769.2
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[  21-  30, 9.38%]: ce = 1.22602234 * 2560; err = 0.37460938 * 2560; time = 0.0085s; samplesPerSecond = 301141.0
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[  31-  40, 12.50%]: ce = 1.18246918 * 2560; err = 0.36015625 * 2560; time = 0.0086s; samplesPerSecond = 297605.2
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[  41-  50, 15.62%]: ce = 1.13529053 * 2560; err = 0.34453125 * 2560; time = 0.0085s; samplesPerSecond = 301716.0
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[  51-  60, 18.75%]: ce = 1.21815300 * 2560; err = 0.36640625 * 2560; time = 0.0086s; samplesPerSecond = 297169.9
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[  61-  70, 21.88%]: ce = 1.14050827 * 2560; err = 0.34140625 * 2560; time = 0.0085s; samplesPerSecond = 301307.6
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[  71-  80, 25.00%]: ce = 1.12378693 * 2560; err = 0.35312500 * 2560; time = 0.0088s; samplesPerSecond = 292431.1
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[  81-  90, 28.12%]: ce = 1.14636002 * 2560; err = 0.33906250 * 2560; time = 0.0086s; samplesPerSecond = 297826.8
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[  91- 100, 31.25%]: ce = 1.12752228 * 2560; err = 0.34843750 * 2560; time = 0.0085s; samplesPerSecond = 300649.4
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 101- 110, 34.38%]: ce = 1.14752197 * 2560; err = 0.34414062 * 2560; time = 0.0085s; samplesPerSecond = 302500.4
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 111- 120, 37.50%]: ce = 1.12730484 * 2560; err = 0.34140625 * 2560; time = 0.0085s; samplesPerSecond = 302075.6
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 121- 130, 40.62%]: ce = 1.11186981 * 2560; err = 0.34179688 * 2560; time = 0.0088s; samplesPerSecond = 290780.2
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 131- 140, 43.75%]: ce = 1.07041931 * 2560; err = 0.32617188 * 2560; time = 0.0085s; samplesPerSecond = 301183.6
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 141- 150, 46.88%]: ce = 1.05150299 * 2560; err = 0.31250000 * 2560; time = 0.0087s; samplesPerSecond = 294090.6
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 151- 160, 50.00%]: ce = 1.06874542 * 2560; err = 0.33007812 * 2560; time = 0.0085s; samplesPerSecond = 302682.8
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 161- 170, 53.12%]: ce = 1.14110870 * 2560; err = 0.34687500 * 2560; time = 0.0086s; samplesPerSecond = 299096.9
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 171- 180, 56.25%]: ce = 1.13898926 * 2560; err = 0.36132812 * 2560; time = 0.0086s; samplesPerSecond = 299030.5
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 181- 190, 59.38%]: ce = 1.08064728 * 2560; err = 0.33437500 * 2560; time = 0.0085s; samplesPerSecond = 301318.3
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 191- 200, 62.50%]: ce = 1.07247162 * 2560; err = 0.33984375 * 2560; time = 0.0087s; samplesPerSecond = 293689.1
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 201- 210, 65.62%]: ce = 1.06161499 * 2560; err = 0.32539062 * 2560; time = 0.0085s; samplesPerSecond = 302075.6
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 211- 220, 68.75%]: ce = 1.09126740 * 2560; err = 0.33242187 * 2560; time = 0.0086s; samplesPerSecond = 297882.2
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 221- 230, 71.88%]: ce = 1.11266785 * 2560; err = 0.34492187 * 2560; time = 0.0085s; samplesPerSecond = 302836.7
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 231- 240, 75.00%]: ce = 1.12638855 * 2560; err = 0.35273437 * 2560; time = 0.0086s; samplesPerSecond = 297958.5
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 241- 250, 78.12%]: ce = 1.08986816 * 2560; err = 0.33984375 * 2560; time = 0.0087s; samplesPerSecond = 295676.9
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 251- 260, 81.25%]: ce = 1.06911316 * 2560; err = 0.33398438 * 2560; time = 0.0088s; samplesPerSecond = 292280.8
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 261- 270, 84.38%]: ce = 1.06766663 * 2560; err = 0.32460937 * 2560; time = 0.0085s; samplesPerSecond = 302629.1
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 271- 280, 87.50%]: ce = 1.09992981 * 2560; err = 0.33203125 * 2560; time = 0.0085s; samplesPerSecond = 301872.6
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 281- 290, 90.62%]: ce = 1.02154846 * 2560; err = 0.32539062 * 2560; time = 0.0087s; samplesPerSecond = 294273.2
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 291- 300, 93.75%]: ce = 1.07519226 * 2560; err = 0.33281250 * 2560; time = 0.0085s; samplesPerSecond = 302718.6
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 301- 310, 96.88%]: ce = 1.06713867 * 2560; err = 0.32812500 * 2560; time = 0.0086s; samplesPerSecond = 299380.2
01/17/2018 06:13:47:  Epoch[ 2 of 2]-Minibatch[ 311- 320, 100.00%]: ce = 1.05164185 * 2560; err = 0.32890625 * 2560; time = 0.0085s; samplesPerSecond = 301776.5
01/17/2018 06:13:47: Finished Epoch[ 2 of 2]: [Training] ce = 1.11149302 * 81920; err = 0.34122314 * 81920; totalSamplesSeen = 163840; learningRatePerSample = 0.003125; epochTime=0.279821s
01/17/2018 06:13:47: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/Pre2/cntkSpeech'

01/17/2018 06:13:47: Action "train" complete.


01/17/2018 06:13:47: ##############################################################################
01/17/2018 06:13:47: #                                                                            #
01/17/2018 06:13:47: # addLayer3 command (edit action)                                            #
01/17/2018 06:13:47: #                                                                            #
01/17/2018 06:13:47: ##############################################################################


01/17/2018 06:13:47: Action "edit" complete.


01/17/2018 06:13:47: ##############################################################################
01/17/2018 06:13:47: #                                                                            #
01/17/2018 06:13:47: # speechTrain command (train action)                                         #
01/17/2018 06:13:47: #                                                                            #
01/17/2018 06:13:47: ##############################################################################

01/17/2018 06:13:47: 
Starting from checkpoint. Loading network from '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.0'.
NDLBuilder Using GPU 0
reading script file /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.scp ... 948 entries
total 132 state names in state list /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/state.list
htkmlfreader: reading MLF file /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.mlf ... total 948 entries
...............................................................................................feature set 0: 252734 frames in 948 out of 948 utterances
label set 0: 129 classes
minibatchutterancesource: 948 utterances grouped into 3 chunks, av. chunk size: 316.0 utterances, 84244.7 frames
01/17/2018 06:13:47: 
Model has 29 nodes. Using GPU 0.

01/17/2018 06:13:47: Training criterion:   ce = CrossEntropyWithSoftmax
01/17/2018 06:13:47: Evaluation criterion: err = ClassificationError

01/17/2018 06:13:47: Training 779396 parameters in 8 out of 8 parameter tensors and 20 nodes with gradient:

01/17/2018 06:13:47: 	Node 'HL1.W' (LearnableParameter operation) : [512 x 363]
01/17/2018 06:13:47: 	Node 'HL1.b' (LearnableParameter operation) : [512 x 1]
01/17/2018 06:13:47: 	Node 'HL2.W' (LearnableParameter operation) : [512 x 512]
01/17/2018 06:13:47: 	Node 'HL2.b' (LearnableParameter operation) : [512 x 1]
01/17/2018 06:13:47: 	Node 'HL3.W' (LearnableParameter operation) : [512 x 512]
01/17/2018 06:13:47: 	Node 'HL3.b' (LearnableParameter operation) : [512 x 1]
01/17/2018 06:13:47: 	Node 'OL.W' (LearnableParameter operation) : [132 x 512]
01/17/2018 06:13:47: 	Node 'OL.b' (LearnableParameter operation) : [132 x 1]

01/17/2018 06:13:47: No PreCompute nodes found, or all already computed. Skipping pre-computation step.

01/17/2018 06:13:47: Starting Epoch 1: learning rate per sample = 0.003125  effective momentum = 0.900117  momentum as time constant = 2432.7 samples
minibatchiterator: epoch 0: frames [0..81920] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 10.0 ms

01/17/2018 06:13:47: Starting minibatch loop.
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[   1-  10, 3.12%]: ce = 3.98869972 * 2560; err = 0.81562500 * 2560; time = 0.0138s; samplesPerSecond = 185394.4
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[  11-  20, 6.25%]: ce = 2.65266838 * 2560; err = 0.64531250 * 2560; time = 0.0110s; samplesPerSecond = 232761.1
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[  21-  30, 9.38%]: ce = 2.04071579 * 2560; err = 0.54687500 * 2560; time = 0.0110s; samplesPerSecond = 231751.8
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[  31-  40, 12.50%]: ce = 1.74825745 * 2560; err = 0.47539063 * 2560; time = 0.0111s; samplesPerSecond = 230917.7
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[  41-  50, 15.62%]: ce = 1.57756348 * 2560; err = 0.44921875 * 2560; time = 0.0110s; samplesPerSecond = 233342.4
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[  51-  60, 18.75%]: ce = 1.47807083 * 2560; err = 0.41835937 * 2560; time = 0.0110s; samplesPerSecond = 232973.0
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[  61-  70, 21.88%]: ce = 1.44050140 * 2560; err = 0.41015625 * 2560; time = 0.0114s; samplesPerSecond = 224176.0
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[  71-  80, 25.00%]: ce = 1.36226807 * 2560; err = 0.39726563 * 2560; time = 0.0110s; samplesPerSecond = 231697.3
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[  81-  90, 28.12%]: ce = 1.28130646 * 2560; err = 0.37578125 * 2560; time = 0.0109s; samplesPerSecond = 233798.5
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[  91- 100, 31.25%]: ce = 1.30515137 * 2560; err = 0.40195313 * 2560; time = 0.0109s; samplesPerSecond = 233965.2
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: ce = 1.28546295 * 2560; err = 0.38984375 * 2560; time = 0.0115s; samplesPerSecond = 223446.1
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: ce = 1.27684479 * 2560; err = 0.38281250 * 2560; time = 0.0112s; samplesPerSecond = 229144.3
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: ce = 1.24204254 * 2560; err = 0.38281250 * 2560; time = 0.0110s; samplesPerSecond = 233783.5
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: ce = 1.30829010 * 2560; err = 0.38320312 * 2560; time = 0.0109s; samplesPerSecond = 234147.1
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: ce = 1.24720459 * 2560; err = 0.36367187 * 2560; time = 0.0113s; samplesPerSecond = 226653.0
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: ce = 1.26371307 * 2560; err = 0.38242188 * 2560; time = 0.0109s; samplesPerSecond = 234294.9
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: ce = 1.20174866 * 2560; err = 0.36210938 * 2560; time = 0.0109s; samplesPerSecond = 234507.4
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: ce = 1.20651245 * 2560; err = 0.36718750 * 2560; time = 0.0112s; samplesPerSecond = 229193.5
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: ce = 1.21452942 * 2560; err = 0.36718750 * 2560; time = 0.0109s; samplesPerSecond = 233990.8
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: ce = 1.20404053 * 2560; err = 0.37617187 * 2560; time = 0.0114s; samplesPerSecond = 224752.6
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: ce = 1.20572510 * 2560; err = 0.36875000 * 2560; time = 0.0109s; samplesPerSecond = 234402.2
01/17/2018 06:13:47:  Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: ce = 1.14164734 * 2560; err = 0.34765625 * 2560; time = 0.0112s; samplesPerSecond = 229483.2
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: ce = 1.14932861 * 2560; err = 0.34921875 * 2560; time = 0.0109s; samplesPerSecond = 234412.9
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: ce = 1.18699341 * 2560; err = 0.35117188 * 2560; time = 0.0111s; samplesPerSecond = 230888.6
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: ce = 1.16585693 * 2560; err = 0.36054687 * 2560; time = 0.0113s; samplesPerSecond = 226640.9
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: ce = 1.08444214 * 2560; err = 0.33945313 * 2560; time = 0.0109s; samplesPerSecond = 234117.1
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: ce = 1.11162720 * 2560; err = 0.34023437 * 2560; time = 0.0111s; samplesPerSecond = 229608.8
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: ce = 1.17780457 * 2560; err = 0.34687500 * 2560; time = 0.0111s; samplesPerSecond = 231621.8
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: ce = 1.11032715 * 2560; err = 0.34062500 * 2560; time = 0.0113s; samplesPerSecond = 226258.3
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: ce = 1.13506470 * 2560; err = 0.34648438 * 2560; time = 0.0109s; samplesPerSecond = 234316.4
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: ce = 1.12134094 * 2560; err = 0.34101562 * 2560; time = 0.0109s; samplesPerSecond = 234309.9
01/17/2018 06:13:48:  Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: ce = 1.12438660 * 2560; err = 0.34335938 * 2560; time = 0.0112s; samplesPerSecond = 229119.7
01/17/2018 06:13:48: Finished Epoch[ 1 of 4]: [Training] ce = 1.40750427 * 81920; err = 0.40214844 * 81920; totalSamplesSeen = 81920; learningRatePerSample = 0.003125; epochTime=0.473469s
01/17/2018 06:13:48: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.1'

01/17/2018 06:13:48: Starting Epoch 2: learning rate per sample = 0.003125  effective momentum = 0.810210  momentum as time constant = 2432.7 samples
minibatchiterator: epoch 1: frames [81920..163840] (first utterance at frame 81920), data subset 0 of 1, with 1 datapasses

01/17/2018 06:13:48: Starting minibatch loop.
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[   1-  10, 6.25%]: ce = 1.46610394 * 5120; err = 0.40996094 * 5120; time = 0.0184s; samplesPerSecond = 278114.3
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[  11-  20, 12.50%]: ce = 1.50110569 * 5120; err = 0.41347656 * 5120; time = 0.0167s; samplesPerSecond = 306956.3
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[  21-  30, 18.75%]: ce = 1.21108513 * 5120; err = 0.36640625 * 5120; time = 0.0165s; samplesPerSecond = 310675.8
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[  31-  40, 25.00%]: ce = 1.12810822 * 5120; err = 0.34023437 * 5120; time = 0.0165s; samplesPerSecond = 310896.6
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[  41-  50, 31.25%]: ce = 1.11897316 * 5120; err = 0.33847656 * 5120; time = 0.0164s; samplesPerSecond = 311420.4
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[  51-  60, 37.50%]: ce = 1.13299255 * 5120; err = 0.34335938 * 5120; time = 0.0165s; samplesPerSecond = 311229.2
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[  61-  70, 43.75%]: ce = 1.08451233 * 5120; err = 0.33515625 * 5120; time = 0.0168s; samplesPerSecond = 304367.0
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[  71-  80, 50.00%]: ce = 1.07491379 * 5120; err = 0.32695313 * 5120; time = 0.0167s; samplesPerSecond = 306016.9
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[  81-  90, 56.25%]: ce = 1.14153519 * 5120; err = 0.35410156 * 5120; time = 0.0164s; samplesPerSecond = 311526.5
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[  91- 100, 62.50%]: ce = 1.06857758 * 5120; err = 0.33339844 * 5120; time = 0.0165s; samplesPerSecond = 310472.4
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[ 101- 110, 68.75%]: ce = 1.05950546 * 5120; err = 0.33046875 * 5120; time = 0.0165s; samplesPerSecond = 310162.0
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[ 111- 120, 75.00%]: ce = 1.13561249 * 5120; err = 0.35058594 * 5120; time = 0.0174s; samplesPerSecond = 294691.5
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[ 121- 130, 81.25%]: ce = 1.12639160 * 5120; err = 0.35410156 * 5120; time = 0.0216s; samplesPerSecond = 236795.9
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[ 131- 140, 87.50%]: ce = 1.10322723 * 5120; err = 0.33828125 * 5120; time = 0.0222s; samplesPerSecond = 230725.2
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[ 141- 150, 93.75%]: ce = 1.04754944 * 5120; err = 0.33144531 * 5120; time = 0.0168s; samplesPerSecond = 304774.6
01/17/2018 06:13:48:  Epoch[ 2 of 4]-Minibatch[ 151- 160, 100.00%]: ce = 1.05628357 * 5120; err = 0.32441406 * 5120; time = 0.0164s; samplesPerSecond = 311448.8
01/17/2018 06:13:48: Finished Epoch[ 2 of 4]: [Training] ce = 1.15352983 * 81920; err = 0.34942627 * 81920; totalSamplesSeen = 163840; learningRatePerSample = 0.003125; epochTime=0.28221s
01/17/2018 06:13:48: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.2'

01/17/2018 06:13:48: Starting Epoch 3: learning rate per sample = 0.003125  effective momentum = 0.810210  momentum as time constant = 2432.7 samples
minibatchiterator: epoch 2: frames [163840..245760] (first utterance at frame 163840), data subset 0 of 1, with 1 datapasses

01/17/2018 06:13:48: Starting minibatch loop.
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[   1-  10, 6.25%]: ce = 1.11074848 * 5120; err = 0.34375000 * 5120; time = 0.0172s; samplesPerSecond = 297989.7
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[  11-  20, 12.50%]: ce = 1.10125542 * 5120; err = 0.34550781 * 5120; time = 0.0165s; samplesPerSecond = 309844.8
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[  21-  30, 18.75%]: ce = 1.08591633 * 5120; err = 0.34277344 * 5120; time = 0.0165s; samplesPerSecond = 310143.3
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[  31-  40, 25.00%]: ce = 1.10742760 * 5120; err = 0.33554688 * 5120; time = 0.0218s; samplesPerSecond = 234876.4
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[  41-  50, 31.25%]: ce = 1.12246552 * 5120; err = 0.33886719 * 5120; time = 0.0223s; samplesPerSecond = 229409.2
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[  51-  60, 37.50%]: ce = 1.08610725 * 5120; err = 0.33730469 * 5120; time = 0.0188s; samplesPerSecond = 272066.9
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[  61-  70, 43.75%]: ce = 1.08662262 * 5120; err = 0.33417969 * 5120; time = 0.0167s; samplesPerSecond = 306893.7
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[  71-  80, 50.00%]: ce = 1.06978607 * 5120; err = 0.32246094 * 5120; time = 0.0165s; samplesPerSecond = 311155.4
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[  81-  90, 56.25%]: ce = 1.02804794 * 5120; err = 0.31328125 * 5120; time = 0.0171s; samplesPerSecond = 299688.6
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[  91- 100, 62.50%]: ce = 1.04875183 * 5120; err = 0.31875000 * 5120; time = 0.0167s; samplesPerSecond = 307459.5
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[ 101- 110, 68.75%]: ce = 1.05174637 * 5120; err = 0.33476563 * 5120; time = 0.0168s; samplesPerSecond = 304930.7
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[ 111- 120, 75.00%]: ce = 1.07829895 * 5120; err = 0.33593750 * 5120; time = 0.0165s; samplesPerSecond = 310792.8
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[ 121- 130, 81.25%]: ce = 1.05014038 * 5120; err = 0.31875000 * 5120; time = 0.0165s; samplesPerSecond = 311049.5
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[ 131- 140, 87.50%]: ce = 1.02171173 * 5120; err = 0.32167969 * 5120; time = 0.0166s; samplesPerSecond = 307910.7
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[ 141- 150, 93.75%]: ce = 1.04328461 * 5120; err = 0.32851562 * 5120; time = 0.0165s; samplesPerSecond = 311153.5
01/17/2018 06:13:48:  Epoch[ 3 of 4]-Minibatch[ 151- 160, 100.00%]: ce = 1.01844177 * 5120; err = 0.31875000 * 5120; time = 0.0164s; samplesPerSecond = 311365.4
01/17/2018 06:13:48: Finished Epoch[ 3 of 4]: [Training] ce = 1.06942205 * 81920; err = 0.33067627 * 81920; totalSamplesSeen = 245760; learningRatePerSample = 0.003125; epochTime=0.28293s
01/17/2018 06:13:48: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.3'

01/17/2018 06:13:48: Starting Epoch 4: learning rate per sample = 0.003125  effective momentum = 0.810210  momentum as time constant = 2432.7 samples
minibatchiterator: epoch 3: frames [245760..327680] (first utterance at frame 245760), data subset 0 of 1, with 1 datapasses

01/17/2018 06:13:48: Starting minibatch loop.
01/17/2018 06:13:48:  Epoch[ 4 of 4]-Minibatch[   1-  10, 6.25%]: ce = 1.03536406 * 5120; err = 0.31777344 * 5120; time = 0.0171s; samplesPerSecond = 299341.7
01/17/2018 06:13:48:  Epoch[ 4 of 4]-Minibatch[  11-  20, 12.50%]: ce = 1.03895218 * 4926; err = 0.32541616 * 4926; time = 0.0570s; samplesPerSecond = 86407.7
01/17/2018 06:13:48:  Epoch[ 4 of 4]-Minibatch[  21-  30, 18.75%]: ce = 1.00940247 * 5120; err = 0.32109375 * 5120; time = 0.0165s; samplesPerSecond = 310790.9
01/17/2018 06:13:48:  Epoch[ 4 of 4]-Minibatch[  31-  40, 25.00%]: ce = 0.99019489 * 5120; err = 0.31230469 * 5120; time = 0.0165s; samplesPerSecond = 309708.0
01/17/2018 06:13:48:  Epoch[ 4 of 4]-Minibatch[  41-  50, 31.25%]: ce = 0.99245567 * 5120; err = 0.31425781 * 5120; time = 0.0170s; samplesPerSecond = 301920.6
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[  51-  60, 37.50%]: ce = 1.00989609 * 5120; err = 0.32246094 * 5120; time = 0.0164s; samplesPerSecond = 311706.7
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[  61-  70, 43.75%]: ce = 1.01605911 * 5120; err = 0.31718750 * 5120; time = 0.0165s; samplesPerSecond = 311187.6
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[  71-  80, 50.00%]: ce = 1.00204391 * 5120; err = 0.31464844 * 5120; time = 0.0168s; samplesPerSecond = 305287.1
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[  81-  90, 56.25%]: ce = 0.99435730 * 5120; err = 0.30527344 * 5120; time = 0.0165s; samplesPerSecond = 311089.2
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[  91- 100, 62.50%]: ce = 0.99423981 * 5120; err = 0.30605469 * 5120; time = 0.0164s; samplesPerSecond = 311517.0
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[ 101- 110, 68.75%]: ce = 1.01819534 * 5120; err = 0.31035156 * 5120; time = 0.0167s; samplesPerSecond = 307478.0
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[ 111- 120, 75.00%]: ce = 1.04231644 * 5120; err = 0.32695313 * 5120; time = 0.0165s; samplesPerSecond = 310508.2
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[ 121- 130, 81.25%]: ce = 0.98021393 * 5120; err = 0.30078125 * 5120; time = 0.0166s; samplesPerSecond = 308417.0
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[ 131- 140, 87.50%]: ce = 0.97062073 * 5120; err = 0.30136719 * 5120; time = 0.0166s; samplesPerSecond = 307799.6
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[ 141- 150, 93.75%]: ce = 0.98007813 * 5120; err = 0.31074219 * 5120; time = 0.0165s; samplesPerSecond = 310921.1
01/17/2018 06:13:49:  Epoch[ 4 of 4]-Minibatch[ 151- 160, 100.00%]: ce = 0.97195587 * 5120; err = 0.29687500 * 5120; time = 0.0164s; samplesPerSecond = 311310.5
01/17/2018 06:13:49: Finished Epoch[ 4 of 4]: [Training] ce = 1.00270529 * 81920; err = 0.31278076 * 81920; totalSamplesSeen = 327680; learningRatePerSample = 0.003125; epochTime=0.310465s
01/17/2018 06:13:49: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech'

01/17/2018 06:13:49: Action "train" complete.


01/17/2018 06:13:49: ##############################################################################
01/17/2018 06:13:49: #                                                                            #
01/17/2018 06:13:49: # replaceCriterionNode command (edit action)                                 #
01/17/2018 06:13:49: #                                                                            #
01/17/2018 06:13:49: ##############################################################################


01/17/2018 06:13:49: Action "edit" complete.


01/17/2018 06:13:49: ##############################################################################
01/17/2018 06:13:49: #                                                                            #
01/17/2018 06:13:49: # sequenceTrain command (train action)                                       #
01/17/2018 06:13:49: #                                                                            #
01/17/2018 06:13:49: ##############################################################################

01/17/2018 06:13:49: 
Starting from checkpoint. Loading network from '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.sequence.0'.
NDLBuilder Using GPU 0
simplesenonehmm: reading '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/model.overalltying', '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/state.list', '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/model.transprob'
simplesenonehmm: 83253 units with 45 unique HMMs, 132 tied states, and 45 trans matrices read
reading script file /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.scp ... 948 entries
trainlayer: OOV-exclusion code enabled, but no unigram specified to derive the word set from, so you won't get OOV exclusion
total 132 state names in state list /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/state.list
htkmlfreader: reading MLF file /tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/glob_0000.mlf ... total 948 entries
archive: opening 80 lattice-archive TOC files ('/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/TestData/CY2SCH010061231_1369712653.numden.lats.toc' etc.).................................................................................. 923 total lattices referenced in 80 archive files
. [no lattice for An4/454/454/an70-meht-b]....... [no lattice for An4/89/89/an6-fjmd-b].. [no lattice for An4/683/683/an364-mmkw-b].. [no lattice for An4/476/476/an256-mewl-b].... [no lattice for An4/2/2/an253-fash-b]...............................................................................feature set 0: 250814 frames in 923 out of 948 utterances
minibatchutterancesource: out of 948 files, 0 files not found in label set and 25 have no lattice
label set 0: 129 classes
minibatchutterancesource: 923 utterances grouped into 3 chunks, av. chunk size: 307.7 utterances, 83604.7 frames
01/17/2018 06:13:49: 
Model has 29 nodes. Using GPU 0.

01/17/2018 06:13:49: Training criterion:   ce = SequenceWithSoftmax
01/17/2018 06:13:49: Evaluation criterion: err = ClassificationError

01/17/2018 06:13:49: Training 779396 parameters in 8 out of 8 parameter tensors and 21 nodes with gradient:

01/17/2018 06:13:49: 	Node 'HL1.W' (LearnableParameter operation) : [512 x 363]
01/17/2018 06:13:49: 	Node 'HL1.b' (LearnableParameter operation) : [512 x 1]
01/17/2018 06:13:49: 	Node 'HL2.W' (LearnableParameter operation) : [512 x 512]
01/17/2018 06:13:49: 	Node 'HL2.b' (LearnableParameter operation) : [512 x 1]
01/17/2018 06:13:49: 	Node 'HL3.W' (LearnableParameter operation) : [512 x 512]
01/17/2018 06:13:49: 	Node 'HL3.b' (LearnableParameter operation) : [512 x 1]
01/17/2018 06:13:49: 	Node 'OL.W' (LearnableParameter operation) : [132 x 512]
01/17/2018 06:13:49: 	Node 'OL.b' (LearnableParameter operation) : [132 x 1]

01/17/2018 06:13:49: No PreCompute nodes found, or all already computed. Skipping pre-computation step.
Setting Hsmoothing weight to 0.95 and frame-dropping threshhold to 1e-10
Setting SeqGammar-related parameters: amf=14.00, lmf=14.00, wp=0.00, bMMIFactor=0.00, usesMBR=false

01/17/2018 06:13:49: Starting Epoch 1: learning rate per sample = 0.000002  effective momentum = 0.995898  momentum as time constant = 2432.7 samples
minibatchiterator: epoch 0: frames [0..81920] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 10.0 ms

01/17/2018 06:13:53: Starting minibatch loop.
dengamma value 1.085592
dengamma value 1.071446
dengamma value 1.041484
dengamma value 1.071582
dengamma value 1.007229
dengamma value 1.049507
dengamma value 1.016768
dengamma value 1.039771
dengamma value 0.990910
dengamma value 1.047892
01/17/2018 06:13:55:  Epoch[ 1 of 3]-Minibatch[   1-  10, 0.12%]: ce = 0.08076315 * 3030; err = 0.33168317 * 3030; time = 1.6586s; samplesPerSecond = 1826.9
dengamma value 0.984427
dengamma value 1.030267
dengamma value 1.040782
dengamma value 1.082328
dengamma value 1.058775
dengamma value 1.007891
dengamma value 0.990956
dengamma value 0.978955
dengamma value 0.930530
dengamma value 1.109135
01/17/2018 06:13:55:  Epoch[ 1 of 3]-Minibatch[  11-  20, 0.24%]: ce = 0.08711072 * 2720; err = 0.34117647 * 2720; time = 0.2585s; samplesPerSecond = 10521.8
dengamma value 1.042756
dengamma value 1.121753
dengamma value 1.009193
dengamma value 1.058572
dengamma value 1.058656
dengamma value 1.093998
dengamma value 0.993350
dengamma value 0.937478
dengamma value 1.042927
dengamma value 0.931578
01/17/2018 06:13:55:  Epoch[ 1 of 3]-Minibatch[  21-  30, 0.37%]: ce = 0.09052124 * 2460; err = 0.33902439 * 2460; time = 0.2185s; samplesPerSecond = 11260.4
dengamma value 1.045144
dengamma value 1.041941
dengamma value 1.013000
dengamma value 1.052089
dengamma value 1.045131
dengamma value 1.085248
dengamma value 1.035781
dengamma value 1.057185
dengamma value 1.096295
dengamma value 1.079609
01/17/2018 06:13:56:  Epoch[ 1 of 3]-Minibatch[  31-  40, 0.49%]: ce = 0.08195799 * 3390; err = 0.28525074 * 3390; time = 0.3471s; samplesPerSecond = 9766.4
dengamma value 1.136612
dengamma value 1.057186
dengamma value 1.072541
dengamma value 1.072105
dengamma value 0.958897
dengamma value 1.091572
dengamma value 1.028559
dengamma value 1.062915
dengamma value 1.017316
dengamma value 1.038177
01/17/2018 06:13:56:  Epoch[ 1 of 3]-Minibatch[  41-  50, 0.61%]: ce = 0.06604747 * 2630; err = 0.32395437 * 2630; time = 0.2762s; samplesPerSecond = 9521.9
dengamma value 1.129306
dengamma value 1.044747
dengamma value 1.020697
dengamma value 1.052896
dengamma value 1.014278
dengamma value 1.080108
dengamma value 0.981639
dengamma value 1.035126
dengamma value 1.108665
dengamma value 1.025858
01/17/2018 06:13:56:  Epoch[ 1 of 3]-Minibatch[  51-  60, 0.73%]: ce = 0.08203647 * 2640; err = 0.32803030 * 2640; time = 0.2828s; samplesPerSecond = 9335.7
dengamma value 1.043133
dengamma value 1.056635
dengamma value 1.000702
dengamma value 0.978152
dengamma value 0.977975
dengamma value 1.113169
dengamma value 1.097272
dengamma value 1.037914
dengamma value 1.053032
dengamma value 1.072180
01/17/2018 06:13:56:  Epoch[ 1 of 3]-Minibatch[  61-  70, 0.85%]: ce = 0.08675747 * 3260; err = 0.30644172 * 3260; time = 0.3582s; samplesPerSecond = 9101.8
dengamma value 1.082317
dengamma value 1.051591
dengamma value 1.043167
dengamma value 1.069331
dengamma value 1.092880
dengamma value 1.088755
dengamma value 1.046199
dengamma value 1.006319
dengamma value 1.061234
dengamma value 1.083249
01/17/2018 06:13:57:  Epoch[ 1 of 3]-Minibatch[  71-  80, 0.98%]: ce = 0.08414011 * 2890; err = 0.26574394 * 2890; time = 0.2899s; samplesPerSecond = 9969.5
dengamma value 1.014453
dengamma value 1.065781
dengamma value 0.968443
dengamma value 1.056673
dengamma value 1.095087
dengamma value 1.079427
dengamma value 1.101784
dengamma value 0.999354
dengamma value 0.982962
dengamma value 1.019914
01/17/2018 06:13:57:  Epoch[ 1 of 3]-Minibatch[  81-  90, 1.10%]: ce = 0.08594954 * 2940; err = 0.34319728 * 2940; time = 0.3085s; samplesPerSecond = 9530.9
dengamma value 1.101123
dengamma value 1.041250
dengamma value 1.146876
dengamma value 1.047008
dengamma value 1.028141
dengamma value 1.090393
dengamma value 1.060193
dengamma value 1.056387
dengamma value 1.025950
dengamma value 1.110658
01/17/2018 06:13:57:  Epoch[ 1 of 3]-Minibatch[  91- 100, 1.22%]: ce = 0.07818488 * 2650; err = 0.27962264 * 2650; time = 0.2982s; samplesPerSecond = 8885.4
dengamma value 1.015076
dengamma value 1.031803
dengamma value 1.081819
dengamma value 1.026175
dengamma value 1.035693
dengamma value 1.123004
dengamma value 1.044640
dengamma value 1.068157
dengamma value 1.025591
dengamma value 0.981567
01/17/2018 06:13:58:  Epoch[ 1 of 3]-Minibatch[ 101- 110, 1.34%]: ce = 0.08591167 * 2410; err = 0.32655602 * 2410; time = 0.2493s; samplesPerSecond = 9667.2
dengamma value 1.048710
dengamma value 1.067718
dengamma value 1.050569
dengamma value 1.012881
dengamma value 1.050871
dengamma value 1.089000
dengamma value 1.059327
dengamma value 1.126727
dengamma value 1.008317
dengamma value 0.985952
01/17/2018 06:13:58:  Epoch[ 1 of 3]-Minibatch[ 111- 120, 1.46%]: ce = 0.07874051 * 2700; err = 0.28555556 * 2700; time = 0.2803s; samplesPerSecond = 9634.2
dengamma value 1.006722
dengamma value 1.031585
dengamma value 1.040546
dengamma value 1.106877
dengamma value 1.077732
dengamma value 1.066431
dengamma value 1.079516
dengamma value 1.044075
dengamma value 1.053967
dengamma value 0.997020
01/17/2018 06:13:58:  Epoch[ 1 of 3]-Minibatch[ 121- 130, 1.59%]: ce = 0.08081588 * 2380; err = 0.31764706 * 2380; time = 0.2464s; samplesPerSecond = 9657.4
dengamma value 1.044012
dengamma value 0.942471
dengamma value 1.079897
dengamma value 0.946740
dengamma value 1.078905
dengamma value 1.044360
dengamma value 1.068945
dengamma value 1.101668
dengamma value 1.046401
dengamma value 0.963575
01/17/2018 06:13:58:  Epoch[ 1 of 3]-Minibatch[ 131- 140, 1.71%]: ce = 0.07699183 * 2630; err = 0.33916350 * 2630; time = 0.2956s; samplesPerSecond = 8898.1
dengamma value 1.021407
dengamma value 1.040773
dengamma value 1.047931
dengamma value 1.003701
dengamma value 1.089679
dengamma value 0.956122
dengamma value 1.068233
dengamma value 1.043382
dengamma value 1.117534
dengamma value 1.068613
01/17/2018 06:13:59:  Epoch[ 1 of 3]-Minibatch[ 141- 150, 1.83%]: ce = 0.08335426 * 3100; err = 0.30774194 * 3100; time = 0.3468s; samplesPerSecond = 8937.6
dengamma value 1.050635
dengamma value 0.916675
dengamma value 1.007561
dengamma value 0.993326
dengamma value 1.042851
dengamma value 1.053824
dengamma value 0.972084
dengamma value 0.970353
dengamma value 1.000634
dengamma value 1.047755
01/17/2018 06:13:59:  Epoch[ 1 of 3]-Minibatch[ 151- 160, 1.95%]: ce = 0.07805598 * 2720; err = 0.38161765 * 2720; time = 0.2767s; samplesPerSecond = 9831.9
dengamma value 1.135599
dengamma value 1.012613
dengamma value 1.025694
dengamma value 1.052878
dengamma value 1.020981
dengamma value 0.997493
dengamma value 1.034745
dengamma value 1.037847
dengamma value 1.054560
dengamma value 0.977525
01/17/2018 06:13:59:  Epoch[ 1 of 3]-Minibatch[ 161- 170, 2.08%]: ce = 0.08179663 * 3000; err = 0.33000000 * 3000; time = 0.2873s; samplesPerSecond = 10441.9
dengamma value 1.201735
dengamma value 1.091450
dengamma value 1.082086
dengamma value 1.110209
dengamma value 1.021330
dengamma value 1.074389
dengamma value 1.091724
dengamma value 1.039205
dengamma value 1.026192
dengamma value 1.008452
01/17/2018 06:14:00:  Epoch[ 1 of 3]-Minibatch[ 171- 180, 2.20%]: ce = 0.07435676 * 3370; err = 0.26884273 * 3370; time = 0.3542s; samplesPerSecond = 9513.3
dengamma value 1.075135
dengamma value 1.101025
dengamma value 1.063082
dengamma value 0.943519
dengamma value 1.067226
dengamma value 0.975263
dengamma value 1.068642
dengamma value 0.975010
dengamma value 1.063940
dengamma value 1.073604
01/17/2018 06:14:00:  Epoch[ 1 of 3]-Minibatch[ 181- 190, 2.32%]: ce = 0.07882024 * 2600; err = 0.36961538 * 2600; time = 0.2898s; samplesPerSecond = 8971.3
dengamma value 1.074046
dengamma value 1.107414
dengamma value 1.035950
dengamma value 1.015494
dengamma value 1.050559
dengamma value 0.999433
dengamma value 1.131953
dengamma value 0.962575
dengamma value 1.117206
dengamma value 0.984125
01/17/2018 06:14:00:  Epoch[ 1 of 3]-Minibatch[ 191- 200, 2.44%]: ce = 0.08103253 * 2600; err = 0.32538462 * 2600; time = 0.2816s; samplesPerSecond = 9232.8
dengamma value 1.064740
dengamma value 1.038668
dengamma value 0.953534
dengamma value 1.040677
dengamma value 1.066185
dengamma value 0.995870
dengamma value 1.004703
dengamma value 1.092931
dengamma value 1.117450
dengamma value 1.100689
01/17/2018 06:14:01:  Epoch[ 1 of 3]-Minibatch[ 201- 210, 2.56%]: ce = 0.07680261 * 2300; err = 0.32086957 * 2300; time = 0.2526s; samplesPerSecond = 9104.0
dengamma value 0.993670
dengamma value 1.007561
dengamma value 0.977259
dengamma value 1.048659
dengamma value 1.028513
dengamma value 1.037645
dengamma value 1.135694
dengamma value 1.017722
dengamma value 1.053688
dengamma value 1.034561
01/17/2018 06:14:01:  Epoch[ 1 of 3]-Minibatch[ 211- 220, 2.69%]: ce = 0.08678676 * 2800; err = 0.33214286 * 2800; time = 0.2998s; samplesPerSecond = 9338.4
dengamma value 1.042492
dengamma value 1.062759
dengamma value 1.077777
dengamma value 1.071892
dengamma value 0.988225
dengamma value 1.013145
dengamma value 1.034390
dengamma value 1.031910
dengamma value 1.086222
dengamma value 1.005992
01/17/2018 06:14:01:  Epoch[ 1 of 3]-Minibatch[ 221- 230, 2.81%]: ce = 0.08701869 * 2590; err = 0.33822394 * 2590; time = 0.2755s; samplesPerSecond = 9399.8
dengamma value 0.978663
dengamma value 1.041660
dengamma value 1.075005
dengamma value 1.053110
dengamma value 1.031181
dengamma value 1.059810
dengamma value 0.978642
dengamma value 1.123866
dengamma value 1.086384
dengamma value 1.077236
01/17/2018 06:14:01:  Epoch[ 1 of 3]-Minibatch[ 231- 240, 2.93%]: ce = 0.07835511 * 2610; err = 0.32605364 * 2610; time = 0.2714s; samplesPerSecond = 9615.9
dengamma value 1.089461
dengamma value 1.066701
dengamma value 1.058816
dengamma value 1.068505
dengamma value 1.058705
dengamma value 1.089026
dengamma value 1.086952
dengamma value 1.017248
dengamma value 1.041065
dengamma value 1.005495
01/17/2018 06:14:02:  Epoch[ 1 of 3]-Minibatch[ 241- 250, 3.05%]: ce = 0.08003601 * 2400; err = 0.33166667 * 2400; time = 0.2400s; samplesPerSecond = 9999.5
dengamma value 1.050215
dengamma value 1.046810
dengamma value 1.077963
dengamma value 1.060864
dengamma value 1.021722
dengamma value 1.031329
dengamma value 1.120931
dengamma value 1.066482
dengamma value 1.042617
dengamma value 1.005963
01/17/2018 06:14:02:  Epoch[ 1 of 3]-Minibatch[ 251- 260, 3.17%]: ce = 0.07966202 * 3200; err = 0.30562500 * 3200; time = 0.3370s; samplesPerSecond = 9495.4
dengamma value 1.084082
dengamma value 1.033501
dengamma value 1.088335
dengamma value 1.026204
dengamma value 1.070161
dengamma value 1.016248
dengamma value 1.077002
dengamma value 1.029063
dengamma value 1.063997
dengamma value 0.996906
01/17/2018 06:14:02:  Epoch[ 1 of 3]-Minibatch[ 261- 270, 3.30%]: ce = 0.08803492 * 3570; err = 0.30644258 * 3570; time = 0.4042s; samplesPerSecond = 8832.0
dengamma value 0.999609
dengamma value 1.059149
dengamma value 1.058708
dengamma value 0.994577
dengamma value 1.049965
dengamma value 0.990715
dengamma value 1.030390
dengamma value 1.114157
dengamma value 0.964035
dengamma value 1.045304
01/17/2018 06:14:03:  Epoch[ 1 of 3]-Minibatch[ 271- 280, 3.42%]: ce = 0.09686586 * 2510; err = 0.36294821 * 2510; time = 0.2425s; samplesPerSecond = 10348.7
dengamma value 0.989013
dengamma value 1.093533
dengamma value 0.903680
dengamma value 1.055913
dengamma value 1.074011
dengamma value 1.077036
dengamma value 1.072860
dengamma value 1.026948
dengamma value 1.056026
dengamma value 1.055009
01/17/2018 06:14:03:  Epoch[ 1 of 3]-Minibatch[ 281- 290, 3.54%]: ce = 0.07571347 * 3400; err = 0.36588235 * 3400; time = 0.3602s; samplesPerSecond = 9439.6
dengamma value 1.112779
dengamma value 1.043813
dengamma value 1.001891
01/17/2018 06:14:03: Finished Epoch[ 1 of 3]: [Training] ce = 0.08176528 * 82104; err = 0.32305369 * 82104; totalSamplesSeen = 82104; learningRatePerSample = 2e-06; epochTime=13.9714s
01/17/2018 06:14:03: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.sequence.1'

01/17/2018 06:14:03: Starting Epoch 2: learning rate per sample = 0.000002  effective momentum = 0.995898  momentum as time constant = 2432.7 samples
minibatchiterator: epoch 1: frames [81920..163840] (first utterance at frame 82104), data subset 0 of 1, with 1 datapasses

01/17/2018 06:14:03: Starting minibatch loop.
dengamma value 1.055532
dengamma value 1.055691
dengamma value 1.027819
dengamma value 0.930571
dengamma value 1.098315
dengamma value 1.033223
dengamma value 1.046419
dengamma value 1.041331
dengamma value 1.037116
dengamma value 1.026451
01/17/2018 06:14:03:  Epoch[ 2 of 3]-Minibatch[   1-  10, 0.12%]: ce = 0.08531331 * 2880; err = 0.31909722 * 2880; time = 0.2918s; samplesPerSecond = 9869.9
dengamma value 1.023625
dengamma value 1.112623
dengamma value 1.030199
dengamma value 1.003669
dengamma value 1.059372
dengamma value 1.044742
dengamma value 1.037744
dengamma value 1.049540
dengamma value 1.046089
dengamma value 1.100100
01/17/2018 06:14:04:  Epoch[ 2 of 3]-Minibatch[  11-  20, 0.24%]: ce = 0.08071604 * 2560; err = 0.29687500 * 2560; time = 0.2665s; samplesPerSecond = 9607.5
dengamma value 1.049389
dengamma value 1.043479
dengamma value 1.032212
dengamma value 1.128287
dengamma value 1.006187
dengamma value 1.034054
dengamma value 0.979706
dengamma value 1.003432
dengamma value 1.072972
dengamma value 1.035142
01/17/2018 06:14:04:  Epoch[ 2 of 3]-Minibatch[  21-  30, 0.37%]: ce = 0.08563805 * 2240; err = 0.32678571 * 2240; time = 0.2389s; samplesPerSecond = 9377.0
dengamma value 1.043645
dengamma value 1.006029
dengamma value 1.079517
dengamma value 1.061947
dengamma value 1.002906
dengamma value 1.002316
dengamma value 1.003873
dengamma value 1.066900
dengamma value 1.014893
dengamma value 1.044063
01/17/2018 06:14:04:  Epoch[ 2 of 3]-Minibatch[  31-  40, 0.49%]: ce = 0.08828906 * 2940; err = 0.32585034 * 2940; time = 0.3201s; samplesPerSecond = 9183.3
dengamma value 1.061005
dengamma value 1.033555
dengamma value 1.067614
dengamma value 1.034329
dengamma value 1.073618
dengamma value 1.036117
dengamma value 1.044576
dengamma value 1.044307
dengamma value 1.012920
dengamma value 1.057996
01/17/2018 06:14:04:  Epoch[ 2 of 3]-Minibatch[  41-  50, 0.61%]: ce = 0.08318363 * 2570; err = 0.30739300 * 2570; time = 0.2417s; samplesPerSecond = 10632.9
dengamma value 0.994694
dengamma value 1.026047
dengamma value 1.093941
dengamma value 1.041647
dengamma value 1.065649
dengamma value 1.029694
dengamma value 1.005609
dengamma value 1.047832
dengamma value 1.027709
dengamma value 1.026578
01/17/2018 06:14:05:  Epoch[ 2 of 3]-Minibatch[  51-  60, 0.73%]: ce = 0.08676719 * 2800; err = 0.29464286 * 2800; time = 0.2775s; samplesPerSecond = 10091.7
dengamma value 1.000809
dengamma value 1.016340
dengamma value 1.020434
dengamma value 1.067570
dengamma value 0.996505
dengamma value 1.099938
dengamma value 1.113480
dengamma value 1.033836
dengamma value 1.106495
dengamma value 1.101523
01/17/2018 06:14:05:  Epoch[ 2 of 3]-Minibatch[  61-  70, 0.85%]: ce = 0.07946878 * 2540; err = 0.27519685 * 2540; time = 0.2895s; samplesPerSecond = 8773.3
dengamma value 1.029639
dengamma value 1.055266
dengamma value 1.063874
dengamma value 0.989733
dengamma value 0.998129
dengamma value 1.061384
dengamma value 0.996896
dengamma value 0.974640
dengamma value 1.007075
dengamma value 1.065890
01/17/2018 06:14:05:  Epoch[ 2 of 3]-Minibatch[  71-  80, 0.98%]: ce = 0.08434347 * 2180; err = 0.35412844 * 2180; time = 0.2081s; samplesPerSecond = 10474.9
dengamma value 1.017184
dengamma value 1.012097
dengamma value 0.978125
dengamma value 0.970466
dengamma value 1.074648
dengamma value 1.112950
dengamma value 1.050000
dengamma value 1.082663
dengamma value 1.043013
dengamma value 1.003298
01/17/2018 06:14:06:  Epoch[ 2 of 3]-Minibatch[  81-  90, 1.10%]: ce = 0.09151962 * 3060; err = 0.34183007 * 3060; time = 0.3142s; samplesPerSecond = 9737.8
dengamma value 1.059898
dengamma value 1.099280
dengamma value 1.066548
dengamma value 1.018401
dengamma value 1.121812
dengamma value 1.062095
dengamma value 1.017651
dengamma value 0.997267
dengamma value 0.914559
dengamma value 0.998489
01/17/2018 06:14:06:  Epoch[ 2 of 3]-Minibatch[  91- 100, 1.22%]: ce = 0.08309687 * 3280; err = 0.33567073 * 3280; time = 0.3484s; samplesPerSecond = 9414.3
dengamma value 1.046067
dengamma value 1.001055
dengamma value 1.099291
dengamma value 1.024097
dengamma value 1.006391
dengamma value 0.989429
dengamma value 1.050647
dengamma value 1.051602
dengamma value 1.053942
dengamma value 1.003344
01/17/2018 06:14:06:  Epoch[ 2 of 3]-Minibatch[ 101- 110, 1.34%]: ce = 0.09013575 * 3270; err = 0.30764526 * 3270; time = 0.3439s; samplesPerSecond = 9508.8
dengamma value 1.063836
dengamma value 0.990877
dengamma value 1.040015
dengamma value 1.033130
dengamma value 1.096731
dengamma value 1.059945
dengamma value 1.023546
dengamma value 1.081310
dengamma value 1.104864
dengamma value 1.065360
01/17/2018 06:14:07:  Epoch[ 2 of 3]-Minibatch[ 111- 120, 1.46%]: ce = 0.08824501 * 2560; err = 0.30859375 * 2560; time = 0.2900s; samplesPerSecond = 8826.1
dengamma value 1.022503
dengamma value 1.052030
dengamma value 1.038282
dengamma value 1.030547
dengamma value 1.073923
dengamma value 1.014216
dengamma value 1.085011
dengamma value 1.049791
dengamma value 1.019830
dengamma value 1.051453
01/17/2018 06:14:07:  Epoch[ 2 of 3]-Minibatch[ 121- 130, 1.59%]: ce = 0.08079401 * 2820; err = 0.30957447 * 2820; time = 0.2921s; samplesPerSecond = 9654.7
dengamma value 1.014071
dengamma value 1.016334
dengamma value 1.049769
dengamma value 1.007258
dengamma value 1.129850
dengamma value 1.017397
dengamma value 1.033014
dengamma value 0.998862
dengamma value 1.038332
dengamma value 1.029170
01/17/2018 06:14:07:  Epoch[ 2 of 3]-Minibatch[ 131- 140, 1.71%]: ce = 0.08786631 * 2420; err = 0.37396694 * 2420; time = 0.2511s; samplesPerSecond = 9636.1
dengamma value 1.018770
dengamma value 1.039732
dengamma value 1.012364
dengamma value 1.047999
dengamma value 1.063220
dengamma value 1.052385
dengamma value 1.011471
dengamma value 1.047765
dengamma value 1.053254
dengamma value 1.004589
01/17/2018 06:14:07:  Epoch[ 2 of 3]-Minibatch[ 141- 150, 1.83%]: ce = 0.07974327 * 2040; err = 0.34607843 * 2040; time = 0.2222s; samplesPerSecond = 9180.0
dengamma value 1.037013
dengamma value 1.053469
dengamma value 1.043653
dengamma value 1.084301
dengamma value 1.058606
dengamma value 0.995751
dengamma value 1.111082
dengamma value 0.992886
dengamma value 0.949674
dengamma value 1.076012
01/17/2018 06:14:08:  Epoch[ 2 of 3]-Minibatch[ 151- 160, 1.95%]: ce = 0.08470424 * 3130; err = 0.33450479 * 3130; time = 0.3174s; samplesPerSecond = 9862.2
dengamma value 1.051913
dengamma value 1.100035
dengamma value 1.070585
dengamma value 1.060229
dengamma value 1.013943
dengamma value 1.071160
dengamma value 1.003355
dengamma value 1.048227
dengamma value 0.974099
dengamma value 1.040547
01/17/2018 06:14:08:  Epoch[ 2 of 3]-Minibatch[ 161- 170, 2.08%]: ce = 0.08746751 * 2600; err = 0.35192308 * 2600; time = 0.2921s; samplesPerSecond = 8902.6
dengamma value 1.041360
dengamma value 1.002137
dengamma value 1.075130
dengamma value 0.964181
dengamma value 1.019639
dengamma value 1.007310
dengamma value 0.959827
dengamma value 1.017657
dengamma value 1.066223
dengamma value 1.008215
01/17/2018 06:14:08:  Epoch[ 2 of 3]-Minibatch[ 171- 180, 2.20%]: ce = 0.09280985 * 2340; err = 0.36324786 * 2340; time = 0.2421s; samplesPerSecond = 9663.9
dengamma value 1.077937
dengamma value 1.102771
dengamma value 1.068618
dengamma value 1.058994
dengamma value 1.042251
dengamma value 0.989695
dengamma value 1.021048
dengamma value 1.038544
dengamma value 1.062225
dengamma value 0.877903
01/17/2018 06:14:08:  Epoch[ 2 of 3]-Minibatch[ 181- 190, 2.32%]: ce = 0.08775357 * 2590; err = 0.34092664 * 2590; time = 0.2640s; samplesPerSecond = 9811.3
dengamma value 1.139447
dengamma value 1.074509
dengamma value 0.970616
dengamma value 1.035687
dengamma value 0.997429
dengamma value 1.004575
dengamma value 1.038732
dengamma value 1.045316
dengamma value 1.043586
dengamma value 1.078847
01/17/2018 06:14:09:  Epoch[ 2 of 3]-Minibatch[ 191- 200, 2.44%]: ce = 0.09020552 * 2640; err = 0.31477273 * 2640; time = 0.2812s; samplesPerSecond = 9389.2
dengamma value 1.094233
dengamma value 1.109460
dengamma value 0.996544
dengamma value 1.080061
dengamma value 1.012535
dengamma value 1.095349
dengamma value 1.018496
dengamma value 1.094128
dengamma value 1.073617
dengamma value 1.046527
01/17/2018 06:14:09:  Epoch[ 2 of 3]-Minibatch[ 201- 210, 2.56%]: ce = 0.08888249 * 2840; err = 0.29330986 * 2840; time = 0.2955s; samplesPerSecond = 9611.0
dengamma value 1.029322
dengamma value 0.995435
dengamma value 1.001968
dengamma value 1.058125
dengamma value 0.985309
dengamma value 1.057485
dengamma value 1.101614
dengamma value 0.945845
dengamma value 1.016905
dengamma value 1.060629
01/17/2018 06:14:09:  Epoch[ 2 of 3]-Minibatch[ 211- 220, 2.69%]: ce = 0.08596441 * 2450; err = 0.36326531 * 2450; time = 0.2510s; samplesPerSecond = 9759.9
dengamma value 1.114126
dengamma value 1.038641
dengamma value 1.056247
dengamma value 1.030473
dengamma value 1.115677
dengamma value 1.097108
dengamma value 1.060199
dengamma value 1.037110
dengamma value 1.070492
dengamma value 1.010478
01/17/2018 06:14:10:  Epoch[ 2 of 3]-Minibatch[ 221- 230, 2.81%]: ce = 0.08903824 * 3180; err = 0.30345912 * 3180; time = 0.3232s; samplesPerSecond = 9839.0
dengamma value 0.962883
dengamma value 1.099859
dengamma value 1.026106
dengamma value 1.034343
dengamma value 1.002216
dengamma value 1.068640
dengamma value 0.994709
dengamma value 1.073168
dengamma value 0.973229
dengamma value 1.055619
01/17/2018 06:14:10:  Epoch[ 2 of 3]-Minibatch[ 231- 240, 2.93%]: ce = 0.08457557 * 2970; err = 0.34040404 * 2970; time = 0.3431s; samplesPerSecond = 8657.2
dengamma value 1.049301
dengamma value 1.028378
dengamma value 1.084516
dengamma value 1.085254
dengamma value 1.047650
dengamma value 1.024163
dengamma value 1.023000
dengamma value 0.997498
dengamma value 1.009478
dengamma value 1.061405
01/17/2018 06:14:10:  Epoch[ 2 of 3]-Minibatch[ 241- 250, 3.05%]: ce = 0.08899573 * 2830; err = 0.28162544 * 2830; time = 0.2916s; samplesPerSecond = 9706.0
dengamma value 1.016204
dengamma value 1.077538
dengamma value 1.094559
dengamma value 1.005716
dengamma value 1.095233
dengamma value 0.959960
dengamma value 1.027039
dengamma value 1.008221
dengamma value 1.030748
dengamma value 1.075858
01/17/2018 06:14:10:  Epoch[ 2 of 3]-Minibatch[ 251- 260, 3.17%]: ce = 0.08077261 * 2600; err = 0.33807692 * 2600; time = 0.2557s; samplesPerSecond = 10167.7
dengamma value 1.104895
dengamma value 1.034957
dengamma value 1.095047
dengamma value 1.067658
dengamma value 1.061504
dengamma value 1.097830
dengamma value 1.051797
dengamma value 1.081107
dengamma value 1.123100
dengamma value 1.096504
01/17/2018 06:14:11:  Epoch[ 2 of 3]-Minibatch[ 261- 270, 3.30%]: ce = 0.07701253 * 2770; err = 0.27220217 * 2770; time = 0.3056s; samplesPerSecond = 9064.6
dengamma value 1.039314
dengamma value 1.023742
dengamma value 1.097859
dengamma value 1.010545
dengamma value 1.112140
dengamma value 1.087165
dengamma value 1.167716
dengamma value 1.022084
dengamma value 1.092489
dengamma value 1.092712
01/17/2018 06:14:11:  Epoch[ 2 of 3]-Minibatch[ 271- 280, 3.42%]: ce = 0.07772022 * 2450; err = 0.33224490 * 2450; time = 0.2847s; samplesPerSecond = 8604.6
dengamma value 0.962385
dengamma value 0.987686
dengamma value 1.084248
dengamma value 1.091608
dengamma value 1.059737
dengamma value 1.029773
dengamma value 1.053498
dengamma value 1.048346
dengamma value 1.059496
dengamma value 1.111459
01/17/2018 06:14:11:  Epoch[ 2 of 3]-Minibatch[ 281- 290, 3.54%]: ce = 0.08094880 * 2730; err = 0.34285714 * 2730; time = 0.2762s; samplesPerSecond = 9885.3
dengamma value 1.081760
dengamma value 0.997461
dengamma value 1.025925
dengamma value 0.962628
dengamma value 1.010243
dengamma value 1.008534
dengamma value 1.010068
dengamma value 1.039498
dengamma value 1.067412
dengamma value 1.010161
01/17/2018 06:14:12:  Epoch[ 2 of 3]-Minibatch[ 291- 300, 3.66%]: ce = 0.08989098 * 2440; err = 0.33524590 * 2440; time = 0.2507s; samplesPerSecond = 9731.7
dengamma value 1.047928
dengamma value 1.027296
dengamma value 0.968702
dengamma value 1.066740
01/17/2018 06:14:12: Finished Epoch[ 2 of 3]: [Training] ce = 0.08551011 * 81852; err = 0.32346186 * 81852; totalSamplesSeen = 163956; learningRatePerSample = 2e-06; epochTime=8.58415s
01/17/2018 06:14:12: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.sequence.2'

01/17/2018 06:14:12: Starting Epoch 3: learning rate per sample = 0.000002  effective momentum = 0.995898  momentum as time constant = 2432.7 samples
minibatchiterator: epoch 2: frames [163840..245760] (first utterance at frame 163956), data subset 0 of 1, with 1 datapasses

01/17/2018 06:14:12: Starting minibatch loop.
dengamma value 1.110620
dengamma value 1.029372
dengamma value 0.909070
dengamma value 1.025470
dengamma value 1.018780
dengamma value 1.034084
dengamma value 1.046380
dengamma value 1.046755
dengamma value 1.072120
dengamma value 0.957179
01/17/2018 06:14:12:  Epoch[ 3 of 3]-Minibatch[   1-  10, 0.12%]: ce = 0.08976399 * 2810; err = 0.32206406 * 2810; time = 0.2785s; samplesPerSecond = 10088.4
dengamma value 1.036484
dengamma value 1.133932
dengamma value 1.088240
dengamma value 1.072873
dengamma value 1.069420
dengamma value 1.023992
dengamma value 1.172189
dengamma value 1.120598
dengamma value 1.083350
dengamma value 1.018375
01/17/2018 06:14:12:  Epoch[ 3 of 3]-Minibatch[  11-  20, 0.24%]: ce = 0.07974844 * 1970; err = 0.31065990 * 1970; time = 0.2055s; samplesPerSecond = 9588.2
dengamma value 1.066813
dengamma value 1.035697
dengamma value 1.028125
dengamma value 1.000217
dengamma value 1.080851
dengamma value 1.062682
dengamma value 1.040857
dengamma value 1.042033
dengamma value 1.001099
dengamma value 1.065784
01/17/2018 06:14:13:  Epoch[ 3 of 3]-Minibatch[  21-  30, 0.37%]: ce = 0.08615116 * 3050; err = 0.30098361 * 3050; time = 0.3057s; samplesPerSecond = 9977.9
dengamma value 1.002968
dengamma value 1.057437
dengamma value 1.054964
dengamma value 0.990713
dengamma value 1.128236
dengamma value 0.994843
dengamma value 0.997885
dengamma value 1.049572
dengamma value 1.080403
dengamma value 1.020483
01/17/2018 06:14:13:  Epoch[ 3 of 3]-Minibatch[  31-  40, 0.49%]: ce = 0.09187803 * 2230; err = 0.33408072 * 2230; time = 0.2408s; samplesPerSecond = 9262.5
dengamma value 1.017574
dengamma value 1.053316
dengamma value 1.023524
dengamma value 1.023568
dengamma value 1.097434
dengamma value 1.058037
dengamma value 1.035606
dengamma value 1.027653
dengamma value 0.985067
dengamma value 1.051370
01/17/2018 06:14:13:  Epoch[ 3 of 3]-Minibatch[  41-  50, 0.61%]: ce = 0.08913338 * 2350; err = 0.33872340 * 2350; time = 0.2484s; samplesPerSecond = 9460.4
dengamma value 1.090492
dengamma value 0.975457
dengamma value 1.062478
dengamma value 1.046687
dengamma value 1.049056
dengamma value 1.066050
dengamma value 1.066178
dengamma value 1.089744
dengamma value 1.011770
dengamma value 0.969172
01/17/2018 06:14:13:  Epoch[ 3 of 3]-Minibatch[  51-  60, 0.73%]: ce = 0.08740967 * 2850; err = 0.32736842 * 2850; time = 0.2959s; samplesPerSecond = 9631.5
dengamma value 0.992644
dengamma value 1.065563
dengamma value 1.068766
dengamma value 1.081797
dengamma value 1.024862
dengamma value 1.074377
dengamma value 1.095989
dengamma value 1.096939
dengamma value 1.033580
dengamma value 1.124714
01/17/2018 06:14:14:  Epoch[ 3 of 3]-Minibatch[  61-  70, 0.85%]: ce = 0.07824388 * 3020; err = 0.30496689 * 3020; time = 0.3627s; samplesPerSecond = 8327.3
dengamma value 1.075945
dengamma value 0.967298
dengamma value 1.062012
dengamma value 1.085542
dengamma value 0.939321
dengamma value 1.023675
dengamma value 1.031839
dengamma value 1.008364
dengamma value 1.051158
dengamma value 1.007159
01/17/2018 06:14:14:  Epoch[ 3 of 3]-Minibatch[  71-  80, 0.98%]: ce = 0.09580436 * 2220; err = 0.31846847 * 2220; time = 0.2161s; samplesPerSecond = 10271.3
dengamma value 1.022387
dengamma value 1.016255
dengamma value 1.101983
dengamma value 1.130538
dengamma value 1.094908
dengamma value 1.032455
dengamma value 1.047819
dengamma value 1.117904
dengamma value 1.040730
dengamma value 1.044204
01/17/2018 06:14:14:  Epoch[ 3 of 3]-Minibatch[  81-  90, 1.10%]: ce = 0.08683124 * 3110; err = 0.27556270 * 3110; time = 0.3728s; samplesPerSecond = 8341.8
dengamma value 1.012881
dengamma value 0.972043
dengamma value 1.010598
dengamma value 1.012210
dengamma value 1.021703
dengamma value 1.084161
dengamma value 0.989334
dengamma value 1.039798
dengamma value 1.109533
dengamma value 1.003094
01/17/2018 06:14:15:  Epoch[ 3 of 3]-Minibatch[  91- 100, 1.22%]: ce = 0.08075247 * 2560; err = 0.36171875 * 2560; time = 0.2734s; samplesPerSecond = 9362.2
dengamma value 1.015362
dengamma value 1.067755
dengamma value 1.146163
dengamma value 1.113834
dengamma value 1.070827
dengamma value 1.052331
dengamma value 1.067634
dengamma value 1.046399
dengamma value 1.080231
dengamma value 0.997883
01/17/2018 06:14:15:  Epoch[ 3 of 3]-Minibatch[ 101- 110, 1.34%]: ce = 0.08150468 * 2780; err = 0.31330935 * 2780; time = 0.2828s; samplesPerSecond = 9831.1
dengamma value 1.022686
dengamma value 1.093580
dengamma value 1.098417
dengamma value 0.943056
dengamma value 0.963957
dengamma value 1.071598
dengamma value 1.043102
dengamma value 1.021944
dengamma value 1.076690
dengamma value 1.042206
01/17/2018 06:14:15:  Epoch[ 3 of 3]-Minibatch[ 111- 120, 1.46%]: ce = 0.08977700 * 2520; err = 0.32539683 * 2520; time = 0.2760s; samplesPerSecond = 9130.2
dengamma value 1.045006
dengamma value 1.009606
dengamma value 0.949925
dengamma value 1.094652
dengamma value 1.125336
dengamma value 1.009857
dengamma value 1.028800
dengamma value 1.095774
dengamma value 1.091655
dengamma value 1.041859
01/17/2018 06:14:15:  Epoch[ 3 of 3]-Minibatch[ 121- 130, 1.59%]: ce = 0.08659356 * 2580; err = 0.32674419 * 2580; time = 0.2682s; samplesPerSecond = 9619.2
dengamma value 1.039658
dengamma value 1.044926
dengamma value 1.009398
dengamma value 1.067034
dengamma value 1.030120
dengamma value 1.091430
dengamma value 1.108022
dengamma value 0.975504
dengamma value 1.054705
dengamma value 0.964947
01/17/2018 06:14:16:  Epoch[ 3 of 3]-Minibatch[ 131- 140, 1.71%]: ce = 0.08245974 * 2450; err = 0.33795918 * 2450; time = 0.2248s; samplesPerSecond = 10899.9
dengamma value 0.961656
dengamma value 1.023292
dengamma value 1.032177
dengamma value 1.103381
dengamma value 0.993618
dengamma value 1.039190
dengamma value 1.100431
dengamma value 0.991350
dengamma value 1.051405
dengamma value 1.055651
01/17/2018 06:14:16:  Epoch[ 3 of 3]-Minibatch[ 141- 150, 1.83%]: ce = 0.08342914 * 2290; err = 0.33144105 * 2290; time = 0.2172s; samplesPerSecond = 10543.6
dengamma value 1.215251
dengamma value 1.033652
dengamma value 0.981108
dengamma value 1.051735
dengamma value 1.090808
dengamma value 1.049925
dengamma value 1.003535
dengamma value 1.041929
dengamma value 1.101351
dengamma value 1.064355
01/17/2018 06:14:16:  Epoch[ 3 of 3]-Minibatch[ 151- 160, 1.95%]: ce = 0.08108236 * 3000; err = 0.28866667 * 3000; time = 0.2911s; samplesPerSecond = 10304.3
dengamma value 1.220061
dengamma value 1.075476
dengamma value 1.055778
dengamma value 1.056912
dengamma value 0.945300
dengamma value 1.059656
dengamma value 1.070068
dengamma value 1.052835
dengamma value 0.966801
dengamma value 1.081473
01/17/2018 06:14:16:  Epoch[ 3 of 3]-Minibatch[ 161- 170, 2.08%]: ce = 0.07432711 * 2510; err = 0.32191235 * 2510; time = 0.2307s; samplesPerSecond = 10880.4
dengamma value 0.986975
dengamma value 1.069280
dengamma value 0.992996
dengamma value 1.056349
dengamma value 1.065694
dengamma value 1.063554
dengamma value 1.012869
dengamma value 1.100636
dengamma value 1.091889
dengamma value 1.061539
01/17/2018 06:14:17:  Epoch[ 3 of 3]-Minibatch[ 171- 180, 2.20%]: ce = 0.08651006 * 2610; err = 0.28429119 * 2610; time = 0.2834s; samplesPerSecond = 9209.0
dengamma value 1.078677
dengamma value 1.089467
dengamma value 1.077103
dengamma value 1.086366
dengamma value 0.991664
dengamma value 1.097497
dengamma value 1.023349
dengamma value 1.062920
dengamma value 1.059059
dengamma value 1.067178
01/17/2018 06:14:17:  Epoch[ 3 of 3]-Minibatch[ 181- 190, 2.32%]: ce = 0.08637085 * 2400; err = 0.32500000 * 2400; time = 0.2584s; samplesPerSecond = 9286.9
dengamma value 1.121711
dengamma value 1.081118
dengamma value 1.050125
dengamma value 1.058810
dengamma value 1.050910
dengamma value 1.032030
dengamma value 1.069189
dengamma value 1.062502
dengamma value 1.022371
dengamma value 1.053790
01/17/2018 06:14:17:  Epoch[ 3 of 3]-Minibatch[ 191- 200, 2.44%]: ce = 0.08598218 * 2590; err = 0.26409266 * 2590; time = 0.2820s; samplesPerSecond = 9183.9
dengamma value 1.117986
dengamma value 1.021991
dengamma value 1.069555
dengamma value 1.009389
dengamma value 1.066734
dengamma value 1.045691
dengamma value 1.068117
dengamma value 1.057114
dengamma value 0.983971
dengamma value 1.052096
01/17/2018 06:14:17:  Epoch[ 3 of 3]-Minibatch[ 201- 210, 2.56%]: ce = 0.07587930 * 2460; err = 0.38292683 * 2460; time = 0.2672s; samplesPerSecond = 9205.8
dengamma value 1.035046
dengamma value 1.090712
dengamma value 1.002802
dengamma value 1.049562
dengamma value 1.083582
dengamma value 1.000723
dengamma value 1.039193
dengamma value 1.057987
dengamma value 1.015598
dengamma value 1.029933
01/17/2018 06:14:18:  Epoch[ 3 of 3]-Minibatch[ 211- 220, 2.69%]: ce = 0.08145695 * 2810; err = 0.33843416 * 2810; time = 0.2708s; samplesPerSecond = 10375.8
dengamma value 1.038936
dengamma value 1.026142
dengamma value 1.054163
dengamma value 1.046108
dengamma value 1.060587
dengamma value 1.126679
dengamma value 1.143463
dengamma value 1.006359
dengamma value 1.002895
dengamma value 0.998035
01/17/2018 06:14:18:  Epoch[ 3 of 3]-Minibatch[ 221- 230, 2.81%]: ce = 0.08805300 * 2550; err = 0.32000000 * 2550; time = 0.2479s; samplesPerSecond = 10286.9
dengamma value 1.060534
dengamma value 1.049096
dengamma value 1.007791
dengamma value 1.022076
dengamma value 1.038340
dengamma value 1.041723
dengamma value 0.971923
dengamma value 0.989889
dengamma value 1.038491
dengamma value 1.044168
01/17/2018 06:14:18:  Epoch[ 3 of 3]-Minibatch[ 231- 240, 2.93%]: ce = 0.08712971 * 2810; err = 0.33451957 * 2810; time = 0.3109s; samplesPerSecond = 9039.5
dengamma value 1.133881
dengamma value 0.942559
dengamma value 1.037781
dengamma value 1.107558
dengamma value 0.986449
dengamma value 0.987818
dengamma value 0.972796
dengamma value 1.086559
dengamma value 1.029544
dengamma value 0.996371
01/17/2018 06:14:18:  Epoch[ 3 of 3]-Minibatch[ 241- 250, 3.05%]: ce = 0.09031242 * 2540; err = 0.34488189 * 2540; time = 0.2562s; samplesPerSecond = 9913.8
dengamma value 1.042929
dengamma value 1.048133
dengamma value 1.010652
dengamma value 1.016824
dengamma value 1.062092
dengamma value 1.044980
dengamma value 1.044335
dengamma value 1.050591
dengamma value 1.051212
dengamma value 1.019243
01/17/2018 06:14:19:  Epoch[ 3 of 3]-Minibatch[ 251- 260, 3.17%]: ce = 0.09112798 * 2790; err = 0.32186380 * 2790; time = 0.3080s; samplesPerSecond = 9059.9
dengamma value 1.071873
dengamma value 1.070938
dengamma value 1.022120
dengamma value 1.113178
dengamma value 1.042713
dengamma value 1.067616
dengamma value 1.100166
dengamma value 1.080709
dengamma value 1.063488
dengamma value 1.055098
01/17/2018 06:14:19:  Epoch[ 3 of 3]-Minibatch[ 261- 270, 3.30%]: ce = 0.08365939 * 3920; err = 0.27091837 * 3920; time = 0.4715s; samplesPerSecond = 8314.6
dengamma value 1.091349
dengamma value 0.989684
dengamma value 0.994667
dengamma value 1.046689
dengamma value 1.044966
dengamma value 1.050924
dengamma value 1.077392
dengamma value 1.021501
dengamma value 1.024190
dengamma value 1.033486
01/17/2018 06:14:20:  Epoch[ 3 of 3]-Minibatch[ 271- 280, 3.42%]: ce = 0.07644210 * 3370; err = 0.37299703 * 3370; time = 0.3337s; samplesPerSecond = 10099.9
dengamma value 1.017464
dengamma value 1.032005
dengamma value 1.025641
dengamma value 1.049554
dengamma value 1.063896
dengamma value 1.048646
dengamma value 1.003035
dengamma value 1.073714
dengamma value 1.069834
dengamma value 1.071686
01/17/2018 06:14:20:  Epoch[ 3 of 3]-Minibatch[ 281- 290, 3.54%]: ce = 0.09546398 * 2930; err = 0.31843003 * 2930; time = 0.3320s; samplesPerSecond = 8825.1
dengamma value 1.018920
dengamma value 1.021554
dengamma value 1.026362
dengamma value 1.066456
dengamma value 1.108099
dengamma value 1.128842
dengamma value 1.056879
dengamma value 1.014008
dengamma value 1.025607
dengamma value 1.029172
01/17/2018 06:14:20:  Epoch[ 3 of 3]-Minibatch[ 291- 300, 3.66%]: ce = 0.07975518 * 2590; err = 0.31698842 * 2590; time = 0.2701s; samplesPerSecond = 9589.1
dengamma value 1.029867
dengamma value 1.076383
dengamma value 1.023832
dengamma value 1.040295
dengamma value 1.033142
01/17/2018 06:14:20: Finished Epoch[ 3 of 3]: [Training] ce = 0.08469954 * 82070; err = 0.32064092 * 82070; totalSamplesSeen = 246026; learningRatePerSample = 2e-06; epochTime=8.62946s
01/17/2018 06:14:20: SGD: Saving checkpoint model '/tmp/cntk-test-20180117061317.742222/Speech/DNN_SequenceTraining@release_gpu/models/cntkSpeech.sequence'

01/17/2018 06:14:21: Action "train" complete.

01/17/2018 06:14:21: __COMPLETED__