Slice

Slice - 13

Version

• name: Slice (GitHub)

• domain: main

• since_version: 13

• function: False

• support_level: SupportType.COMMON

• shape inference: True

This version of the operator has been available since version 13.

Summary

Produces a slice of the input tensor along multiple axes. Similar to numpy: https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html Slices uses starts, ends, axes and steps inputs to specify the start and end dimension and step for each axis in the list of axes, it uses this information to slice the input data tensor. If a negative value is passed for any of the start or end indices, it represents number of elements before the end of that dimension. If the value passed to start or end is larger than the n (the number of elements in this dimension), it represents n. For slicing to the end of a dimension with unknown size, it is recommended to pass in INT_MAX when sclicing forward and ‘INT_MIN’ when slicing backward. If a negative value is passed for step, it represents slicing backward. However step value cannot be 0. If axes are omitted, they are set to [0, …, ndim-1]. If steps are omitted, they are set to [1, …, 1] of length len(starts) Example 1:

data = [

[1, 2, 3, 4], [5, 6, 7, 8],

] axes = [0, 1] starts = [1, 0] ends = [2, 3] steps = [1, 2] result = [

[5, 7],

]

Example 2:

data = [

[1, 2, 3, 4], [5, 6, 7, 8],

] starts = [0, 1] ends = [-1, 1000] result = [

[2, 3, 4],

]

Inputs

Between 3 and 5 inputs.

• data (heterogeneous) - T: Tensor of data to extract slices from.

• starts (heterogeneous) - Tind: 1-D tensor of starting indices of corresponding axis in axes

• ends (heterogeneous) - Tind: 1-D tensor of ending indices (exclusive) of corresponding axis in axes

• axes (optional, heterogeneous) - Tind: 1-D tensor of axes that starts and ends apply to. Negative value means counting dimensions from the back. Accepted range is [-r, r-1] where r = rank(data).

• steps (optional, heterogeneous) - Tind: 1-D tensor of slice step of corresponding axis in axes. Negative value means slicing backward. ‘steps’ cannot be 0. Defaults to 1.

Outputs

• output (heterogeneous) - T: Sliced data tensor.

Type Constraints

• T in ( tensor(bfloat16), tensor(bool), tensor(complex128), tensor(complex64), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8) ): Constrain input and output types to all tensor types.

• Tind in ( tensor(int32), tensor(int64) ): Constrain indices to integer types

Examples

slice

node = onnx.helper.make_node(
    'Slice',
    inputs=['x', 'starts', 'ends', 'axes', 'steps'],
    outputs=['y'],
)

x = np.random.randn(20, 10, 5).astype(np.float32)
y = x[0:3, 0:10]
starts = np.array([0, 0], dtype=np.int64)
ends = np.array([3, 10], dtype=np.int64)
axes = np.array([0, 1], dtype=np.int64)
steps = np.array([1, 1], dtype=np.int64)

expect(node, inputs=[x, starts, ends, axes, steps], outputs=[y],
       name='test_slice')

slice_neg

node = onnx.helper.make_node(
    'Slice',
    inputs=['x', 'starts', 'ends', 'axes', 'steps'],
    outputs=['y'],
)

x = np.random.randn(20, 10, 5).astype(np.float32)
starts = np.array([0], dtype=np.int64)
ends = np.array([-1], dtype=np.int64)
axes = np.array([1], dtype=np.int64)
steps = np.array([1], dtype=np.int64)
y = x[:, 0:-1]

expect(node, inputs=[x, starts, ends, axes, steps], outputs=[y],
       name='test_slice_neg')

slice_start_out_of_bounds

node = onnx.helper.make_node(
    'Slice',
    inputs=['x', 'starts', 'ends', 'axes', 'steps'],
    outputs=['y'],
)

x = np.random.randn(20, 10, 5).astype(np.float32)
starts = np.array([1000], dtype=np.int64)
ends = np.array([1000], dtype=np.int64)
axes = np.array([1], dtype=np.int64)
steps = np.array([1], dtype=np.int64)
y = x[:, 1000:1000]

expect(node, inputs=[x, starts, ends, axes, steps], outputs=[y],
       name='test_slice_start_out_of_bounds')

slice_end_out_of_bounds

node = onnx.helper.make_node(
    'Slice',
    inputs=['x', 'starts', 'ends', 'axes', 'steps'],
    outputs=['y'],
)

x = np.random.randn(20, 10, 5).astype(np.float32)
starts = np.array([1], dtype=np.int64)
ends = np.array([1000], dtype=np.int64)
axes = np.array([1], dtype=np.int64)
steps = np.array([1], dtype=np.int64)
y = x[:, 1:1000]

expect(node, inputs=[x, starts, ends, axes, steps], outputs=[y],
       name='test_slice_end_out_of_bounds')

slice_default_axes

node = onnx.helper.make_node(
    'Slice',
    inputs=['x', 'starts', 'ends'],
    outputs=['y'],
)

x = np.random.randn(20, 10, 5).astype(np.float32)
starts = np.array([0, 0, 3], dtype=np.int64)
ends = np.array([20, 10, 4], dtype=np.int64)
y = x[:, :, 3:4]

expect(node, inputs=[x, starts, ends], outputs=[y],
       name='test_slice_default_axes')

slice_default_steps

node = onnx.helper.make_node(
    'Slice',
    inputs=['x', 'starts', 'ends', 'axes'],
    outputs=['y'],
)

x = np.random.randn(20, 10, 5).astype(np.float32)
starts = np.array([0, 0, 3], dtype=np.int64)
ends = np.array([20, 10, 4], dtype=np.int64)
axes = np.array([0, 1, 2], dtype=np.int64)
y = x[:, :, 3:4]

expect(node, inputs=[x, starts, ends, axes], outputs=[y],
       name='test_slice_default_steps')

slice_neg_steps

node = onnx.helper.make_node(
    'Slice',
    inputs=['x', 'starts', 'ends', 'axes', 'steps'],
    outputs=['y'],
)

x = np.random.randn(20, 10, 5).astype(np.float32)
starts = np.array([20, 10, 4], dtype=np.int64)
ends = np.array([0, 0, 1], dtype=np.int64)
axes = np.array([0, 1, 2], dtype=np.int64)
steps = np.array([-1, -3, -2]).astype(np.int64)
y = x[20:0:-1, 10:0:-3, 4:1:-2]

expect(node, inputs=[x, starts, ends, axes, steps], outputs=[y],
       name='test_slice_neg_steps')

slice_negative_axes

node = onnx.helper.make_node(
    'Slice',
    inputs=['x', 'starts', 'ends', 'axes'],
    outputs=['y'],
)

x = np.random.randn(20, 10, 5).astype(np.float32)
starts = np.array([0, 0, 3], dtype=np.int64)
ends = np.array([20, 10, 4], dtype=np.int64)
axes = np.array([0, -2, -1], dtype=np.int64)
y = x[:, :, 3:4]

expect(node, inputs=[x, starts, ends, axes], outputs=[y],
       name='test_slice_negative_axes')

Differences

0	0	Produces a slice of the input tensor along multiple axes. Similar to numpy:	Produces a slice of the input tensor along multiple axes. Similar to numpy:
1	1	https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html	https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html
2	2	Slices uses starts, ends, axes and steps inputs to specify the start and end	Slices uses starts, ends, axes and steps inputs to specify the start and end
3	3	dimension and step for each axis in the list of axes, it uses this information to	dimension and step for each axis in the list of axes, it uses this information to
4	4	slice the input data tensor. If a negative value is passed for any of the	slice the input data tensor. If a negative value is passed for any of the
5	5	start or end indices, it represents number of elements before the end of that	start or end indices, it represents number of elements before the end of that
6	6	dimension. If the value passed to start or end is larger than the n (the	dimension. If the value passed to start or end is larger than the n (the
7	7	number of elements in this dimension), it represents n. For slicing to the	number of elements in this dimension), it represents n. For slicing to the
8	8	end of a dimension with unknown size, it is recommended to pass in INT_MAX	end of a dimension with unknown size, it is recommended to pass in INT_MAX
9	9	when sclicing forward and 'INT_MIN' when slicing backward.	when sclicing forward and 'INT_MIN' when slicing backward.
10	10	If a negative value is passed for step, it represents slicing backward.	If a negative value is passed for step, it represents slicing backward.
11	11	However step value cannot be 0.	However step value cannot be 0.
12	12	If axes are omitted, they are set to [0, ..., ndim-1].	If axes are omitted, they are set to [0, ..., ndim-1].
13	13	If steps are omitted, they are set to [1, ..., 1] of length len(starts)	If steps are omitted, they are set to [1, ..., 1] of length len(starts)
14	14	Example 1:	Example 1:
15	15	data = [	data = [
16	16	[1, 2, 3, 4],	[1, 2, 3, 4],
17	17	[5, 6, 7, 8],	[5, 6, 7, 8],
18	18	]	]
19	19	axes = [0, 1]	axes = [0, 1]
20	20	starts = [1, 0]	starts = [1, 0]
21	21	ends = [2, 3]	ends = [2, 3]
22	22	steps = [1, 2]	steps = [1, 2]
23	23	result = [	result = [
24	24	[5, 7],	[5, 7],
25	25	]	]
26	26	Example 2:	Example 2:
27	27	data = [	data = [
28	28	[1, 2, 3, 4],	[1, 2, 3, 4],
29	29	[5, 6, 7, 8],	[5, 6, 7, 8],
30	30	]	]
31	31	starts = [0, 1]	starts = [0, 1]
32	32	ends = [-1, 1000]	ends = [-1, 1000]
33	33	result = [	result = [
34	34	[2, 3, 4],	[2, 3, 4],
35	35	]	]
36	36
37	37	**Inputs**	**Inputs**
38	38
39	39	Between 3 and 5 inputs.	Between 3 and 5 inputs.
40	40
41	41	* **data** (heterogeneous) - **T**:	* **data** (heterogeneous) - **T**:
42	42	Tensor of data to extract slices from.	Tensor of data to extract slices from.
43	43	* **starts** (heterogeneous) - **Tind**:	* **starts** (heterogeneous) - **Tind**:
44	44	1-D tensor of starting indices of corresponding axis in axes	1-D tensor of starting indices of corresponding axis in axes
45	45	* **ends** (heterogeneous) - **Tind**:	* **ends** (heterogeneous) - **Tind**:
46	46	1-D tensor of ending indices (exclusive) of corresponding axis in	1-D tensor of ending indices (exclusive) of corresponding axis in
47	47	axes	axes
48	48	* **axes** (optional, heterogeneous) - **Tind**:	* **axes** (optional, heterogeneous) - **Tind**:
49	49	1-D tensor of axes that starts and ends apply to. Negative value	1-D tensor of axes that starts and ends apply to. Negative value
50	50	means counting dimensions from the back. Accepted range is [-r, r-1]	means counting dimensions from the back. Accepted range is [-r, r-1]
51	51	where r = rank(data).	where r = rank(data).
52	52	* **steps** (optional, heterogeneous) - **Tind**:	* **steps** (optional, heterogeneous) - **Tind**:
53	53	1-D tensor of slice step of corresponding axis in axes. Negative	1-D tensor of slice step of corresponding axis in axes. Negative
54	54	value means slicing backward. 'steps' cannot be 0. Defaults to 1.	value means slicing backward. 'steps' cannot be 0. Defaults to 1.
55	55
56	56	**Outputs**	**Outputs**
57	57
58	58	* **output** (heterogeneous) - **T**:	* **output** (heterogeneous) - **T**:
59	59	Sliced data tensor.	Sliced data tensor.
60	60
61	61	**Type Constraints**	**Type Constraints**
62	62
63	63	* **T** in (	* **T** in (
	64		tensor(bfloat16),
64	65	tensor(bool),	tensor(bool),
65	66	tensor(complex128),	tensor(complex128),
66	67	tensor(complex64),	tensor(complex64),
67	68	tensor(double),	tensor(double),
68	69	tensor(float),	tensor(float),
69	70	tensor(float16),	tensor(float16),
70	71	tensor(int16),	tensor(int16),
71	72	tensor(int32),	tensor(int32),
72	73	tensor(int64),	tensor(int64),
73	74	tensor(int8),	tensor(int8),
74	75	tensor(string),	tensor(string),
75	76	tensor(uint16),	tensor(uint16),
76	77	tensor(uint32),	tensor(uint32),
77	78	tensor(uint64),	tensor(uint64),
78	79	tensor(uint8)	tensor(uint8)
79	80	):	):
80	81	Constrain input and output types to all tensor types.	Constrain input and output types to all tensor types.
81	82	* **Tind** in (	* **Tind** in (
82	83	tensor(int32),	tensor(int32),
83	84	tensor(int64)	tensor(int64)
84	85	):	):
85	86	Constrain indices to integer types	Constrain indices to integer types

Slice - 11

Version

• name: Slice (GitHub)

• domain: main

• since_version: 11

• function: False

• support_level: SupportType.COMMON

• shape inference: True

This version of the operator has been available since version 11.

Summary

Produces a slice of the input tensor along multiple axes. Similar to numpy: https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html Slices uses starts, ends, axes and steps inputs to specify the start and end dimension and step for each axis in the list of axes, it uses this information to slice the input data tensor. If a negative value is passed for any of the start or end indices, it represents number of elements before the end of that dimension. If the value passed to start or end is larger than the n (the number of elements in this dimension), it represents n. For slicing to the end of a dimension with unknown size, it is recommended to pass in INT_MAX when sclicing forward and ‘INT_MIN’ when slicing backward. If a negative value is passed for step, it represents slicing backward. However step value cannot be 0. If axes are omitted, they are set to [0, …, ndim-1]. If steps are omitted, they are set to [1, …, 1] of length len(starts) Example 1:

data = [

[1, 2, 3, 4], [5, 6, 7, 8],

] axes = [0, 1] starts = [1, 0] ends = [2, 3] steps = [1, 2] result = [

[5, 7],

]

Example 2:

data = [

[1, 2, 3, 4], [5, 6, 7, 8],

] starts = [0, 1] ends = [-1, 1000] result = [

[2, 3, 4],

]

Inputs

Between 3 and 5 inputs.

• data (heterogeneous) - T: Tensor of data to extract slices from.

• starts (heterogeneous) - Tind: 1-D tensor of starting indices of corresponding axis in axes

• ends (heterogeneous) - Tind: 1-D tensor of ending indices (exclusive) of corresponding axis in axes

• axes (optional, heterogeneous) - Tind: 1-D tensor of axes that starts and ends apply to. Negative value means counting dimensions from the back. Accepted range is [-r, r-1] where r = rank(data).

• steps (optional, heterogeneous) - Tind: 1-D tensor of slice step of corresponding axis in axes. Negative value means slicing backward. ‘steps’ cannot be 0. Defaults to 1.

Outputs

• output (heterogeneous) - T: Sliced data tensor.

Type Constraints

• T in ( tensor(bool), tensor(complex128), tensor(complex64), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8) ): Constrain input and output types to all tensor types.

• Tind in ( tensor(int32), tensor(int64) ): Constrain indices to integer types

Differences

0	0	Produces a slice of the input tensor along multiple axes. Similar to numpy:	Produces a slice of the input tensor along multiple axes. Similar to numpy:
1	1	https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html	https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html
2	2	Slices uses starts, ends, axes and steps inputs to specify the start and end	Slices uses starts, ends, axes and steps inputs to specify the start and end
3	3	dimension and step for each axis in the list of axes, it uses this information to	dimension and step for each axis in the list of axes, it uses this information to
4	4	slice the input data tensor. If a negative value is passed for any of the	slice the input data tensor. If a negative value is passed for any of the
5	5	start or end indices, it represent number of elements before the end of that	start or end indices, it represents number of elements before the end of that
6	6	dimension. If the value passed to start or end is larger than the n (the	dimension. If the value passed to start or end is larger than the n (the
7	7	number of elements in this dimension), it represents n. For slicing to the	number of elements in this dimension), it represents n. For slicing to the
8	8	end of a dimension with unknown size, it is recommended to pass in INT_MAX.	end of a dimension with unknown size, it is recommended to pass in INT_MAX
	9		when sclicing forward and 'INT_MIN' when slicing backward.
9	10	If a negative value is passed for step, it represents slicing backward.	If a negative value is passed for step, it represents slicing backward.
	11		However step value cannot be 0.
10	12	If axes are omitted, they are set to [0, ..., ndim-1].	If axes are omitted, they are set to [0, ..., ndim-1].
11	13	If steps are omitted, they are set to [1, ..., 1] of length len(starts)	If steps are omitted, they are set to [1, ..., 1] of length len(starts)
12	14	Example 1:	Example 1:
13	15	data = [	data = [
14	16	[1, 2, 3, 4],	[1, 2, 3, 4],
15	17	[5, 6, 7, 8],	[5, 6, 7, 8],
16	18	]	]
17	19	axes = [0, 1]	axes = [0, 1]
18	20	starts = [1, 0]	starts = [1, 0]
19	21	ends = [2, 3]	ends = [2, 3]
20	22	steps = [1, 2]	steps = [1, 2]
21	23	result = [	result = [
22	24	[5, 7],	[5, 7],
23	25	]	]
24	26	Example 2:	Example 2:
25	27	data = [	data = [
26	28	[1, 2, 3, 4],	[1, 2, 3, 4],
27	29	[5, 6, 7, 8],	[5, 6, 7, 8],
28	30	]	]
29	31	starts = [0, 1]	starts = [0, 1]
30	32	ends = [-1, 1000]	ends = [-1, 1000]
31	33	result = [	result = [
32	34	[2, 3, 4],	[2, 3, 4],
33	35	]	]
34	36
35	37	**Inputs**	**Inputs**
36	38
37	39	Between 3 and 5 inputs.	Between 3 and 5 inputs.
38	40
39	41	* **data** (heterogeneous) - **T**:	* **data** (heterogeneous) - **T**:
40	42	Tensor of data to extract slices from.	Tensor of data to extract slices from.
41	43	* **starts** (heterogeneous) - **Tind**:	* **starts** (heterogeneous) - **Tind**:
42	44	1-D tensor of starting indices of corresponding axis in axes	1-D tensor of starting indices of corresponding axis in axes
43	45	* **ends** (heterogeneous) - **Tind**:	* **ends** (heterogeneous) - **Tind**:
44	46	1-D tensor of ending indices (exclusive) of corresponding axis in	1-D tensor of ending indices (exclusive) of corresponding axis in
45	47	axes	axes
46	48	* **axes** (optional, heterogeneous) - **Tind**:	* **axes** (optional, heterogeneous) - **Tind**:
47	49	1-D tensor of axes that starts and ends apply to.	1-D tensor of axes that starts and ends apply to. Negative value
	50		means counting dimensions from the back. Accepted range is [-r, r-1]
	51		where r = rank(data).
48	52	* **steps** (optional, heterogeneous) - **Tind**:	* **steps** (optional, heterogeneous) - **Tind**:
49	53	1-D tensor of slice step of corresponding axis in axes. Default to	1-D tensor of slice step of corresponding axis in axes. Negative
50		1.
	54		value means slicing backward. 'steps' cannot be 0. Defaults to 1.
51	55
52	56	**Outputs**	**Outputs**
53	57
54	58	* **output** (heterogeneous) - **T**:	* **output** (heterogeneous) - **T**:
55	59	Sliced data tensor.	Sliced data tensor.
56	60
57	61	**Type Constraints**	**Type Constraints**
58	62
59	63	* **T** in (	* **T** in (
60	64	tensor(bool),	tensor(bool),
61	65	tensor(complex128),	tensor(complex128),
62	66	tensor(complex64),	tensor(complex64),
63	67	tensor(double),	tensor(double),
64	68	tensor(float),	tensor(float),
65	69	tensor(float16),	tensor(float16),
66	70	tensor(int16),	tensor(int16),
67	71	tensor(int32),	tensor(int32),
68	72	tensor(int64),	tensor(int64),
69	73	tensor(int8),	tensor(int8),
70	74	tensor(string),	tensor(string),
71	75	tensor(uint16),	tensor(uint16),
72	76	tensor(uint32),	tensor(uint32),
73	77	tensor(uint64),	tensor(uint64),
74	78	tensor(uint8)	tensor(uint8)
75	79	):	):
76	80	Constrain input and output types to all tensor types.	Constrain input and output types to all tensor types.
77	81	* **Tind** in (	* **Tind** in (
78	82	tensor(int32),	tensor(int32),
79	83	tensor(int64)	tensor(int64)
80	84	):	):
81	85	Constrain indices to integer types	Constrain indices to integer types

Slice - 10

Version

• name: Slice (GitHub)

• domain: main

• since_version: 10

• function: False

• support_level: SupportType.COMMON

• shape inference: True

This version of the operator has been available since version 10.

Summary

Produces a slice of the input tensor along multiple axes. Similar to numpy: https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html Slices uses starts, ends, axes and steps inputs to specify the start and end dimension and step for each axis in the list of axes, it uses this information to slice the input data tensor. If a negative value is passed for any of the start or end indices, it represent number of elements before the end of that dimension. If the value passed to start or end is larger than the n (the number of elements in this dimension), it represents n. For slicing to the end of a dimension with unknown size, it is recommended to pass in INT_MAX. If a negative value is passed for step, it represents slicing backward. If axes are omitted, they are set to [0, …, ndim-1]. If steps are omitted, they are set to [1, …, 1] of length len(starts) Example 1:

data = [

[1, 2, 3, 4], [5, 6, 7, 8],

] axes = [0, 1] starts = [1, 0] ends = [2, 3] steps = [1, 2] result = [

[5, 7],

]

Example 2:

data = [

[1, 2, 3, 4], [5, 6, 7, 8],

] starts = [0, 1] ends = [-1, 1000] result = [

[2, 3, 4],

]

Inputs

Between 3 and 5 inputs.

• data (heterogeneous) - T: Tensor of data to extract slices from.

• starts (heterogeneous) - Tind: 1-D tensor of starting indices of corresponding axis in axes

• ends (heterogeneous) - Tind: 1-D tensor of ending indices (exclusive) of corresponding axis in axes

• axes (optional, heterogeneous) - Tind: 1-D tensor of axes that starts and ends apply to.

• steps (optional, heterogeneous) - Tind: 1-D tensor of slice step of corresponding axis in axes. Default to 1.

Outputs

• output (heterogeneous) - T: Sliced data tensor.

Type Constraints

• T in ( tensor(bool), tensor(complex128), tensor(complex64), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8) ): Constrain input and output types to all tensor types.

• Tind in ( tensor(int32), tensor(int64) ): Constrain indices to integer types

Differences

0	0	Produces a slice of the input tensor along multiple axes. Similar to numpy:	Produces a slice of the input tensor along multiple axes. Similar to numpy:
1	1	https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html	https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html
2	2	Slices uses axes, starts and ends attributes to specify the start and end	Slices uses starts, ends, axes and steps inputs to specify the start and end
3	3	dimension for each axis in the list of axes, it uses this information to	dimension and step for each axis in the list of axes, it uses this information to
4	4	slice the input data tensor. If a negative value is passed for any of the	slice the input data tensor. If a negative value is passed for any of the
5	5	start or end indices, it represent number of elements before the end of that	start or end indices, it represent number of elements before the end of that
6	6	dimension. If the value passed to start or end is larger than the n (the	dimension. If the value passed to start or end is larger than the n (the
7	7	number of elements in this dimension), it represents n. For slicing to the	number of elements in this dimension), it represents n. For slicing to the
8	8	end of a dimension with unknown size, it is recommended to pass in INT_MAX.	end of a dimension with unknown size, it is recommended to pass in INT_MAX.
	9		If a negative value is passed for step, it represents slicing backward.
9	10	If axes are omitted, they are set to [0, ..., ndim-1].	If axes are omitted, they are set to [0, ..., ndim-1].
	11		If steps are omitted, they are set to [1, ..., 1] of length len(starts)
10	12	Example 1:	Example 1:
11	13	data = [	data = [
12	14	[1, 2, 3, 4],	[1, 2, 3, 4],
13	15	[5, 6, 7, 8],	[5, 6, 7, 8],
14	16	]	]
15	17	axes = [0, 1]	axes = [0, 1]
16	18	starts = [1, 0]	starts = [1, 0]
17	19	ends = [2, 3]	ends = [2, 3]
	20		steps = [1, 2]
18	21	result = [	result = [
19	22	[5, 6, 7],	[5, 7],
20	23	]	]
21	24	Example 2:	Example 2:
22	25	data = [	data = [
23	26	[1, 2, 3, 4],	[1, 2, 3, 4],
24	27	[5, 6, 7, 8],	[5, 6, 7, 8],
25	28	]	]
26	29	starts = [0, 1]	starts = [0, 1]
27	30	ends = [-1, 1000]	ends = [-1, 1000]
28	31	result = [	result = [
29	32	[2, 3, 4],	[2, 3, 4],
30	33	]	]
31	34
32		**Attributes**
33
34		* **axes**:
35	35	Axes that starts and ends apply to. It's optional. If not	**Inputs**
	36
36	37	present, will be treated as [0, 1, ..., len(starts) - 1].	Between 3 and 5 inputs.
	38
	39		* **data** (heterogeneous) - **T**:
	40		Tensor of data to extract slices from.
	41		* **starts** (heterogeneous) - **Tind**:
37	42	* **ends** (required):	1-D tensor of starting indices of corresponding axis in axes
	43		* **ends** (heterogeneous) - **Tind**:
38	44	Ending indices (exclusive) of corresponding axis in axes	1-D tensor of ending indices (exclusive) of corresponding axis in
39		* **starts** (required):
40		Starting indices of corresponding axis in axes
41
	45		axes
42	46	**Inputs**	* **axes** (optional, heterogeneous) - **Tind**:
43
	47		1-D tensor of axes that starts and ends apply to.
44	48	* **data** (heterogeneous) - **T**:	* **steps** (optional, heterogeneous) - **Tind**:
45		Tensor of data to extract slices from.
	49		1-D tensor of slice step of corresponding axis in axes. Default to
	50		1.
46	51
47	52	**Outputs**	**Outputs**
48	53
49	54	* **output** (heterogeneous) - **T**:	* **output** (heterogeneous) - **T**:
50	55	Sliced data tensor.	Sliced data tensor.
51	56
52	57	**Type Constraints**	**Type Constraints**
53	58
54	59	* **T** in (	* **T** in (
55	60	tensor(bool),	tensor(bool),
56	61	tensor(complex128),	tensor(complex128),
57	62	tensor(complex64),	tensor(complex64),
58	63	tensor(double),	tensor(double),
59	64	tensor(float),	tensor(float),
60	65	tensor(float16),	tensor(float16),
61	66	tensor(int16),	tensor(int16),
62	67	tensor(int32),	tensor(int32),
63	68	tensor(int64),	tensor(int64),
64	69	tensor(int8),	tensor(int8),
65	70	tensor(string),	tensor(string),
66	71	tensor(uint16),	tensor(uint16),
67	72	tensor(uint32),	tensor(uint32),
68	73	tensor(uint64),	tensor(uint64),
69	74	tensor(uint8)	tensor(uint8)
70	75	):	):
71	76	Constrain input and output types to all tensor types.	Constrain input and output types to all tensor types.
	77		* **Tind** in (
	78		tensor(int32),
	79		tensor(int64)
	80		):
	81		Constrain indices to integer types

Slice - 1

Version

• name: Slice (GitHub)

• domain: main

• since_version: 1

• function: False

• support_level: SupportType.COMMON

• shape inference: True

This version of the operator has been available since version 1.

Summary

Produces a slice of the input tensor along multiple axes. Similar to numpy: https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html Slices uses axes, starts and ends attributes to specify the start and end dimension for each axis in the list of axes, it uses this information to slice the input data tensor. If a negative value is passed for any of the start or end indices, it represent number of elements before the end of that dimension. If the value passed to start or end is larger than the n (the number of elements in this dimension), it represents n. For slicing to the end of a dimension with unknown size, it is recommended to pass in INT_MAX. If axes are omitted, they are set to [0, …, ndim-1]. Example 1:

data = [

[1, 2, 3, 4], [5, 6, 7, 8],

] axes = [0, 1] starts = [1, 0] ends = [2, 3] result = [

[5, 6, 7],

]

Example 2:

data = [

[1, 2, 3, 4], [5, 6, 7, 8],

] starts = [0, 1] ends = [-1, 1000] result = [

[2, 3, 4],

]

Attributes

• axes: Axes that starts and ends apply to. It’s optional. If not present, will be treated as [0, 1, …, len(starts) - 1].

• ends (required): Ending indices (exclusive) of corresponding axis in axes`

• starts (required): Starting indices of corresponding axis in axes

Inputs

• data (heterogeneous) - T: Tensor of data to extract slices from.

Outputs

• output (heterogeneous) - T: Sliced data tensor.

Type Constraints

• T in ( tensor(bool), tensor(complex128), tensor(complex64), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8) ): Constrain input and output types to all tensor types.

previous

Size

next

Softmax