public <T> SparseReorder<T> sparseReorder(Operand<Long> inputIndices, Operand<T> inputValues, Operand<Long> inputShape)
SparseReorder
operationinputIndices
- 2-D. `N x R` matrix with the indices of non-empty values in ainputValues
- 1-D. `N` non-empty values corresponding to `input_indices`.inputShape
- 1-D. Shape of the input SparseTensor.SparseReorder
public <T extends Number,U extends Number> SparseSegmentSqrtNGrad<T> sparseSegmentSqrtNGrad(Operand<T> grad, Operand<U> indices, Operand<Integer> segmentIds, Operand<Integer> outputDim0)
SparseSegmentSqrtNGrad
operationgrad
- gradient propagated to the SparseSegmentSqrtN op.indices
- indices passed to the corresponding SparseSegmentSqrtN op.segmentIds
- segment_ids passed to the corresponding SparseSegmentSqrtN op.outputDim0
- dimension 0 of "data" passed to SparseSegmentSqrtN op.SparseSegmentSqrtNGrad
public <T extends Number,U extends Number> SparseSegmentSqrtN<T> sparseSegmentSqrtN(Operand<T> data, Operand<U> indices, Operand<Integer> segmentIds)
SparseSegmentSqrtN
operationdata
- indices
- A 1-D tensor. Has same rank as `segment_ids`.segmentIds
- A 1-D tensor. Values should be sorted and can be repeated.SparseSegmentSqrtN
public <T> SparseFillEmptyRows<T> sparseFillEmptyRows(Operand<Long> indices, Operand<T> values, Operand<Long> denseShape, Operand<T> defaultValue)
SparseFillEmptyRows
operationindices
- 2-D. the indices of the sparse tensor.values
- 1-D. the values of the sparse tensor.denseShape
- 1-D. the shape of the sparse tensor.defaultValue
- 0-D. default value to insert into location `[row, 0, ..., 0]`SparseFillEmptyRows
public <T> SparseAccumulatorApplyGradient sparseAccumulatorApplyGradient(Operand<String> handle, Operand<Long> localStep, Operand<Long> gradientIndices, Operand<T> gradientValues, Operand<Long> gradientShape, Boolean hasKnownShape)
SparseAccumulatorApplyGradient
operationhandle
- The handle to a accumulator.localStep
- The local_step value at which the sparse gradient was computed.gradientIndices
- Indices of the sparse gradient to be accumulated. Must be agradientValues
- Values are the non-zero slices of the gradient, and must havegradientShape
- Shape of the sparse gradient to be accumulated.hasKnownShape
- Boolean indicating whether gradient_shape is unknown, in whichSparseAccumulatorApplyGradient
public <U,T extends Number> SparseTensorDenseMatMul<U> sparseTensorDenseMatMul(Operand<T> aIndices, Operand<U> aValues, Operand<Long> aShape, Operand<U> b, SparseTensorDenseMatMul.Options... options)
SparseTensorDenseMatMul
operationaIndices
- 2-D. The `indices` of the `SparseTensor`, size `[nnz, 2]` Matrix.aValues
- 1-D. The `values` of the `SparseTensor`, size `[nnz]` Vector.aShape
- 1-D. The `shape` of the `SparseTensor`, size `[2]` Vector.b
- 2-D. A dense Matrix.options
- carries optional attributes valuesSparseTensorDenseMatMul
public <U,T extends Number> SparseTensorDenseAdd<U> sparseTensorDenseAdd(Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b)
SparseTensorDenseAdd
operationaIndices
- 2-D. The `indices` of the `SparseTensor`, with shape `[nnz, ndims]`.aValues
- 1-D. The `values` of the `SparseTensor`, with shape `[nnz]`.aShape
- 1-D. The `shape` of the `SparseTensor`, with shape `[ndims]`.b
- `ndims`-D Tensor. With shape `a_shape`.SparseTensorDenseAdd
public <T extends Number,U extends Number,V extends Number> SparseSegmentSqrtNWithNumSegments<T> sparseSegmentSqrtNWithNumSegments(Operand<T> data, Operand<U> indices, Operand<Integer> segmentIds, Operand<V> numSegments)
SparseSegmentSqrtNWithNumSegments
operationdata
- indices
- A 1-D tensor. Has same rank as `segment_ids`.segmentIds
- A 1-D tensor. Values should be sorted and can be repeated.numSegments
- Should equal the number of distinct segment IDs.SparseSegmentSqrtNWithNumSegments
public <U,T> DeserializeSparse<U> deserializeSparse(Operand<T> serializedSparse, Class<U> dtype)
DeserializeSparse
operationserializedSparse
- The serialized `SparseTensor` objects. The last dimensiondtype
- The `dtype` of the serialized `SparseTensor` objects.DeserializeSparse
public <T> TakeManySparseFromTensorsMap<T> takeManySparseFromTensorsMap(Operand<Long> sparseHandles, Class<T> dtype, TakeManySparseFromTensorsMap.Options... options)
TakeManySparseFromTensorsMap
operationsparseHandles
- 1-D, The `N` serialized `SparseTensor` objects.dtype
- The `dtype` of the `SparseTensor` objects stored in theoptions
- carries optional attributes valuesTakeManySparseFromTensorsMap
public <T extends Number> SparseReduceMaxSparse<T> sparseReduceMaxSparse(Operand<Long> inputIndices, Operand<T> inputValues, Operand<Long> inputShape, Operand<Integer> reductionAxes, SparseReduceMaxSparse.Options... options)
SparseReduceMaxSparse
operationinputIndices
- 2-D. `N x R` matrix with the indices of non-empty values in ainputValues
- 1-D. `N` non-empty values corresponding to `input_indices`.inputShape
- 1-D. Shape of the input SparseTensor.reductionAxes
- 1-D. Length-`K` vector containing the reduction axes.options
- carries optional attributes valuesSparseReduceMaxSparse
public <T,U extends Number> SparseAdd<T> sparseAdd(Operand<Long> aIndices, Operand<T> aValues, Operand<Long> aShape, Operand<Long> bIndices, Operand<T> bValues, Operand<Long> bShape, Operand<U> thresh)
SparseAdd
operationaIndices
- 2-D. The `indices` of the first `SparseTensor`, size `[nnz, ndims]` Matrix.aValues
- 1-D. The `values` of the first `SparseTensor`, size `[nnz]` Vector.aShape
- 1-D. The `shape` of the first `SparseTensor`, size `[ndims]` Vector.bIndices
- 2-D. The `indices` of the second `SparseTensor`, size `[nnz, ndims]` Matrix.bValues
- 1-D. The `values` of the second `SparseTensor`, size `[nnz]` Vector.bShape
- 1-D. The `shape` of the second `SparseTensor`, size `[ndims]` Vector.thresh
- 0-D. The magnitude threshold that determines if an output value/indexSparseAdd
public <T> AddSparseToTensorsMap addSparseToTensorsMap(Operand<Long> sparseIndices, Operand<T> sparseValues, Operand<Long> sparseShape, AddSparseToTensorsMap.Options... options)
AddSparseToTensorsMap
operationsparseIndices
- 2-D. The `indices` of the `SparseTensor`.sparseValues
- 1-D. The `values` of the `SparseTensor`.sparseShape
- 1-D. The `shape` of the `SparseTensor`.options
- carries optional attributes valuesAddSparseToTensorsMap
public <T extends Number,U extends Number> SparseSegmentMeanGrad<T> sparseSegmentMeanGrad(Operand<T> grad, Operand<U> indices, Operand<Integer> segmentIds, Operand<Integer> outputDim0)
SparseSegmentMeanGrad
operationgrad
- gradient propagated to the SparseSegmentMean op.indices
- indices passed to the corresponding SparseSegmentMean op.segmentIds
- segment_ids passed to the corresponding SparseSegmentMean op.outputDim0
- dimension 0 of "data" passed to SparseSegmentMean op.SparseSegmentMeanGrad
public <T extends Number,U extends Number> SparseMatMul sparseMatMul(Operand<T> a, Operand<U> b, SparseMatMul.Options... options)
SparseMatMul
operationa
- b
- options
- carries optional attributes valuesSparseMatMul
public <T> SparseFillEmptyRowsGrad<T> sparseFillEmptyRowsGrad(Operand<Long> reverseIndexMap, Operand<T> gradValues)
SparseFillEmptyRowsGrad
operationreverseIndexMap
- 1-D. The reverse index map from SparseFillEmptyRows.gradValues
- 1-D. The gradients from backprop.SparseFillEmptyRowsGrad
public <T> SparseSlice<T> sparseSlice(Operand<Long> indices, Operand<T> values, Operand<Long> shape, Operand<Long> start, Operand<Long> size)
SparseSlice
operationindices
- 2-D tensor represents the indices of the sparse tensor.values
- 1-D tensor represents the values of the sparse tensor.shape
- 1-D. tensor represents the shape of the sparse tensor.start
- 1-D. tensor represents the start of the slice.size
- 1-D. tensor represents the size of the slice.SparseSlice
public <T> SparseReduceSumSparse<T> sparseReduceSumSparse(Operand<Long> inputIndices, Operand<T> inputValues, Operand<Long> inputShape, Operand<Integer> reductionAxes, SparseReduceSumSparse.Options... options)
SparseReduceSumSparse
operationinputIndices
- 2-D. `N x R` matrix with the indices of non-empty values in ainputValues
- 1-D. `N` non-empty values corresponding to `input_indices`.inputShape
- 1-D. Shape of the input SparseTensor.reductionAxes
- 1-D. Length-`K` vector containing the reduction axes.options
- carries optional attributes valuesSparseReduceSumSparse
public <T> SparseDenseCwiseDiv<T> sparseDenseCwiseDiv(Operand<Long> spIndices, Operand<T> spValues, Operand<Long> spShape, Operand<T> dense)
SparseDenseCwiseDiv
operationspIndices
- 2-D. `N x R` matrix with the indices of non-empty values in aspValues
- 1-D. `N` non-empty values corresponding to `sp_indices`.spShape
- 1-D. Shape of the input SparseTensor.dense
- `R`-D. The dense Tensor operand.SparseDenseCwiseDiv
public <T> SparseReduceSum<T> sparseReduceSum(Operand<Long> inputIndices, Operand<T> inputValues, Operand<Long> inputShape, Operand<Integer> reductionAxes, SparseReduceSum.Options... options)
SparseReduceSum
operationinputIndices
- 2-D. `N x R` matrix with the indices of non-empty values in ainputValues
- 1-D. `N` non-empty values corresponding to `input_indices`.inputShape
- 1-D. Shape of the input SparseTensor.reductionAxes
- 1-D. Length-`K` vector containing the reduction axes.options
- carries optional attributes valuesSparseReduceSum
public <T> SparseSparseMinimum<T> sparseSparseMinimum(Operand<Long> aIndices, Operand<T> aValues, Operand<Long> aShape, Operand<Long> bIndices, Operand<T> bValues, Operand<Long> bShape)
SparseSparseMinimum
operationaIndices
- 2-D. `N x R` matrix with the indices of non-empty values in aaValues
- 1-D. `N` non-empty values corresponding to `a_indices`.aShape
- 1-D. Shape of the input SparseTensor.bIndices
- counterpart to `a_indices` for the other operand.bValues
- counterpart to `a_values` for the other operand; must be of the same dtype.bShape
- counterpart to `a_shape` for the other operand; the two shapes must be equal.SparseSparseMinimum
public <T extends Number,U extends Number,V extends Number> SparseSegmentSumWithNumSegments<T> sparseSegmentSumWithNumSegments(Operand<T> data, Operand<U> indices, Operand<Integer> segmentIds, Operand<V> numSegments)
SparseSegmentSumWithNumSegments
operationdata
- indices
- A 1-D tensor. Has same rank as `segment_ids`.segmentIds
- A 1-D tensor. Values should be sorted and can be repeated.numSegments
- Should equal the number of distinct segment IDs.SparseSegmentSumWithNumSegments
public <T extends Number,U extends Number> SparseSegmentMean<T> sparseSegmentMean(Operand<T> data, Operand<U> indices, Operand<Integer> segmentIds)
SparseSegmentMean
operationdata
- indices
- A 1-D tensor. Has same rank as `segment_ids`.segmentIds
- A 1-D tensor. Values should be sorted and can be repeated.SparseSegmentMean
public <T> AddManySparseToTensorsMap addManySparseToTensorsMap(Operand<Long> sparseIndices, Operand<T> sparseValues, Operand<Long> sparseShape, AddManySparseToTensorsMap.Options... options)
AddManySparseToTensorsMap
operationsparseIndices
- 2-D. The `indices` of the minibatch `SparseTensor`.sparseValues
- 1-D. The `values` of the minibatch `SparseTensor`.sparseShape
- 1-D. The `shape` of the minibatch `SparseTensor`.options
- carries optional attributes valuesAddManySparseToTensorsMap
public <T> SparseToSparseSetOperation<T> sparseToSparseSetOperation(Operand<Long> set1Indices, Operand<T> set1Values, Operand<Long> set1Shape, Operand<Long> set2Indices, Operand<T> set2Values, Operand<Long> set2Shape, String setOperation, SparseToSparseSetOperation.Options... options)
SparseToSparseSetOperation
operationset1Indices
- 2D `Tensor`, indices of a `SparseTensor`. Must be in row-majorset1Values
- 1D `Tensor`, values of a `SparseTensor`. Must be in row-majorset1Shape
- 1D `Tensor`, shape of a `SparseTensor`. `set1_shape[0...n-1]` mustset2Indices
- 2D `Tensor`, indices of a `SparseTensor`. Must be in row-majorset2Values
- 1D `Tensor`, values of a `SparseTensor`. Must be in row-majorset2Shape
- 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` mustsetOperation
- options
- carries optional attributes valuesSparseToSparseSetOperation
public <T> DenseToSparseSetOperation<T> denseToSparseSetOperation(Operand<T> set1, Operand<Long> set2Indices, Operand<T> set2Values, Operand<Long> set2Shape, String setOperation, DenseToSparseSetOperation.Options... options)
DenseToSparseSetOperation
operationset1
- `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.set2Indices
- 2D `Tensor`, indices of a `SparseTensor`. Must be in row-majorset2Values
- 1D `Tensor`, values of a `SparseTensor`. Must be in row-majorset2Shape
- 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` mustsetOperation
- options
- carries optional attributes valuesDenseToSparseSetOperation
public <T> SparseSplit<T> sparseSplit(Operand<Long> splitDim, Operand<Long> indices, Operand<T> values, Operand<Long> shape, Long numSplit)
SparseSplit
operationsplitDim
- 0-D. The dimension along which to split. Must be in the rangeindices
- 2-D tensor represents the indices of the sparse tensor.values
- 1-D tensor represents the values of the sparse tensor.shape
- 1-D. tensor represents the shape of the sparse tensor.numSplit
- The number of ways to split.SparseSplit
public <T> DenseToDenseSetOperation<T> denseToDenseSetOperation(Operand<T> set1, Operand<T> set2, String setOperation, DenseToDenseSetOperation.Options... options)
DenseToDenseSetOperation
operationset1
- `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.set2
- `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set1`.setOperation
- options
- carries optional attributes valuesDenseToDenseSetOperation
public <T> SparseAccumulatorTakeGradient<T> sparseAccumulatorTakeGradient(Operand<String> handle, Operand<Integer> numRequired, Class<T> dtype)
SparseAccumulatorTakeGradient
operationhandle
- The handle to a SparseConditionalAccumulator.numRequired
- Number of gradients required before we return an aggregate.dtype
- The data type of accumulated gradients. Needs to correspond to the typeSparseAccumulatorTakeGradient
public <T> SparseDenseCwiseMul<T> sparseDenseCwiseMul(Operand<Long> spIndices, Operand<T> spValues, Operand<Long> spShape, Operand<T> dense)
SparseDenseCwiseMul
operationspIndices
- 2-D. `N x R` matrix with the indices of non-empty values in aspValues
- 1-D. `N` non-empty values corresponding to `sp_indices`.spShape
- 1-D. Shape of the input SparseTensor.dense
- `R`-D. The dense Tensor operand.SparseDenseCwiseMul
public <T> SparseConditionalAccumulator sparseConditionalAccumulator(Class<T> dtype, Shape shape, SparseConditionalAccumulator.Options... options)
SparseConditionalAccumulator
operationdtype
- The type of the value being accumulated.shape
- The shape of the values.options
- carries optional attributes valuesSparseConditionalAccumulator
public <T extends Number> SparseReduceMax<T> sparseReduceMax(Operand<Long> inputIndices, Operand<T> inputValues, Operand<Long> inputShape, Operand<Integer> reductionAxes, SparseReduceMax.Options... options)
SparseReduceMax
operationinputIndices
- 2-D. `N x R` matrix with the indices of non-empty values in ainputValues
- 1-D. `N` non-empty values corresponding to `input_indices`.inputShape
- 1-D. Shape of the input SparseTensor.reductionAxes
- 1-D. Length-`K` vector containing the reduction axes.options
- carries optional attributes valuesSparseReduceMax
public <U,T extends Number> SparseToDense<U> sparseToDense(Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues, Operand<U> defaultValue, SparseToDense.Options... options)
SparseToDense
operationsparseIndices
- 0-D, 1-D, or 2-D. `sparse_indices[i]` contains the completeoutputShape
- 1-D. Shape of the dense output tensor.sparseValues
- 1-D. Values corresponding to each row of `sparse_indices`,defaultValue
- Scalar value to set for indices not specified inoptions
- carries optional attributes valuesSparseToDense
public <T> SparseDenseCwiseAdd<T> sparseDenseCwiseAdd(Operand<Long> spIndices, Operand<T> spValues, Operand<Long> spShape, Operand<T> dense)
SparseDenseCwiseAdd
operationspIndices
- 2-D. `N x R` matrix with the indices of non-empty values in aspValues
- 1-D. `N` non-empty values corresponding to `sp_indices`.spShape
- 1-D. Shape of the input SparseTensor.dense
- `R`-D. The dense Tensor operand.SparseDenseCwiseAdd
public <T,U> SparseCross<T> sparseCross(Iterable<Operand<Long>> indices, Iterable<Operand<?>> values, Iterable<Operand<Long>> shapes, Iterable<Operand<?>> denseInputs, Boolean hashedOutput, Long numBuckets, Long hashKey, Class<T> outType, Class<U> internalType)
SparseCross
operationindices
- 2-D. Indices of each input `SparseTensor`.values
- 1-D. values of each `SparseTensor`.shapes
- 1-D. Shapes of each `SparseTensor`.denseInputs
- 2-D. Columns represented by dense `Tensor`.hashedOutput
- If true, returns the hash of the cross instead of the string.numBuckets
- It is used if hashed_output is true.hashKey
- Specify the hash_key that will be used by the `FingerprintCat64`outType
- internalType
- SparseCross
public <T> SparseSliceGrad<T> sparseSliceGrad(Operand<T> backpropValGrad, Operand<Long> inputIndices, Operand<Long> inputStart, Operand<Long> outputIndices)
SparseSliceGrad
operationbackpropValGrad
- 1-D. The gradient with respect toinputIndices
- 2-D. The `indices` of the input `SparseTensor`.inputStart
- 1-D. tensor represents the start of the slice.outputIndices
- 2-D. The `indices` of the sliced `SparseTensor`.SparseSliceGrad
public <T extends Number,U extends Number,V extends Number> SparseSegmentMeanWithNumSegments<T> sparseSegmentMeanWithNumSegments(Operand<T> data, Operand<U> indices, Operand<Integer> segmentIds, Operand<V> numSegments)
SparseSegmentMeanWithNumSegments
operationdata
- indices
- A 1-D tensor. Has same rank as `segment_ids`.segmentIds
- A 1-D tensor. Values should be sorted and can be repeated.numSegments
- Should equal the number of distinct segment IDs.SparseSegmentMeanWithNumSegments
public <T extends Number> SparseSoftmax<T> sparseSoftmax(Operand<Long> spIndices, Operand<T> spValues, Operand<Long> spShape)
SparseSoftmax
operationspIndices
- 2-D. `NNZ x R` matrix with the indices of non-empty values in aspValues
- 1-D. `NNZ` non-empty values corresponding to `sp_indices`.spShape
- 1-D. Shape of the input SparseTensor.SparseSoftmax
public <T> SparseConcat<T> sparseConcat(Iterable<Operand<Long>> indices, Iterable<Operand<T>> values, Iterable<Operand<Long>> shapes, Long concatDim)
SparseConcat
operationindices
- 2-D. Indices of each input `SparseTensor`.values
- 1-D. Non-empty values of each `SparseTensor`.shapes
- 1-D. Shapes of each `SparseTensor`.concatDim
- Dimension to concatenate along. Must be in range [-rank, rank),SparseConcat
public <T> SparseAddGrad<T> sparseAddGrad(Operand<T> backpropValGrad, Operand<Long> aIndices, Operand<Long> bIndices, Operand<Long> sumIndices)
SparseAddGrad
operationbackpropValGrad
- 1-D with shape `[nnz(sum)]`. The gradient with respect toaIndices
- 2-D. The `indices` of the `SparseTensor` A, size `[nnz(A), ndims]`.bIndices
- 2-D. The `indices` of the `SparseTensor` B, size `[nnz(B), ndims]`.sumIndices
- 2-D. The `indices` of the sum `SparseTensor`, sizeSparseAddGrad
public <T extends Number> SparseSparseMaximum<T> sparseSparseMaximum(Operand<Long> aIndices, Operand<T> aValues, Operand<Long> aShape, Operand<Long> bIndices, Operand<T> bValues, Operand<Long> bShape)
SparseSparseMaximum
operationaIndices
- 2-D. `N x R` matrix with the indices of non-empty values in aaValues
- 1-D. `N` non-empty values corresponding to `a_indices`.aShape
- 1-D. Shape of the input SparseTensor.bIndices
- counterpart to `a_indices` for the other operand.bValues
- counterpart to `a_values` for the other operand; must be of the same dtype.bShape
- counterpart to `a_shape` for the other operand; the two shapes must be equal.SparseSparseMaximum
public <T extends Number,U extends Number> SparseSegmentSum<T> sparseSegmentSum(Operand<T> data, Operand<U> indices, Operand<Integer> segmentIds)
SparseSegmentSum
operationdata
- indices
- A 1-D tensor. Has same rank as `segment_ids`.segmentIds
- A 1-D tensor. Values should be sorted and can be repeated.SparseSegmentSum
public SparseReshape sparseReshape(Operand<Long> inputIndices, Operand<Long> inputShape, Operand<Long> newShape)
SparseReshape
operationinputIndices
- 2-D. `N x R_in` matrix with the indices of non-empty values in ainputShape
- 1-D. `R_in` vector with the input SparseTensor's dense shape.newShape
- 1-D. `R_out` vector with the requested new dense shape.SparseReshape
Copyright © 2022. All rights reserved.