Mean - version 8#

This page documents version 8 of operator Mean. See Mean for the latest version (since version 13).

  • Domain: ai.onnx

  • Since version: 8

Element-wise mean of each of the input tensors (with Numpy-style broadcasting support). All inputs and outputs must have the same data type.

This operator supports multidirectional (i.e., Numpy-style) broadcasting; for more details please check the doc.

Inputs

  • data_0 (T): List of tensors for mean.

Outputs

  • mean (T): Output tensor.

Type Constraints

  • T: Constrain input and output types to float tensors. Allowed types: tensor(double), tensor(float), tensor(float16).

Examples#

test_cc_mean_column_row_scalar_double_opset_8

Node:
  Mean(data_0, data_1, data_2) -> (result)

Inputs:
  data_0: shape=(2, 1), dtype=float64
    [[0.],
     [3.]]
  data_1: shape=(3,), dtype=float64
    [3., 6., 9.]
  data_2: shape=(), dtype=float64
    3.

Outputs:
  result: shape=(2, 3), dtype=float64
    [[2., 3., 4.],
     [3., 4., 5.]]

test_cc_mean_column_row_scalar_float_opset_8

Node:
  Mean(data_0, data_1, data_2) -> (result)

Inputs:
  data_0: shape=(2, 1), dtype=float32
    [[0.],
     [3.]]
  data_1: shape=(3,), dtype=float32
    [3., 6., 9.]
  data_2: shape=(), dtype=float32
    3.

Outputs:
  result: shape=(2, 3), dtype=float32
    [[2., 3., 4.],
     [3., 4., 5.]]

test_cc_mean_crossed_dimensions_double_opset_8

Node:
  Mean(data_0, data_1) -> (result)

Inputs:
  data_0: shape=(2, 1), dtype=float64
    [[2.],
     [4.]]
  data_1: shape=(1, 3), dtype=float64
    [[2., 4., 6.]]

Outputs:
  result: shape=(2, 3), dtype=float64
    [[2., 3., 4.],
     [3., 4., 5.]]

test_cc_mean_crossed_dimensions_float_opset_8

Node:
  Mean(data_0, data_1) -> (result)

Inputs:
  data_0: shape=(2, 1), dtype=float32
    [[2.],
     [4.]]
  data_1: shape=(1, 3), dtype=float32
    [[2., 4., 6.]]

Outputs:
  result: shape=(2, 3), dtype=float32
    [[2., 3., 4.],
     [3., 4., 5.]]

test_cc_mean_scalar_first_double_opset_8

Node:
  Mean(data_0, data_1) -> (result)

Inputs:
  data_0: shape=(), dtype=float64
    2.
  data_1: shape=(2, 2), dtype=float64
    [[2., 4.],
     [6., 8.]]

Outputs:
  result: shape=(2, 2), dtype=float64
    [[2., 3.],
     [4., 5.]]

test_cc_mean_scalar_first_float_opset_8

Node:
  Mean(data_0, data_1) -> (result)

Inputs:
  data_0: shape=(), dtype=float32
    2.
  data_1: shape=(2, 2), dtype=float32
    [[2., 4.],
     [6., 8.]]

Outputs:
  result: shape=(2, 2), dtype=float32
    [[2., 3.],
     [4., 5.]]

test_cc_mean_scalar_last_double_opset_8

Node:
  Mean(data_0, data_1) -> (result)

Inputs:
  data_0: shape=(2, 2), dtype=float64
    [[2., 4.],
     [6., 8.]]
  data_1: shape=(), dtype=float64
    2.

Outputs:
  result: shape=(2, 2), dtype=float64
    [[2., 3.],
     [4., 5.]]

test_cc_mean_scalar_last_float_opset_8

Node:
  Mean(data_0, data_1) -> (result)

Inputs:
  data_0: shape=(2, 2), dtype=float32
    [[2., 4.],
     [6., 8.]]
  data_1: shape=(), dtype=float32
    2.

Outputs:
  result: shape=(2, 2), dtype=float32
    [[2., 3.],
     [4., 5.]]

test_cc_mean_scalar_row_column_double_opset_8

Node:
  Mean(data_0, data_1, data_2) -> (result)

Inputs:
  data_0: shape=(), dtype=float64
    3.
  data_1: shape=(3,), dtype=float64
    [3., 6., 9.]
  data_2: shape=(2, 1), dtype=float64
    [[0.],
     [3.]]

Outputs:
  result: shape=(2, 3), dtype=float64
    [[2., 3., 4.],
     [3., 4., 5.]]

test_cc_mean_scalar_row_column_float_opset_8

Node:
  Mean(data_0, data_1, data_2) -> (result)

Inputs:
  data_0: shape=(), dtype=float32
    3.
  data_1: shape=(3,), dtype=float32
    [3., 6., 9.]
  data_2: shape=(2, 1), dtype=float32
    [[0.],
     [3.]]

Outputs:
  result: shape=(2, 3), dtype=float32
    [[2., 3., 4.],
     [3., 4., 5.]]

test_cc_mean_successive_rank_expansion_double_opset_8

Node:
  Mean(data_0, data_1, data_2, data_3) -> (result)

Inputs:
  data_0: shape=(), dtype=float64
    4.
  data_1: shape=(3,), dtype=float64
    [0., 4., 8.]
  data_2: shape=(2, 1), dtype=float64
    [[0.],
     [4.]]
  data_3: shape=(2, 1, 1), dtype=float64
    [[[0.]],

     [[8.]]]

Outputs:
  result: shape=(2, 2, 3), dtype=float64
    [[[1., 2., 3.],
      [2., 3., 4.]],

     [[3., 4., 5.],
      [4., 5., 6.]]]

test_cc_mean_successive_rank_expansion_float_opset_8

Node:
  Mean(data_0, data_1, data_2, data_3) -> (result)

Inputs:
  data_0: shape=(), dtype=float32
    4.
  data_1: shape=(3,), dtype=float32
    [0., 4., 8.]
  data_2: shape=(2, 1), dtype=float32
    [[0.],
     [4.]]
  data_3: shape=(2, 1, 1), dtype=float32
    [[[0.]],

     [[8.]]]

Outputs:
  result: shape=(2, 2, 3), dtype=float32
    [[[1., 2., 3.],
      [2., 3., 4.]],

     [[3., 4., 5.],
      [4., 5., 6.]]]

test_cc_mean_two_scalars_column_double_opset_8

Node:
  Mean(data_0, data_1, data_2) -> (result)

Inputs:
  data_0: shape=(), dtype=float64
    3.
  data_1: shape=(), dtype=float64
    6.
  data_2: shape=(2, 1), dtype=float64
    [[0.],
     [3.]]

Outputs:
  result: shape=(2, 1), dtype=float64
    [[3.],
     [4.]]

test_cc_mean_two_scalars_column_float_opset_8

Node:
  Mean(data_0, data_1, data_2) -> (result)

Inputs:
  data_0: shape=(), dtype=float32
    3.
  data_1: shape=(), dtype=float32
    6.
  data_2: shape=(2, 1), dtype=float32
    [[0.],
     [3.]]

Outputs:
  result: shape=(2, 1), dtype=float32
    [[3.],
     [4.]]

Differences with previous version (6)#

SchemaDiff: Mean (domain 'ai.onnx')

  • old version: 6

  • new version: 8

  • breaking: no

Documentation:

  • line similarity: 0.00 (+4/-2 lines)

--- Mean v6
+++ Mean v8
@@ -1,2 +1,4 @@
-Element-wise mean of each of the input tensors. All inputs and outputs must
-have the same shape and data type.
+Element-wise mean of each of the input tensors (with Numpy-style broadcasting support).
+All inputs and outputs must have the same data type.
+
+This operator supports **multidirectional (i.e., Numpy-style) broadcasting**; for more details please check [the doc](Broadcasting.md).