Informazioni di riferimento sulla funzione Full BrainScript

Articolo
03/09/2017

Questa sezione fornisce informazioni sulle funzioni predefinite di BrainScript.

Le dichiarazioni di tutte le funzioni predefinite sono disponibili nella CNTK.core.bs che si trova accanto al file binario CNTK.

Le operazioni e i livelli primitivi vengono dichiarati nello spazio dei nomi globale. Le operazioni aggiuntive vengono dichiarate negli spazi dei nomi e verranno fornite con il rispettivo prefisso (ad esempio, BS.RNN.LSTMP).

Strati

DenseLayer {outDim, bias= true, activation=Identity, init='uniform', initValueScale=1}
ConvolutionalLayer {numOutputChannels, filterShape, activation = Identity,
init = "uniform", initValueScale = 1,
stride = 1, pad = false, lowerPad = 0, upperPad = 0, bias=true}
MaxPoolingLayer {filterShape, stride = 1, pad = false, lowerPad = 0, upperPad = 0}
AveragePoolingLayer {filterShape, stride = 1, pad = false, lowerPad = 0, upperPad = 0}
EmbeddingLayer {outDim, embeddingPath = '', transpose = false}
RecurrentLSTMLayer {outputDim, cellShape = None, goBackwards = false, enableSelfStabilization = false}
DelayLayer {T=1, defaultHiddenActivation=0}
Dropout
BatchNormalizationLayer {spatialRank = 0, initialScale = 1, normalizationTimeConstant = 0, blendTimeConstant = 0, epsilon = 0.00001, useCntkEngine = true}
LayerNormalizationLayer {initialScale = 1, initialBias = 0}
StabilizerLayer{}
FeatureMVNLayer{}

Compilazione di livelli

Sequential()

Funzioni di attivazione

Sigmoid(x)
Tanh(x)
ReLU(x)
Softmax(x, axis=0)
LogSoftmax(x)
Hardmax(x)

Operazioni elementwise, unarie

Abs (x)
Ceil (x)
Cosine (x)
Clip (x, minValue, maxValue)
Exp (x)
Floor (x)
Log (x)
Negate (x)
-x
BS.Boolean.Not (b)
!x
Reciprocal (x)
Round (x)
Sin (x)
Sqrt (x)

Operazioni element per elemento, binario

ElementTimes (x, y)
x .* y
Minus (x, y)
x - y
Plus (x, y)
x + y`
LogPlus (x, y)
Less (x, y)
Equal (x, y)
Greater (x, y)
GreaterEqual (x, y)
NotEqual (x, y)
LessEqual (x, y)
BS.Boolean.And (a, b)
BS.Boolean.Or (a, b)
BS.Boolean.Xor (a, b)

Operazioni elementwise, ternarie

BS.Boolean.If (condition, thenVal, elseVal)

Operazioni di convoluzione e prodotto matrice

Times(A, B, outputRank=1)
A * B
TransposeTimes(A, B, outputRank=1)
Convolution(weights, x, kernelShape, mapDims=(0), stride=(1), sharing=(true), autoPadding=(true), lowerPadding=(0), upperPadding=(0), imageLayout='CHW', maxTempMemSizeInSamples=0)
Pooling(x, poolKind/*'max'|'average'*/, kernelShape, stride=(1), autoPadding=(true), lowerPadding=(0), upperPadding=(0), imageLayout='CHW')
ROIPooling(x, rois, roiOutputShape, spatialScale=1.0/16.0)

Parametri e costanti learnable

ParameterTensor {shape, learningRateMultiplier=1.0, init='uniform'/*|gaussian*/, initValueScale=1.0, initValue=0.0, randomSeed=-1, initFromFilePath=''}
Constant {scalarValue, rows = 1, cols = 1}
BS.Constants.Zero, BS.Constants.One
BS.Constants.True, BS.Constants.False, BS.Constants.None
BS.Constants.OnesTensor (shape)
BS.Constants.ZeroSequenceLike (x)

Ingressi

Input (shape, dynamicAxis='', sparse=false, tag='feature')
DynamicAxis{}
EnvironmentInput (propertyName)
Mean (x), InvStdDev (x)

Funzioni di perdita e metriche

CrossEntropyWithSoftmax (targetDistribution, nonNormalizedLogClassPosteriors)
CrossEntropy (targetDistribution, classPosteriors)
Logistic (label, probability)
WeightedLogistic(label, probability, instanceWeight)
ClassificationError (labels, nonNormalizedLogClassPosteriors)
MatrixL1Reg(matrix)
MatrixL2Reg(matrix)
SquareError (x, y)

Riduzioni

ReduceSum (z, axis=None)
ReduceLogSum (z, axis=None)
ReduceMean (z, axis=None)
ReduceMin (z, axis=None)
ReduceMax (z, axis=None)
CosDistance (x, y)
SumElements (z)

Operazioni di training

BatchNormalization (input, scale, bias, runMean, runInvStdDev, spatial, normalizationTimeConstant = 0, blendTimeConstant = 0, epsilon = 0.00001, useCntkEngine = true, imageLayout='CHW')
Dropout (x)
Stabilize (x, enabled=true)
StabilizeElements (x, inputDim=x.dim, enabled=true)
CosDistanceWithNegativeSamples (x, y, numShifts, numNegSamples)

Operazioni di rishaping

CNTK2.Reshape (x, shape, beginAxis=0, endAxis=0)
ReshapeDimension (x, axis, shape) = CNTK2.Reshape (x, shape, beginAxis=axis, endAxis=axis + 1)
FlattenDimensions (x, axis, num) = CNTK2.Reshape (x, 0, beginAxis=axis, endAxis=axis + num)
SplitDimension (x, axis, N) = ReshapeDimension (x, axis, 0:N)
Slice (beginIndex, endIndex, input, axis=1)
BS.Sequences.First (x) = Slice (0, 1, x, axis=-1)
BS.Sequences.Last (x) = Slice (-1, 0, x, axis=-1)
Splice (inputs, axis=1)
TransposeDimensions (x, axis1, axis2)
Transpose (x) = TransposeDimensions (x, 1, 2)
BS.Sequences.BroadcastSequenceAs (type, data1)
BS.Sequences.Gather (where, x)
BS.Sequences.Scatter (where, y)
BS.Sequences.IsFirst (x)
BS.Sequences.IsLast (x)

Ricorrenza

OptimizedRNNStack(weights, input, hiddenDims, numLayers=1, bidirectional=false, recurrentOp='lstm')
BS.Loop.Previous (x, timeStep=1, defaultHiddenActivation=0)
PastValue (shape, x, defaultHiddenActivation=0.1, ...) = BS.Loop.Previous (0, shape, ...)
BS.Loop.Next (x, timeStep=1, defaultHiddenActivation=0)
FutureValue (shape, x, defaultHiddenActivation=0.1, ...) = BS.Loop.Next (0, shape, ...)
LSTMP (outputDim, cellDim=outputDim, x, inputDim=x.shape, aux=BS.Constants.None, auxDim=aux.shape, prevState, enableSelfStabilization=false)
BS.Boolean.Toggle (clk, initialValue=BS.Constants.False)
BS.RNNs.RecurrentLSTMP (outputDim, cellDim=outputDim, x, inputDim=x.shape, previousHook=BS.RNNs.PreviousHC, augmentInputHook=NoAuxInputHook, augmentInputDim=0, layerIndex=0, enableSelfStabilization=false)
BS.RNNs.RecurrentLSTMPStack (layerShapes, cellDims=layerShapes, input, inputShape=input.shape, previousHook=PreviousHC, augmentInputHook=NoAuxInputHook, augmentInputShape=0, enableSelfStabilization=false)
BS.RNNs.RecurrentBirectionalLSTMPStack (layerShapes, cellDims=layerShapes, input, inputShape=input.dim, previousHook=PreviousHC, nextHook=NextHC, enableSelfStabilization=false)

Supporto da sequenza a sequenza

BS.Seq2Seq.CreateAugmentWithFixedWindowAttentionHook (attentionDim, attentionSpan, decoderDynamicAxis, encoderOutput, enableSelfStabilization=false)
BS.Seq2Seq.GreedySequenceDecoderFrom (modelAsTrained)
BS.Seq2Seq.BeamSearchSequenceDecoderFrom (modelAsTrained, beamDepth)

Operazioni per scopi speciali

ClassBasedCrossEntropyWithSoftmax (labelClassDescriptorVectorSequence, mainInputInfo, mainWeight, classLogProbsBeforeSoftmax)

Modifica del modello

BS.Network.Load (pathName)
BS.Network.Edit (inputModel, editFunctions, additionalRoots)
BS.Network.CloneFunction (inputNodes, outputNodes, parameters="learnable" /*|"constant"|"shared"*/)

Altro

Fail (what)
IsSameObject (a, b)
Trace (node, say='', logFrequency=traceFrequency, logFirst=10, logGradientToo=false, onlyUpToRow=100000000, onlyUpToT=100000000, format=[])

Deprecato

ErrorPrediction (labels, nonNormalizedLogClassPosteriors)
ColumnElementTimes (...) = ElementTimes (...)
DiagTimes (...) = ElementTimes (...)
LearnableParameter(...) = Parameter(...)
LookupTable (embeddingMatrix, inputTensor)
RowRepeat (input, numRepeats)
RowSlice (beginIndex, numRows, input) = Slice(beginIndex, beginIndex + numRows, input, axis = 1)
RowStack (inputs)
RowElementTimes (...) = ElementTimes (...)
Scale (...) = ElementTimes (...)
ConstantTensor (scalarVal, shape)
Parameter (outputDim, inputDim, ...) = ParameterTensor ((outputDim:input), ...)
WeightParam (outputDim, inputDim) = Parameter (outputDim, inputDim, init='uniform', initValueScale=1, initOnCPUOnly=true, randomSeed=1)
DiagWeightParam (outputDim) = ParameterTensor ((outputDim), init='uniform', initValueScale=1, initOnCPUOnly=true, randomSeed=1)
BiasParam (dim) = ParameterTensor ((dim), init='fixedValue', value=0.0)
ScalarParam() = BiasParam (1)
SparseInput (shape, dynamicAxis='', tag='feature')
ImageInput (imageWidth, imageHeight, imageChannels, imageLayout='CHW', dynamicAxis='', tag='feature')
SparseImageInput (imageWidth, imageHeight, imageChannels, imageLayout='CHW', dynamicAxis='', tag='feature')
MeanVarNorm(feat) = PerDimMeanVarNormalization(feat, Mean (feat), InvStdDev (feat))
PerDimMeanVarNormalization (x, mean, invStdDev),
PerDimMeanVarDeNormalization (x, mean, invStdDev)
ReconcileDynamicAxis (dataInput, layoutInput)

Condividi tramite