Contents Menu Expand Light mode Dark mode Auto light/dark, in light mode Auto light/dark, in dark mode 跳至内容
ONNX 1.18.0 文档
Light Logo Dark Logo
ONNX 1.18.0 文档
  • ONNX 简介
    • ONNX 概念
    • 使用 Python 的 ONNX
    • 转换器
  • API 参考
    • Protos
    • 序列化
    • onnx.backend
    • onnx.checker
    • onnx.compose
    • onnx._custom_element_types
    • onnx.defs
    • onnx.external_data_helper
    • onnx.helper
    • onnx.hub
    • onnx.inliner
    • onnx.mapping
    • onnx.model_container
    • onnx.numpy_helper
    • onnx.parser
    • onnx.printer
    • onnx.reference
    • onnx.shape_inference
    • onnx.tools
    • onnx.utils
    • onnx.version_converter
  • ONNX 算子
    • 示例算子测试代码
    • Abs
      • Abs - 6 vs 13
      • Abs - 1 vs 13
      • Abs - 1 vs 6
    • Acos
      • Acos - 7 vs 22
    • Acosh
      • Acosh - 9 vs 22
    • Add
      • Add - 13 vs 14
      • Add - 7 vs 14
      • Add - 7 vs 13
      • Add - 6 vs 14
      • Add - 6 vs 13
      • Add - 6 vs 7
      • Add - 1 vs 14
      • Add - 1 vs 13
      • Add - 1 vs 7
      • Add - 1 vs 6
    • AffineGrid
    • And
      • And - 1 vs 7
    • ArgMax
      • ArgMax - 12 vs 13
      • ArgMax - 11 vs 13
      • ArgMax - 11 vs 12
      • ArgMax - 1 vs 13
      • ArgMax - 1 vs 12
      • ArgMax - 1 vs 11
    • ArgMin
      • ArgMin - 12 vs 13
      • ArgMin - 11 vs 13
      • ArgMin - 11 vs 12
      • ArgMin - 1 vs 13
      • ArgMin - 1 vs 12
      • ArgMin - 1 vs 11
    • Asin
      • Asin - 7 vs 22
    • Asinh
      • Asinh - 9 vs 22
    • Atan
      • Atan - 7 vs 22
    • Atanh
      • Atanh - 9 vs 22
    • AveragePool
      • AveragePool - 19 vs 22
      • AveragePool - 11 vs 22
      • AveragePool - 11 vs 19
      • AveragePool - 10 vs 22
      • AveragePool - 10 vs 19
      • AveragePool - 10 vs 11
      • AveragePool - 7 vs 22
      • AveragePool - 7 vs 19
      • AveragePool - 7 vs 11
      • AveragePool - 7 vs 10
      • AveragePool - 1 vs 22
      • AveragePool - 1 vs 19
      • AveragePool - 1 vs 11
      • AveragePool - 1 vs 10
      • AveragePool - 1 vs 7
    • BatchNormalization
      • BatchNormalization - 14 vs 15
      • BatchNormalization - 9 vs 15
      • BatchNormalization - 9 vs 14
      • BatchNormalization - 7 vs 15
      • BatchNormalization - 7 vs 14
      • BatchNormalization - 7 vs 9
      • BatchNormalization - 6 vs 15
      • BatchNormalization - 6 vs 14
      • BatchNormalization - 6 vs 9
      • BatchNormalization - 6 vs 7
      • BatchNormalization - 1 vs 15
      • BatchNormalization - 1 vs 14
      • BatchNormalization - 1 vs 9
      • BatchNormalization - 1 vs 7
      • BatchNormalization - 1 vs 6
    • Bernoulli
      • Bernoulli - 15 vs 22
    • BitShift
    • BitwiseAnd
    • BitwiseNot
    • BitwiseOr
    • BitwiseXor
    • BlackmanWindow
    • Cast
      • Cast - 21 vs 23
      • Cast - 19 vs 23
      • Cast - 19 vs 21
      • Cast - 13 vs 23
      • Cast - 13 vs 21
      • Cast - 13 vs 19
      • Cast - 9 vs 23
      • Cast - 9 vs 21
      • Cast - 9 vs 19
      • Cast - 9 vs 13
      • Cast - 6 vs 23
      • Cast - 6 vs 21
      • Cast - 6 vs 19
      • Cast - 6 vs 13
      • Cast - 6 vs 9
      • Cast - 1 vs 23
      • Cast - 1 vs 21
      • Cast - 1 vs 19
      • Cast - 1 vs 13
      • Cast - 1 vs 9
      • Cast - 1 vs 6
    • CastLike
      • CastLike - 21 vs 23
      • CastLike - 19 vs 23
      • CastLike - 19 vs 21
      • CastLike - 15 vs 23
      • CastLike - 15 vs 21
      • CastLike - 15 vs 19
    • Ceil
      • Ceil - 6 vs 13
      • Ceil - 1 vs 13
      • Ceil - 1 vs 6
    • Celu
    • CenterCropPad
    • Clip
      • Clip - 12 vs 13
      • Clip - 11 vs 13
      • Clip - 11 vs 12
      • Clip - 6 vs 13
      • Clip - 6 vs 12
      • Clip - 6 vs 11
      • Clip - 1 vs 13
      • Clip - 1 vs 12
      • Clip - 1 vs 11
      • Clip - 1 vs 6
    • Col2Im
    • Compress
      • Compress - 9 vs 11
    • Concat
      • Concat - 11 vs 13
      • Concat - 4 vs 13
      • Concat - 4 vs 11
      • Concat - 1 vs 13
      • Concat - 1 vs 11
      • Concat - 1 vs 4
    • ConcatFromSequence
    • Constant
      • Constant - 21 vs 23
      • Constant - 19 vs 23
      • Constant - 19 vs 21
      • Constant - 13 vs 23
      • Constant - 13 vs 21
      • Constant - 13 vs 19
      • Constant - 12 vs 23
      • Constant - 12 vs 21
      • Constant - 12 vs 19
      • Constant - 12 vs 13
      • Constant - 11 vs 23
      • Constant - 11 vs 21
      • Constant - 11 vs 19
      • Constant - 11 vs 13
      • Constant - 11 vs 12
      • Constant - 9 vs 23
      • Constant - 9 vs 21
      • Constant - 9 vs 19
      • Constant - 9 vs 13
      • Constant - 9 vs 12
      • Constant - 9 vs 11
      • Constant - 1 vs 23
      • Constant - 1 vs 21
      • Constant - 1 vs 19
      • Constant - 1 vs 13
      • Constant - 1 vs 12
      • Constant - 1 vs 11
      • Constant - 1 vs 9
    • ConstantOfShape
      • ConstantOfShape - 21 vs 23
      • ConstantOfShape - 20 vs 23
      • ConstantOfShape - 20 vs 21
      • ConstantOfShape - 9 vs 23
      • ConstantOfShape - 9 vs 21
      • ConstantOfShape - 9 vs 20
    • Conv
      • Conv - 11 vs 22
      • Conv - 1 vs 22
      • Conv - 1 vs 11
    • ConvInteger
    • ConvTranspose
      • ConvTranspose - 11 vs 22
      • ConvTranspose - 1 vs 22
      • ConvTranspose - 1 vs 11
    • Cos
      • Cos - 7 vs 22
    • Cosh
      • Cosh - 9 vs 22
    • CumSum
      • CumSum - 11 vs 14
    • DFT
      • DFT - 17 vs 20
    • DeformConv
      • DeformConv - 19 vs 22
    • DepthToSpace
      • DepthToSpace - 11 vs 13
      • DepthToSpace - 1 vs 13
      • DepthToSpace - 1 vs 11
    • DequantizeLinear
      • DequantizeLinear - 21 vs 23
      • DequantizeLinear - 19 vs 23
      • DequantizeLinear - 19 vs 21
      • DequantizeLinear - 13 vs 23
      • DequantizeLinear - 13 vs 21
      • DequantizeLinear - 13 vs 19
      • DequantizeLinear - 10 vs 23
      • DequantizeLinear - 10 vs 21
      • DequantizeLinear - 10 vs 19
      • DequantizeLinear - 10 vs 13
    • Det
      • Det - 11 vs 22
    • Div
      • Div - 13 vs 14
      • Div - 7 vs 14
      • Div - 7 vs 13
      • Div - 6 vs 14
      • Div - 6 vs 13
      • Div - 6 vs 7
      • Div - 1 vs 14
      • Div - 1 vs 13
      • Div - 1 vs 7
      • Div - 1 vs 6
    • Dropout
      • Dropout - 13 vs 22
      • Dropout - 12 vs 22
      • Dropout - 12 vs 13
      • Dropout - 10 vs 22
      • Dropout - 10 vs 13
      • Dropout - 10 vs 12
      • Dropout - 7 vs 22
      • Dropout - 7 vs 13
      • Dropout - 7 vs 12
      • Dropout - 7 vs 10
      • Dropout - 6 vs 22
      • Dropout - 6 vs 13
      • Dropout - 6 vs 12
      • Dropout - 6 vs 10
      • Dropout - 6 vs 7
      • Dropout - 1 vs 22
      • Dropout - 1 vs 13
      • Dropout - 1 vs 12
      • Dropout - 1 vs 10
      • Dropout - 1 vs 7
      • Dropout - 1 vs 6
    • DynamicQuantizeLinear
    • Einsum
    • Elu
      • Elu - 6 vs 22
      • Elu - 1 vs 22
      • Elu - 1 vs 6
    • Equal
      • Equal - 13 vs 19
      • Equal - 11 vs 19
      • Equal - 11 vs 13
      • Equal - 7 vs 19
      • Equal - 7 vs 13
      • Equal - 7 vs 11
      • Equal - 1 vs 19
      • Equal - 1 vs 13
      • Equal - 1 vs 11
      • Equal - 1 vs 7
    • Erf
      • Erf - 9 vs 13
    • Exp
      • Exp - 6 vs 13
      • Exp - 1 vs 13
      • Exp - 1 vs 6
    • Expand
      • Expand - 8 vs 13
    • EyeLike
      • EyeLike - 9 vs 22
    • Flatten
      • Flatten - 21 vs 23
      • Flatten - 13 vs 23
      • Flatten - 13 vs 21
      • Flatten - 11 vs 23
      • Flatten - 11 vs 21
      • Flatten - 11 vs 13
      • Flatten - 9 vs 23
      • Flatten - 9 vs 21
      • Flatten - 9 vs 13
      • Flatten - 9 vs 11
      • Flatten - 1 vs 23
      • Flatten - 1 vs 21
      • Flatten - 1 vs 13
      • Flatten - 1 vs 11
      • Flatten - 1 vs 9
    • Floor
      • Floor - 6 vs 13
      • Floor - 1 vs 13
      • Floor - 1 vs 6
    • GRU
      • GRU - 14 vs 22
      • GRU - 7 vs 22
      • GRU - 7 vs 14
      • GRU - 3 vs 22
      • GRU - 3 vs 14
      • GRU - 3 vs 7
      • GRU - 1 vs 22
      • GRU - 1 vs 14
      • GRU - 1 vs 7
      • GRU - 1 vs 3
    • Gather
      • Gather - 11 vs 13
      • Gather - 1 vs 13
      • Gather - 1 vs 11
    • GatherElements
      • GatherElements - 11 vs 13
    • GatherND
      • GatherND - 12 vs 13
      • GatherND - 11 vs 13
      • GatherND - 11 vs 12
    • Gelu
    • Gemm
      • Gemm - 11 vs 13
      • Gemm - 9 vs 13
      • Gemm - 9 vs 11
      • Gemm - 7 vs 13
      • Gemm - 7 vs 11
      • Gemm - 7 vs 9
      • Gemm - 6 vs 13
      • Gemm - 6 vs 11
      • Gemm - 6 vs 9
      • Gemm - 6 vs 7
      • Gemm - 1 vs 13
      • Gemm - 1 vs 11
      • Gemm - 1 vs 9
      • Gemm - 1 vs 7
      • Gemm - 1 vs 6
    • GlobalAveragePool
      • GlobalAveragePool - 1 vs 22
    • GlobalLpPool
      • GlobalLpPool - 2 vs 22
      • GlobalLpPool - 1 vs 22
      • GlobalLpPool - 1 vs 2
    • GlobalMaxPool
      • GlobalMaxPool - 1 vs 22
    • Greater
      • Greater - 9 vs 13
      • Greater - 7 vs 13
      • Greater - 7 vs 9
      • Greater - 1 vs 13
      • Greater - 1 vs 9
      • Greater - 1 vs 7
    • GreaterOrEqual
      • GreaterOrEqual - 12 vs 16
    • GridSample
      • GridSample - 20 vs 22
      • GridSample - 16 vs 22
      • GridSample - 16 vs 20
    • GroupNormalization
      • GroupNormalization - 18 vs 21
    • HammingWindow
    • HannWindow
    • HardSigmoid
      • HardSigmoid - 6 vs 22
      • HardSigmoid - 1 vs 22
      • HardSigmoid - 1 vs 6
    • HardSwish
      • HardSwish - 14 vs 22
    • Hardmax
      • Hardmax - 11 vs 13
      • Hardmax - 1 vs 13
      • Hardmax - 1 vs 11
    • Identity
      • Identity - 21 vs 23
      • Identity - 19 vs 23
      • Identity - 19 vs 21
      • Identity - 16 vs 23
      • Identity - 16 vs 21
      • Identity - 16 vs 19
      • Identity - 14 vs 23
      • Identity - 14 vs 21
      • Identity - 14 vs 19
      • Identity - 14 vs 16
      • Identity - 13 vs 23
      • Identity - 13 vs 21
      • Identity - 13 vs 19
      • Identity - 13 vs 16
      • Identity - 13 vs 14
      • Identity - 1 vs 23
      • Identity - 1 vs 21
      • Identity - 1 vs 19
      • Identity - 1 vs 16
      • Identity - 1 vs 14
      • Identity - 1 vs 13
    • If
      • If - 21 vs 23
      • If - 19 vs 23
      • If - 19 vs 21
      • If - 16 vs 23
      • If - 16 vs 21
      • If - 16 vs 19
      • If - 13 vs 23
      • If - 13 vs 21
      • If - 13 vs 19
      • If - 13 vs 16
      • If - 11 vs 23
      • If - 11 vs 21
      • If - 11 vs 19
      • If - 11 vs 16
      • If - 11 vs 13
      • If - 1 vs 23
      • If - 1 vs 21
      • If - 1 vs 19
      • If - 1 vs 16
      • If - 1 vs 13
      • If - 1 vs 11
    • ImageDecoder
    • InstanceNormalization
      • InstanceNormalization - 6 vs 22
      • InstanceNormalization - 1 vs 22
      • InstanceNormalization - 1 vs 6
    • IsInf
      • IsInf - 10 vs 20
    • IsNaN
      • IsNaN - 13 vs 20
      • IsNaN - 9 vs 20
      • IsNaN - 9 vs 13
    • LRN
      • LRN - 1 vs 13
    • LSTM
      • LSTM - 14 vs 22
      • LSTM - 7 vs 22
      • LSTM - 7 vs 14
      • LSTM - 1 vs 22
      • LSTM - 1 vs 14
      • LSTM - 1 vs 7
    • LayerNormalization
    • LeakyRelu
      • LeakyRelu - 6 vs 16
      • LeakyRelu - 1 vs 16
      • LeakyRelu - 1 vs 6
    • Less
      • Less - 9 vs 13
      • Less - 7 vs 13
      • Less - 7 vs 9
      • Less - 1 vs 13
      • Less - 1 vs 9
      • Less - 1 vs 7
    • LessOrEqual
      • LessOrEqual - 12 vs 16
    • Log
      • Log - 6 vs 13
      • Log - 1 vs 13
      • Log - 1 vs 6
    • LogSoftmax
      • LogSoftmax - 11 vs 13
      • LogSoftmax - 1 vs 13
      • LogSoftmax - 1 vs 11
    • Loop
      • Loop - 21 vs 23
      • Loop - 19 vs 23
      • Loop - 19 vs 21
      • Loop - 16 vs 23
      • Loop - 16 vs 21
      • Loop - 16 vs 19
      • Loop - 13 vs 23
      • Loop - 13 vs 21
      • Loop - 13 vs 19
      • Loop - 13 vs 16
      • Loop - 11 vs 23
      • Loop - 11 vs 21
      • Loop - 11 vs 19
      • Loop - 11 vs 16
      • Loop - 11 vs 13
      • Loop - 1 vs 23
      • Loop - 1 vs 21
      • Loop - 1 vs 19
      • Loop - 1 vs 16
      • Loop - 1 vs 13
      • Loop - 1 vs 11
    • LpNormalization
      • LpNormalization - 1 vs 22
    • LpPool
      • LpPool - 18 vs 22
      • LpPool - 11 vs 22
      • LpPool - 11 vs 18
      • LpPool - 2 vs 22
      • LpPool - 2 vs 18
      • LpPool - 2 vs 11
      • LpPool - 1 vs 22
      • LpPool - 1 vs 18
      • LpPool - 1 vs 11
      • LpPool - 1 vs 2
    • MatMul
      • MatMul - 9 vs 13
      • MatMul - 1 vs 13
      • MatMul - 1 vs 9
    • MatMulInteger
    • Max
      • Max - 12 vs 13
      • Max - 8 vs 13
      • Max - 8 vs 12
      • Max - 6 vs 13
      • Max - 6 vs 12
      • Max - 6 vs 8
      • Max - 1 vs 13
      • Max - 1 vs 12
      • Max - 1 vs 8
      • Max - 1 vs 6
    • MaxPool
      • MaxPool - 12 vs 22
      • MaxPool - 11 vs 22
      • MaxPool - 11 vs 12
      • MaxPool - 10 vs 22
      • MaxPool - 10 vs 12
      • MaxPool - 10 vs 11
      • MaxPool - 8 vs 22
      • MaxPool - 8 vs 12
      • MaxPool - 8 vs 11
      • MaxPool - 8 vs 10
      • MaxPool - 1 vs 22
      • MaxPool - 1 vs 12
      • MaxPool - 1 vs 11
      • MaxPool - 1 vs 10
      • MaxPool - 1 vs 8
    • MaxRoiPool
      • MaxRoiPool - 1 vs 22
    • MaxUnpool
      • MaxUnpool - 11 vs 22
      • MaxUnpool - 9 vs 22
      • MaxUnpool - 9 vs 11
    • Mean
      • Mean - 8 vs 13
      • Mean - 6 vs 13
      • Mean - 6 vs 8
      • Mean - 1 vs 13
      • Mean - 1 vs 8
      • Mean - 1 vs 6
    • MeanVarianceNormalization
      • MeanVarianceNormalization - 9 vs 13
    • MelWeightMatrix
    • Min
      • Min - 12 vs 13
      • Min - 8 vs 13
      • Min - 8 vs 12
      • Min - 6 vs 13
      • Min - 6 vs 12
      • Min - 6 vs 8
      • Min - 1 vs 13
      • Min - 1 vs 12
      • Min - 1 vs 8
      • Min - 1 vs 6
    • Mish
      • Mish - 18 vs 22
    • Mod
      • Mod - 10 vs 13
    • Mul
      • Mul - 13 vs 14
      • Mul - 7 vs 14
      • Mul - 7 vs 13
      • Mul - 6 vs 14
      • Mul - 6 vs 13
      • Mul - 6 vs 7
      • Mul - 1 vs 14
      • Mul - 1 vs 13
      • Mul - 1 vs 7
      • Mul - 1 vs 6
    • Multinomial
      • Multinomial - 7 vs 22
    • Neg
      • Neg - 6 vs 13
      • Neg - 1 vs 13
      • Neg - 1 vs 6
    • NegativeLogLikelihoodLoss
      • NegativeLogLikelihoodLoss - 13 vs 22
      • NegativeLogLikelihoodLoss - 12 vs 22
      • NegativeLogLikelihoodLoss - 12 vs 13
    • NonMaxSuppression
      • NonMaxSuppression - 10 vs 11
    • NonZero
      • NonZero - 9 vs 13
    • Not
    • OneHot
      • OneHot - 9 vs 11
    • Optional
    • OptionalGetElement
      • OptionalGetElement - 15 vs 18
    • OptionalHasElement
      • OptionalHasElement - 15 vs 18
    • Or
      • Or - 1 vs 7
    • PRelu
      • PRelu - 9 vs 16
      • PRelu - 7 vs 16
      • PRelu - 7 vs 9
      • PRelu - 6 vs 16
      • PRelu - 6 vs 9
      • PRelu - 6 vs 7
      • PRelu - 1 vs 16
      • PRelu - 1 vs 9
      • PRelu - 1 vs 7
      • PRelu - 1 vs 6
    • Pad
      • Pad - 21 vs 23
      • Pad - 19 vs 23
      • Pad - 19 vs 21
      • Pad - 18 vs 23
      • Pad - 18 vs 21
      • Pad - 18 vs 19
      • Pad - 13 vs 23
      • Pad - 13 vs 21
      • Pad - 13 vs 19
      • Pad - 13 vs 18
      • Pad - 11 vs 23
      • Pad - 11 vs 21
      • Pad - 11 vs 19
      • Pad - 11 vs 18
      • Pad - 11 vs 13
      • Pad - 2 vs 23
      • Pad - 2 vs 21
      • Pad - 2 vs 19
      • Pad - 2 vs 18
      • Pad - 2 vs 13
      • Pad - 2 vs 11
      • Pad - 1 vs 23
      • Pad - 1 vs 21
      • Pad - 1 vs 19
      • Pad - 1 vs 18
      • Pad - 1 vs 13
      • Pad - 1 vs 11
      • Pad - 1 vs 2
    • Pow
      • Pow - 13 vs 15
      • Pow - 12 vs 15
      • Pow - 12 vs 13
      • Pow - 7 vs 15
      • Pow - 7 vs 13
      • Pow - 7 vs 12
      • Pow - 1 vs 15
      • Pow - 1 vs 13
      • Pow - 1 vs 12
      • Pow - 1 vs 7
    • QLinearConv
    • QLinearMatMul
      • QLinearMatMul - 10 vs 21
    • QuantizeLinear
      • QuantizeLinear - 21 vs 23
      • QuantizeLinear - 19 vs 23
      • QuantizeLinear - 19 vs 21
      • QuantizeLinear - 13 vs 23
      • QuantizeLinear - 13 vs 21
      • QuantizeLinear - 13 vs 19
      • QuantizeLinear - 10 vs 23
      • QuantizeLinear - 10 vs 21
      • 量化线性 - 10 vs 19
      • 量化线性 - 10 vs 13
    • 循环神经网络 (RNN)
      • RNN - 14 vs 22
      • RNN - 7 vs 22
      • RNN - 7 vs 14
      • RNN - 1 vs 22
      • RNN - 1 vs 14
      • RNN - 1 vs 7
    • 随机正态分布 (RandomNormal)
      • RandomNormal - 1 vs 22
    • 类似随机正态分布 (RandomNormalLike)
      • RandomNormalLike - 1 vs 22
    • 随机均匀分布 (RandomUniform)
      • RandomUniform - 1 vs 22
    • 类似随机均匀分布 (RandomUniformLike)
      • RandomUniformLike - 1 vs 22
    • 范围 (Range)
    • 倒数 (Reciprocal)
      • Reciprocal - 6 vs 13
      • Reciprocal - 1 vs 13
      • Reciprocal - 1 vs 6
    • L1 归约 (ReduceL1)
      • ReduceL1 - 13 vs 18
      • ReduceL1 - 11 vs 18
      • ReduceL1 - 11 vs 13
      • ReduceL1 - 1 vs 18
      • ReduceL1 - 1 vs 13
      • ReduceL1 - 1 vs 11
    • L2 归约 (ReduceL2)
      • ReduceL2 - 13 vs 18
      • ReduceL2 - 11 vs 18
      • ReduceL2 - 11 vs 13
      • ReduceL2 - 1 vs 18
      • ReduceL2 - 1 vs 13
      • ReduceL2 - 1 vs 11
    • 对数和归约 (ReduceLogSum)
      • ReduceLogSum - 13 vs 18
      • ReduceLogSum - 11 vs 18
      • ReduceLogSum - 11 vs 13
      • ReduceLogSum - 1 vs 18
      • ReduceLogSum - 1 vs 13
      • ReduceLogSum - 1 vs 11
    • 对数和指数归约 (ReduceLogSumExp)
      • ReduceLogSumExp - 13 vs 18
      • ReduceLogSumExp - 11 vs 18
      • ReduceLogSumExp - 11 vs 13
      • ReduceLogSumExp - 1 vs 18
      • ReduceLogSumExp - 1 vs 13
      • ReduceLogSumExp - 1 vs 11
    • 最大值归约 (ReduceMax)
      • ReduceMax - 18 vs 20
      • ReduceMax - 13 vs 20
      • ReduceMax - 13 vs 18
      • ReduceMax - 12 vs 20
      • ReduceMax - 12 vs 18
      • ReduceMax - 12 vs 13
      • ReduceMax - 11 vs 20
      • ReduceMax - 11 vs 18
      • ReduceMax - 11 vs 13
      • ReduceMax - 11 vs 12
      • ReduceMax - 1 vs 20
      • ReduceMax - 1 vs 18
      • ReduceMax - 1 vs 13
      • ReduceMax - 1 vs 12
      • ReduceMax - 1 vs 11
    • 平均值归约 (ReduceMean)
      • ReduceMean - 13 vs 18
      • ReduceMean - 11 vs 18
      • ReduceMean - 11 vs 13
      • ReduceMean - 1 vs 18
      • ReduceMean - 1 vs 13
      • ReduceMean - 1 vs 11
    • 最小值归约 (ReduceMin)
      • ReduceMin - 18 vs 20
      • ReduceMin - 13 vs 20
      • ReduceMin - 13 vs 18
      • ReduceMin - 12 vs 20
      • ReduceMin - 12 vs 18
      • ReduceMin - 12 vs 13
      • ReduceMin - 11 vs 20
      • ReduceMin - 11 vs 18
      • ReduceMin - 11 vs 13
      • ReduceMin - 11 vs 12
      • ReduceMin - 1 vs 20
      • ReduceMin - 1 vs 18
      • ReduceMin - 1 vs 13
      • ReduceMin - 1 vs 12
      • ReduceMin - 1 vs 11
    • 积归约 (ReduceProd)
      • ReduceProd - 13 vs 18
      • ReduceProd - 11 vs 18
      • ReduceProd - 11 vs 13
      • ReduceProd - 1 vs 18
      • ReduceProd - 1 vs 13
      • ReduceProd - 1 vs 11
    • 和归约 (ReduceSum)
      • ReduceSum - 11 vs 13
      • ReduceSum - 1 vs 13
      • ReduceSum - 1 vs 11
    • 平方和归约 (ReduceSumSquare)
      • ReduceSumSquare - 13 vs 18
      • ReduceSumSquare - 11 vs 18
      • ReduceSumSquare - 11 vs 13
      • ReduceSumSquare - 1 vs 18
      • ReduceSumSquare - 1 vs 13
      • ReduceSumSquare - 1 vs 11
    • 正则表达式完全匹配 (RegexFullMatch)
    • ReLU
      • ReLU - 13 vs 14
      • ReLU - 6 vs 14
      • ReLU - 6 vs 13
      • ReLU - 1 vs 14
      • ReLU - 1 vs 13
      • ReLU - 1 vs 6
    • 重塑 (Reshape)
      • Reshape - 21 vs 23
      • Reshape - 19 vs 23
      • Reshape - 19 vs 21
      • Reshape - 14 vs 23
      • Reshape - 14 vs 21
      • Reshape - 14 vs 19
      • Reshape - 13 vs 23
      • Reshape - 13 vs 21
      • Reshape - 13 vs 19
      • Reshape - 13 vs 14
      • Reshape - 5 vs 23
      • Reshape - 5 vs 21
      • Reshape - 5 vs 19
      • Reshape - 5 vs 14
      • Reshape - 5 vs 13
      • Reshape - 1 vs 23
      • Reshape - 1 vs 21
      • Reshape - 1 vs 19
      • Reshape - 1 vs 14
      • Reshape - 1 vs 13
      • Reshape - 1 vs 5
    • 调整大小 (Resize)
      • Resize - 18 vs 19
      • Resize - 13 vs 19
      • Resize - 13 vs 18
      • Resize - 11 vs 19
      • Resize - 11 vs 18
      • Resize - 11 vs 13
      • Resize - 10 vs 19
      • Resize - 10 vs 18
      • Resize - 10 vs 13
      • Resize - 10 vs 11
    • 反转序列 (ReverseSequence)
    • RoIAlign
      • RoIAlign - 16 vs 22
      • RoIAlign - 10 vs 22
      • RoIAlign - 10 vs 16
    • 舍入 (Round)
      • Round - 11 vs 22
    • 短时傅里叶变换 (STFT)
    • 扫描 (Scan)
      • Scan - 21 vs 23
      • Scan - 19 vs 23
      • Scan - 19 vs 21
      • Scan - 16 vs 23
      • Scan - 16 vs 21
      • Scan - 16 vs 19
      • Scan - 11 vs 23
      • Scan - 11 vs 21
      • Scan - 11 vs 19
      • Scan - 11 vs 16
      • Scan - 9 vs 23
      • Scan - 9 vs 21
      • Scan - 9 vs 19
      • Scan - 9 vs 16
      • Scan - 9 vs 11
      • Scan - 8 vs 23
      • Scan - 8 vs 21
      • Scan - 8 vs 19
      • Scan - 8 vs 16
      • Scan - 8 vs 11
      • Scan - 8 vs 9
    • 散射 (Scatter)
      • Scatter - 9 vs 11
    • 散射元素 (ScatterElements)
      • ScatterElements - 16 vs 18
      • ScatterElements - 13 vs 18
      • ScatterElements - 13 vs 16
      • ScatterElements - 11 vs 18
      • ScatterElements - 11 vs 16
      • ScatterElements - 11 vs 13
    • N 维散射 (ScatterND)
      • ScatterND - 16 vs 18
      • ScatterND - 13 vs 18
      • ScatterND - 13 vs 16
      • ScatterND - 11 vs 18
      • ScatterND - 11 vs 16
      • ScatterND - 11 vs 13
    • SeLU
      • SeLU - 6 vs 22
      • SeLU - 1 vs 22
      • SeLU - 1 vs 6
    • 序列获取 (SequenceAt)
    • 序列构造 (SequenceConstruct)
    • 序列为空 (SequenceEmpty)
    • 序列删除 (SequenceErase)
    • 序列插入 (SequenceInsert)
    • 序列长度 (SequenceLength)
    • 序列映射 (SequenceMap)
    • 形状 (Shape)
      • Shape - 21 vs 23
      • Shape - 19 vs 23
      • Shape - 19 vs 21
      • Shape - 15 vs 23
      • Shape - 15 vs 21
      • Shape - 15 vs 19
      • Shape - 13 vs 23
      • Shape - 13 vs 21
      • Shape - 13 vs 19
      • Shape - 13 vs 15
      • Shape - 1 vs 23
      • Shape - 1 vs 21
      • Shape - 1 vs 19
      • Shape - 1 vs 15
      • Shape - 1 vs 13
    • 收缩 (Shrink)
    • Sigmoid
      • Sigmoid - 6 vs 13
      • Sigmoid - 1 vs 13
      • Sigmoid - 1 vs 6
    • 符号 (Sign)
      • Sign - 9 vs 13
    • 正弦 (Sin)
      • Sin - 7 vs 22
    • 双曲正弦 (Sinh)
      • Sinh - 9 vs 22
    • 大小 (Size)
      • Size - 21 vs 23
      • Size - 19 vs 23
      • Size - 19 vs 21
      • Size - 13 vs 23
      • Size - 13 vs 21
      • Size - 13 vs 19
      • Size - 1 vs 23
      • Size - 1 vs 21
      • Size - 1 vs 19
      • Size - 1 vs 13
    • 切片 (Slice)
      • Slice - 11 vs 13
      • Slice - 10 vs 13
      • Slice - 10 vs 11
      • Slice - 1 vs 13
      • Slice - 1 vs 11
      • Slice - 1 vs 10
    • Softmax
      • Softmax - 11 vs 13
      • Softmax - 1 vs 13
      • Softmax - 1 vs 11
    • Softmax 交叉熵损失 (SoftmaxCrossEntropyLoss)
      • SoftmaxCrossEntropyLoss - 12 vs 13
    • Softplus
      • Softplus - 1 vs 22
    • Softsign
      • Softsign - 1 vs 22
    • 空间到深度 (SpaceToDepth)
      • SpaceToDepth - 1 vs 13
    • 分割 (Split)
      • Split - 13 vs 18
      • Split - 11 vs 18
      • Split - 11 vs 13
      • Split - 2 vs 18
      • Split - 2 vs 13
      • Split - 2 vs 11
      • Split - 1 vs 18
      • Split - 1 vs 13
      • Split - 1 vs 11
      • Split - 1 vs 2
    • 分割到序列 (SplitToSequence)
    • 平方根 (Sqrt)
      • Sqrt - 6 vs 13
      • Sqrt - 1 vs 13
      • Sqrt - 1 vs 6
    • 压缩 (Squeeze)
      • Squeeze - 21 vs 23
      • Squeeze - 13 vs 23
      • Squeeze - 13 vs 21
      • Squeeze - 11 vs 23
      • Squeeze - 11 vs 21
      • Squeeze - 11 vs 13
      • Squeeze - 1 vs 23
      • Squeeze - 1 vs 21
      • Squeeze - 1 vs 13
      • Squeeze - 1 vs 11
    • 字符串连接 (StringConcat)
    • 字符串标准化 (StringNormalizer)
    • 字符串分割 (StringSplit)
    • 减法 (Sub)
      • Sub - 13 vs 14
      • Sub - 7 vs 14
      • Sub - 7 vs 13
      • Sub - 6 vs 14
      • Sub - 6 vs 13
      • Sub - 6 vs 7
      • Sub - 1 vs 14
      • Sub - 1 vs 13
      • Sub - 1 vs 7
      • Sub - 1 vs 6
    • 求和 (Sum)
      • Sum - 8 vs 13
      • Sum - 6 vs 13
      • Sum - 6 vs 8
      • Sum - 1 vs 13
      • Sum - 1 vs 8
      • Sum - 1 vs 6
    • 正切 (Tan)
      • Tan - 7 vs 22
    • 双曲正切 (Tanh)
      • Tanh - 6 vs 13
      • Tanh - 1 vs 13
      • Tanh - 1 vs 6
    • TF-IDF 向量化器 (TfIdfVectorizer)
    • 阈值 ReLU (ThresholdedRelu)
      • ThresholdedRelu - 10 vs 22
    • 平铺 (Tile)
      • Tile - 6 vs 13
      • Tile - 1 vs 13
      • Tile - 1 vs 6
    • TopK
      • TopK - 10 vs 11
      • TopK - 1 vs 11
      • TopK - 1 vs 10
    • 转置 (Transpose)
      • Transpose - 21 vs 23
      • Transpose - 13 vs 23
      • Transpose - 13 vs 21
      • Transpose - 1 vs 23
      • Transpose - 1 vs 21
      • Transpose - 1 vs 13
    • 三对角线 (Trilu)
    • 唯一值 (Unique)
    • unsqueeze
      • Unsqueeze - 21 vs 23
      • Unsqueeze - 13 vs 23
      • Unsqueeze - 13 vs 21
      • Unsqueeze - 11 vs 23
      • Unsqueeze - 11 vs 21
      • Unsqueeze - 11 vs 13
      • Unsqueeze - 1 vs 23

      • Unsqueeze - 1 对 21
      • Unsqueeze - 1 对 13
      • Unsqueeze - 1 对 11
    • 上采样
      • 上采样 - 9 对 10
      • 上采样 - 7 对 10
      • 上采样 - 7 对 9
      • 上采样 - 1 对 10
      • 上采样 - 1 对 9
      • 上采样 - 1 对 7
    • Where
      • Where - 9 对 16
    • Xor
      • Xor - 1 对 7
    • ai.onnx.ml - ArrayFeatureExtractor
    • ai.onnx.ml - Binarizer
    • ai.onnx.ml - CastMap
    • ai.onnx.ml - CategoryMapper
    • ai.onnx.ml - DictVectorizer
    • ai.onnx.ml - FeatureVectorizer
    • ai.onnx.ml - Imputer
    • ai.onnx.ml - LabelEncoder
      • LabelEncoder - 2 对 4
      • LabelEncoder - 1 对 4
      • LabelEncoder - 1 对 2
    • ai.onnx.ml - LinearClassifier
    • ai.onnx.ml - LinearRegressor
    • ai.onnx.ml - Normalizer
    • ai.onnx.ml - OneHotEncoder
    • ai.onnx.ml - SVMClassifier
    • ai.onnx.ml - SVMRegressor
    • ai.onnx.ml - Scaler
    • ai.onnx.ml - TreeEnsemble
    • ai.onnx.ml - TreeEnsembleClassifier
      • TreeEnsembleClassifier - 3 对 5
      • TreeEnsembleClassifier - 1 对 5
      • TreeEnsembleClassifier - 1 对 3
    • ai.onnx.ml - TreeEnsembleRegressor
      • TreeEnsembleRegressor - 3 对 5
      • TreeEnsembleRegressor - 1 对 5
      • TreeEnsembleRegressor - 1 对 3
    • ai.onnx.ml - ZipMap
    • ai.onnx.preview.training - Adagrad
    • ai.onnx.preview.training - Adam
    • ai.onnx.preview.training - Gradient
    • ai.onnx.preview.training - Momentum
  • 技术细节
    • 以8位存储浮点数
    • 4位整数类型
    • 以4位存储浮点数
  • ONNX 仓库文档
    • 向ONNX添加新的运算符或函数
    • ONNX中的广播
    • 关于ONNX运算符的可微性标签的简要指南
    • 维度表示
    • 外部数据
    • ONNX模型中心
    • 开放神经网络交换中间表示（ONNX IR）规范
    • 实现ONNX后端
    • 管理实验性运算符
    • 元数据
    • ONNX类型
    • ONNX后端测试
    • ONNX版本
    • 正式发布
    • 运算符约定
    • 概述
    • Python API概述
    • ONNX形状推断
    • ONNX文本语法
    • 类型表示
    • ONNX版本转换器
    • ONNX版本控制
返回顶部
查看此页面

技术细节¶

本节深入探讨实现细节、技术描述，比代码文档更深入。

  • 以8位存储浮点数
    • 论文
    • E4M3FN 和 E5M2
    • E4M3FNUZ 和 E5M2FNUZ
    • Cast
  • 4位整数类型
    • 论文
    • Cast
    • 打包和解包
  • 以4位存储浮点数
    • 论文
    • E2M1
    • Cast
    • 打包和解包
下一页
以8位存储浮点数
上一页
ai.onnx.preview.training - Momentum
版权所有 © 2024
使用 Sphinx 和 @pradyunsg 的 Furo 制作