ops

Type Members

1. class All[T] extends Operation[Table, Tensor[Boolean], T]

2. class Any[T] extends Operation[Table, Tensor[Boolean], T]

3. class ApproximateEqual[T] extends Compare[T]

4. class ArgMax[T] extends Operation[Table, Tensor[Int], T]

5. class Assert[T] extends Operation[Table, Activity, T]

   Assert will assert the first input to be true, if not, throw the message in the second input.

6. class Assign[T] extends Operation[Table, Tensor[_], T]

   Update 'ref' by assigning 'value' to it.

7. class AssignGrad[T] extends Operation[Tensor[T], Activity, T]

8. class AvgPoolGrad[T] extends Operation[Table, Tensor[T], T]

9. class BatchMatMul[T, D] extends Operation[Table, Tensor[D], T]

   Multiplies all slices of Tensor x and y (each slice can be viewed as an element of a batch), and arranges the individual results in a single output tensor of the same batch size.

10. class BiasAddGrad[T] extends Operation[Tensor[T], Tensor[T], T]

11. class BroadcastGradientArgs[T] extends Operation[Table, Table, T]

    Given shapes of two tensors, computes the reduction indices for the gradient computation.

12. class Cast[T, D] extends Operation[Tensor[_], Tensor[D], T]

    Casts a tensor to a new type.

13. class Ceil[T, D] extends Operation[Tensor[D], Tensor[D], T]

14. abstract class Compare[T] extends Operation[Table, Tensor[Boolean], T]

15. sealed abstract class ControlOps[T] extends Operation[Activity, Activity, T]

    Control flow related operations

16. class Conv2D[T] extends Operation[Table, Tensor[T], T]

17. class Conv2DBackFilter[T] extends Operation[Activity, Tensor[T], T]

18. class Conv2DTranspose[T] extends Operation[Activity, Tensor[T], T]

    Backward of SpatialConvolution

19. class Conv3D[T] extends Operation[Table, Tensor[T], T]

20. class Conv3DBackpropFilter[T] extends Operation[Table, Tensor[T], T]

21. class Conv3DBackpropFilterV2[T] extends Conv3DBackpropFilter[T]

22. class Conv3DBackpropInput[T] extends Operation[Table, Tensor[T], T]

23. class Conv3DBackpropInputV2[T] extends Conv3DBackpropInput[T]

24. class CrossEntropy[T] extends Operation[Table, Table, T]

    Compute the cross entropy loss and the gradients.

25. class DecodeGif[T] extends DecodeImage[T]

26. class DecodeImage[T] extends Operation[Tensor[ByteString], Tensor[Int], T]

27. class DecodeJpeg[T] extends DecodeImage[T]

28. class DecodePng[T] extends DecodeImage[T]

29. class DecodeRaw[T] extends Operation[Tensor[ByteString], Activity, T]

30. class DepthwiseConv2D[T] extends Operation[Table, Tensor[T], T]

31. class DepthwiseConv2DBackpropFilter[T] extends Operation[Table, Tensor[T], T]

32. class DepthwiseConv2DBackpropInput[T] extends Operation[Table, Tensor[T], T]

33. class Digamma[T, D] extends Operation[Tensor[D], Tensor[D], T]

34. class Dilation2D[T, D] extends Operation[Table, Tensor[D], T]

    Computes the grayscale dilation of 4-D input and 3-D filter tensors.

35. class Dilation2DBackpropFilter[T, D] extends Operation[Table, Tensor[D], T]

36. class Dilation2DBackpropInput[T, D] extends Operation[Table, Tensor[D], T]

37. class EluGrad[T, D] extends UnaryGrad[T, D]

38. class Equal[T] extends Compare[T]

39. class Erf[T, D] extends Operation[Tensor[D], Tensor[D], T]

40. class Erfc[T, D] extends Operation[Tensor[D], Tensor[D], T]

41. class Exp[T, D] extends Operation[Tensor[D], Tensor[D], T]

42. class Expm1[T, D] extends Operation[Tensor[D], Tensor[D], T]

43. class Floor[T] extends Operation[Tensor[_], Tensor[_], T]

44. class FloorDiv[T, D] extends Operation[Table, Tensor[D], T]

45. class FloorMod[T, D] extends Operation[Table, Tensor[D], T]

46. class FusedBatchNorm[T] extends Operation[Table, Table, T]

    This is similar to SpatialBatchNormalization.

47. class FusedBatchNormGrad[T] extends Operation[Table, Table, T]

    This is the gradient operation coressponding to the FusedBatchNorm.

48. class Greater[T] extends Compare[T]

49. class GreaterEqual[T] extends Compare[T]

50. abstract class IdentityControl[T] extends ControlOps[T]

    Control flow related operations, and they just pass the input without modifying them

51. class InTopK[T] extends Operation[Table, Tensor[Boolean], T]

52. class Inv[T, D] extends Operation[Tensor[D], Tensor[D], T]

53. class InvGrad[T, D] extends Operation[Table, Tensor[D], T]

54. class IsFinite[T, D] extends Operation[Tensor[D], Tensor[Boolean], T]

55. class IsInf[T, D] extends Operation[Tensor[D], Tensor[Boolean], T]

56. class IsNan[T, D] extends Operation[Tensor[D], Tensor[Boolean], T]

57. class L2Loss[T] extends Operation[Tensor[_], Tensor[_], T]

58. class LRNGrad[T] extends Operation[Table, Tensor[Float], T]

    LRNGrad calculate the backprop gradients of the Local response normalization layer.

59. class Less[T] extends Compare[T]

60. class LessEqual[T] extends Compare[T]

61. class Lgamma[T, D] extends Operation[Tensor[D], Tensor[D], T]

62. class LogicalAnd[T] extends Operation[Table, Tensor[Boolean], T]

63. class LogicalNot[T] extends Operation[Tensor[Boolean], Tensor[Boolean], T]

64. class LogicalOr[T] extends Operation[Table, Tensor[Boolean], T]

65. class MaxPool[T] extends Operation[Tensor[T], Tensor[T], T]

66. class MaxPoolGrad[T] extends Operation[Table, Tensor[T], T]

67. class Maximum[T, D] extends Operation[Table, Tensor[D], T]

68. sealed class MergeControlNode[T] extends Node[T]

    A wrapper of node for merge operation.

69. class Minimum[T, D] extends Operation[Table, Tensor[D], T]

70. class Mod[T, D] extends Operation[Table, Tensor[D], T]

71. class ModuleToOperation[T] extends Operation[Activity, Activity, T]

    Wrap a nn module to an Operation

72. class NoOp[T] extends Operation[Activity, Activity, T] with WithoutInput

73. class NotEqual[T] extends Operation[Table, Tensor[Boolean], T]

74. class OneHot[T, D] extends Operation[Table, Tensor[D], T]

    OneHot operation returns a one-hot tensor

75. abstract class Operation[A <: Activity, B <: Activity, T] extends AbstractModule[A, B, T]

    Operation is an abstract class which represents the most basic operations An operations has only forward functions and without backward functions.

76. class Pad[T, D] extends Operation[Table, Tensor[D], T]

77. class ParseExample[T] extends Operation[Table, Table, T]

78. class Pow[T] extends Operation[Table, Tensor[_], T]

79. class Prod[T] extends Operation[Tensor[_], Tensor[_], T]

80. class RandomUniform[T, D] extends Operation[Tensor[Int], Tensor[D], T] with RandomNode

81. class RangeOps[T, D] extends Operation[Table, Tensor[D], T]

82. class Rank[T] extends Operation[Tensor[_], Tensor[Int], T]

83. class Relu6Grad[T, D] extends UnaryGrad[T, D]

84. class ReluGrad[T] extends Operation[Table, Tensor[T], T]

85. class ResizeBilinearOps[T] extends Operation[Activity, Tensor[Float], T]

86. class Rint[T] extends Operation[Tensor[Float], Tensor[Float], T]

87. class Round[T, D] extends Operation[Tensor[D], Tensor[D], T]

88. class RsqrtGrad[T, D] extends Operation[Table, Tensor[D], T]

89. class SegmentSum[T] extends Operation[Table, Tensor[T], T]

    Computes the sum along segments of a tensor.

90. class Select[T] extends Operation[Table, Activity, T]

    Selects elements from input, depending on given condition.

91. class SigmoidGrad[T, D] extends Operation[Table, Tensor[D], T]

92. class Sign[T, D] extends Operation[Tensor[D], Tensor[D], T]

93. class Slice[T] extends Operation[Tensor[_], Tensor[_], T]

    This operation extracts a slice of size size from a tensor input starting at the location specified by begin.

94. class SoftplusGrad[T, D] extends UnaryGrad[T, D]

95. class SoftsignGrad[T, D] extends UnaryGrad[T, D]

96. class SqrtGrad[T, D] extends Operation[Table, Tensor[D], T]

97. class SquaredDifference[T] extends Operation[Table, Tensor[_], T]

    Returns (x - y)(x - y) element-wise.

98. class Substr[T] extends Operation[Table, Tensor[ByteString], T]

99. class Sum[T, D] extends Operation[Table, Tensor[D], T]

100. sealed class SwitchControlNode[T] extends Node[T]

     A wrapper of node for switch operation.

101. class TanhGrad[T, D] extends Operation[Table, Tensor[D], T]

102. class Tile[T] extends Operation[Table, Tensor[_], T]

     This operation creates a new tensor by replicating input multiples times.

103. class TopK[T, D] extends Operation[Tensor[D], Table, T]

104. class TruncateDiv[T, D] extends Operation[Table, Tensor[D], T]

105. class TruncatedNormal[T, DataType] extends Operation[Tensor[Int], Tensor[DataType], T]

106. abstract class UnaryGrad[T, D] extends Operation[Table, Tensor[D], T]