Skip to content

Commit

Permalink
Support dynamic 'pads' and 'value' in Pad operator (onnx#2031)
Browse files Browse the repository at this point in the history 
* Pad spec update

* More changes

* More changes

* Make pads and value attributes inputs in Pad operator

* More changes

* More changes

* More cahnges

* More changes

* Remove tab

* Fix formatting issues

* More changes

* Pad spec update

* More changes

* More changes

* Prevent unused variables from generating warnings across all platforms.  (onnx#1930)

* Change the return type for the zipmap operator to match the description in the spec.

* Prevent unused variables from generating warnings across all platforms.
This was observed in onnxruntime when __ONNX_NO_DOC_STRINGS was enabled.

* nit change

* Shape Inference Tests for QOps (onnx#1929)

* fix shape inference and add tests for shape inference

* cosmetic fixes

* plus some formatting

* Fix shape inference for matmul (onnx#1941)

* Fix shape inference for matmul when there is no input shape, and provide better coverage for testcases

* add test for qlinearmatmul as weel

* Make pads and value attributes inputs in Pad operator

* More changes

* More changes

* More cahnges

* More changes

* Remove tab

* Fix formatting issues

* More changes

* Fix line ending issue

* Update Operators md file

* Fix typo

* Updating TestCoverage.md file

* Nit fix

* Fix accidental revert

* PR feedback

* More PR feedback changes

* Optimizer passes changes

* Remove unnecessary line in def file

* More formatting changes

* Add 2 files missed previously

* Commit missed out file

* Fix build break

* Update tests

* Initial commit

* Formatting

* Refactor ParseRawData

* More refactroing changes

* Introduce shape inference failure in case of wrong number of pad values

* Resolve comments

* Avoid vector copy

* Fix comment

* More changes

* Update Changelog.md

* Def changes

* More updates

* Formatting

* Check-in tests

* Update docs

* Update Changelog.md

* Docs update

* Update Changelog.md

* Update Changelog.md

* Resolve comments

* Add docs

* Addign docs

* Update Changelog.md

* Build break fix

* Fix build break

* Add optimizer tests for older opset Pad

* Add back older opset Pad shape inference tests

* Fix build break

* Fix build break

* Account for Value to Constant_value change in some comments
  • Loading branch information
@hariharans29 @wschin
hariharans29 authored and wschin committed Sep 17, 2019
1 parent 8ac885a commit 7ef341b
Show file tree
Hide file tree
Showing 27 changed files with 1,125 additions and 293 deletions.
125 changes: 120 additions & 5 deletions docs/Changelog.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2723,17 +2723,14 @@ This version of the operator has been available since version 1 of the default O
### <a name="Pad-1"></a>**Pad-1**</a>

Given `data` tensor, paddings, mode, and value.

Example:
Insert 0 paddings to the beginning of the second dimension.

data = [
[1.0, 1.2],
[2.3, 3.4],
[4.5, 5.7],
]
paddings = [0, 0, 2, 0]

output = [
[
[0.0, 0.0, 1.0, 1.2],
Expand Down Expand Up @@ -8940,7 +8937,6 @@ This version of the operator has been available since version 9 of the default O
for dimension = axis, and index in source for dimension != axis. For instance, in a 2-D tensor case,
data[indices[i][j]][j] = updates[i][j] if axis = 0, or data[i][indices[i][j]] = updates[i][j] if axis = 1,
where i and j are loop counters from 0 up to the respective size in `updates` - 1.

Example 1:
data = [
[0.0, 0.0, 0.0],
Expand All @@ -8960,7 +8956,6 @@ This version of the operator has been available since version 9 of the default O
[1.0, 0.0, 2.2]
[0.0, 2.1, 1.2]
]

Example 2:
data = [[1.0, 2.0, 3.0, 4.0, 5.0]]
indices = [[1, 3]]
Expand Down Expand Up @@ -11504,6 +11499,126 @@ This version of the operator has been available since version 11 of the default
<dd>Constrain to any tensor type.</dd>
</dl>

### <a name="Pad-11"></a>**Pad-11**</a>

Given a tensor containing the data to be padded (`data`), a tensor containing the number of start and end pad values for axis (`pads`), (optionally) a `mode`, and (optionally) `constant_value`,
a padded tensor (`output`) is generated.

The three supported `modes` are (similar to corresponding modes supported by `numpy.pad`):

1) `constant`(default) - pads with a given constant value as specified by `constant_value` (which defaults to 0)

2) `reflect` - pads with the reflection of the vector mirrored on the first and last values of the vector along each axis

3) `edge` - pads with the edge values of array


Example 1 (`constant` mode):
Insert 0 pads to the beginning of the second dimension.

data =
[
[1.0, 1.2],
[2.3, 3.4],
[4.5, 5.7],
]

pads = [0, 2, 0, 0]

mode = 'constant'

constant_value = 0.0

output =
[
[
[0.0, 0.0, 1.0, 1.2],
[0.0, 0.0, 2.3, 3.4],
[0.0, 0.0, 4.5, 5.7],
],
]


Example 2 (`reflect` mode):
data =
[
[1.0, 1.2],
[2.3, 3.4],
[4.5, 5.7],
]

pads = [0, 2, 0, 0]

mode = 'reflect'

output =
[
[
[1.0, 1.2, 1.0, 1.2],
[2.3, 3.4, 2.3, 3.4],
[4.5, 5.7, 4.5, 5.7],
],
]


Example 3 (`edge` mode):
data =
[
[1.0, 1.2],
[2.3, 3.4],
[4.5, 5.7],
]

pads = [0, 2, 0, 0]

mode = 'edge'

output =
[
[
[1.0, 1.0, 1.0, 1.2],
[2.3, 2.3, 2.3, 3.4],
[4.5, 4.5, 4.5, 5.7],
],
]


#### Version

This version of the operator has been available since version 11 of the default ONNX operator set.

#### Attributes

<dl>
<dt><tt>mode</tt> : string (default is constant)</dt>
<dd>Supported modes: `constant`(default), `reflect`, `edge`</dd>
</dl>

#### Inputs (2 - 3)

<dl>
<dt><tt>data</tt> : T</dt>
<dd>Input tensor.</dd>
<dt><tt>pads</tt> : tensor(int64)</dt>
<dd>Tensor of integers indicating the number of padding elements to add or remove (if negative) at the beginning and end of each axis. For 2D input tensor, it is the number of pixels. `pads` should be a 1D tensor of shape [2 * input_rank]. `pads` format should be: [x1_begin, x2_begin,...,x1_end, x2_end,...], where xi_begin is the number of pad values added at the beginning of axis `i` and xi_end, the number of pad values added at the end of axis `i`.</dd>
<dt><tt>constant_value</tt> (optional) : T</dt>
<dd>(Optional) A scalar value to be used if the mode chosen is `constant` (by default it is 0).</dd>
</dl>

#### Outputs

<dl>
<dt><tt>output</tt> : T</dt>
<dd>Tensor after padding.</dd>
</dl>

#### Type Constraints

<dl>
<dt><tt>T</tt> : tensor(uint8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(int8), tensor(int16), tensor(int32), tensor(int64), tensor(float16), tensor(float), tensor(double)</dt>
<dd>Constrains input and output to only numeric types.</dd>
</dl>

### <a name="Range-11"></a>**Range-11**</a>

Generate a tensor containing a sequence of numbers that begin at `start` and extends by increments of `delta`
Expand Down
129 changes: 97 additions & 32 deletions docs/Operators.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9986,45 +9986,110 @@ expect(node, inputs=[x, slope], outputs=[y],

### <a name="Pad"></a><a name="pad">**Pad**</a>

Given `data` tensor, pads, mode, and value.
Example:
Given a tensor containing the data to be padded (`data`), a tensor containing the number of start and end pad values for axis (`pads`), (optionally) a `mode`, and (optionally) `constant_value`,
a padded tensor (`output`) is generated.

The three supported `modes` are (similar to corresponding modes supported by `numpy.pad`):

1) `constant`(default) - pads with a given constant value as specified by `constant_value` (which defaults to 0)

2) `reflect` - pads with the reflection of the vector mirrored on the first and last values of the vector along each axis

3) `edge` - pads with the edge values of array


Example 1 (`constant` mode):
Insert 0 pads to the beginning of the second dimension.
data = [

data =
[
[1.0, 1.2],
[2.3, 3.4],
[4.5, 5.7],
]
]

pads = [0, 2, 0, 0]
output = [

mode = 'constant'

constant_value = 0.0

output =
[
[
[0.0, 0.0, 1.0, 1.2],
[0.0, 0.0, 2.3, 3.4],
[0.0, 0.0, 4.5, 5.7],
],
]


Example 2 (`reflect` mode):
data =
[
[1.0, 1.2],
[2.3, 3.4],
[4.5, 5.7],
]

pads = [0, 2, 0, 0]

mode = 'reflect'

output =
[
[
[1.0, 1.2, 1.0, 1.2],
[2.3, 3.4, 2.3, 3.4],
[4.5, 5.7, 4.5, 5.7],
],
]


Example 3 (`edge` mode):
data =
[
[1.0, 1.2],
[2.3, 3.4],
[4.5, 5.7],
]

pads = [0, 2, 0, 0]

mode = 'edge'

output =
[
[
[1.0, 1.0, 1.0, 1.2],
[2.3, 2.3, 2.3, 3.4],
[4.5, 4.5, 4.5, 5.7],
],
]


#### Version

This version of the operator has been available since version 2 of the default ONNX operator set.
This version of the operator has been available since version 11 of the default ONNX operator set.

Other versions of this operator: <a href="Changelog.md#Pad-1">Pad-1</a>
Other versions of this operator: <a href="Changelog.md#Pad-1">Pad-1</a>, <a href="Changelog.md#Pad-2">Pad-2</a>

#### Attributes

<dl>
<dt><tt>mode</tt> : string (default is constant)</dt>
<dd>Three modes: constant(default), reflect, edge</dd>
<dt><tt>pads</tt> : list of ints (required)</dt>
<dd>List of integers indicating the number of padding elements to add or remove (if negative) at the beginning and end of each axis. For 2D it is the number of pixels. `pads` rank should be double of the input's rank. `pads` format should be as follow [x1_begin, x2_begin...x1_end, x2_end,...], where xi_begin the number of pixels added at the beginning of axis `i` and xi_end, the number of pixels added at the end of axis `i`.</dd>
<dt><tt>value</tt> : float (default is 0.0)</dt>
<dd>One float, indicates the value to be filled.</dd>
<dd>Supported modes: `constant`(default), `reflect`, `edge`</dd>
</dl>

#### Inputs
#### Inputs (2 - 3)

<dl>
<dt><tt>data</tt> : T</dt>
<dd>Input tensor.</dd>
<dt><tt>pads</tt> : tensor(int64)</dt>
<dd>Tensor of integers indicating the number of padding elements to add or remove (if negative) at the beginning and end of each axis. For 2D input tensor, it is the number of pixels. `pads` should be a 1D tensor of shape [2 * input_rank]. `pads` format should be: [x1_begin, x2_begin,...,x1_end, x2_end,...], where xi_begin is the number of pad values added at the beginning of axis `i` and xi_end, the number of pad values added at the end of axis `i`.</dd>
<dt><tt>constant_value</tt> (optional) : T</dt>
<dd>(Optional) A scalar value to be used if the mode chosen is `constant` (by default it is 0).</dd>
</dl>

#### Outputs
Expand All @@ -10037,8 +10102,8 @@ Other versions of this operator: <a href="Changelog.md#Pad-1">Pad-1</a>
#### Type Constraints

<dl>
<dt><tt>T</tt> : tensor(float16), tensor(float), tensor(double)</dt>
<dd>Constrain input and output types to float tensors.</dd>
<dt><tt>T</tt> : tensor(uint8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(int8), tensor(int16), tensor(int32), tensor(int64), tensor(float16), tensor(float), tensor(double)</dt>
<dd>Constrains input and output to only numeric types.</dd>
</dl>


Expand All @@ -10050,21 +10115,21 @@ Other versions of this operator: <a href="Changelog.md#Pad-1">Pad-1</a>
```python
node = onnx.helper.make_node(
'Pad',
inputs=['x'],
inputs=['x', 'pads', 'value'],
outputs=['y'],
mode='constant',
value=1.2,
pads=[0, 0, 1, 3, 0, 0, 2, 4],
mode='constant'
)
x = np.random.randn(1, 3, 4, 5).astype(np.float32)
y = np.pad(
pads = np.array([0, 0, 1, 3, 0, 0, 2, 4]).astype(np.int64) # pad order [x1_begin, x2_begin, ..., x1_end, x2_end, ...]
value = np.float32(1.2)
y = pad_impl(
x,
pad_width=((0, 0), (0, 0), (1, 2), (3, 4)),
mode='constant',
constant_values=1.2,
pads,
'constant',
1.2
)

expect(node, inputs=[x], outputs=[y],
expect(node, inputs=[x, pads, value], outputs=[y],
name='test_constant_pad')
```

Expand All @@ -10078,19 +10143,19 @@ expect(node, inputs=[x], outputs=[y],
for mode in ['edge', 'reflect']:
node = onnx.helper.make_node(
'Pad',
inputs=['x'],
inputs=['x', 'pads'],
outputs=['y'],
mode=mode,
pads=[0, 0, 1, 1, 0, 0, 1, 1]
mode=mode
)
x = np.random.randn(1, 3, 4, 5).astype(np.float32)
y = np.pad(
x = np.random.randn(1, 3, 4, 5).astype(np.int32)
pads = np.array([0, 0, 1, 1, 0, 0, 1, 1]).astype(np.int64) # pad order [x1_begin, x2_begin, ..., x1_end, x2_end, ...]
y = pad_impl(
x,
pad_width=((0, 0), (0, 0), (1, 1), (1, 1)),
mode=mode,
pads,
mode
)

expect(node, inputs=[x], outputs=[y],
expect(node, inputs=[x, pads], outputs=[y],
name='test_{}_pad'.format(mode))
```

Expand Down
Loading

0 comments on commit 7ef341b

Please sign in to comment.