Support ScatterND operator in ONNX

Add GatherND

* Update to include ONNX Foundation WG

Added link to Gitter and description of the newly formed Foundation WG. Co-leaders Jim Spohrer (IBM) and Ryan Loney (Intel)

* Updated description of Foundation WG

Revised the description of the working group for ONNX Foundation

* Update working-groups.md

…he Range op (#2287)

with type default values even though they are not in the stream.

Move map and sequence types to onnx domain, this is the first step of merging onnx-ml and onnx types.

* Fix link to community docs in readme

Addresses #2255

* Update README.md

* Update README.md

* Update managingexperimentalops.md

* Update managingexperimentalops.md

* Rename managingexperimentalops.md to ManagingExperimentalOps.md

* Added negative axes for slice and squeeze opset 11

* added negative axes support for squeeze, unsqueeze, flatten

* added support for negative axes to all the existing ops

* fixed minor if condition missed for axis attr in flatten

* fixed test name for flatten with negative axes

* updated unsqueeze and softmax tests with fix for failures

* fixed typo

* Updating Split op documentations and version

* fixed typo in unsqueeze model

* fixed dim check for unsqueeze

* fixed type cast

* test fix for build failure

* updating onnx model for unsqueeze test

* fixed minor error in type casting

* added test for int64 input to 'where' op

* added onnx model files and docs for test 'where' op with long input

* added missing doc updates

… that do not have matching graph inputs. (#2135)

* Update helper.py

An IR v4 model is not required to have matching graph inputs for all initializers. Update printable_graph to allow for this and output the name, type and shape of initializers with no matching graph input.

* Add test for printable_graph

Add test and tweak messaging

* Fix comment formatting

…lements', 'OneHot' (#2260)

* modified gather docs to support negative indices

* added support for negatice indices to gather_elements and scatter_elements

* fixed documentation formatting as per comments

* Added negatice indices to docs for OneHot op

* GatheND spec for negative indices editted

* fix for comments

* fixed formatting for gather op

* Update onnx/defs/tensor/defs.cc

Co-Authored-By: Wei-Sheng Chin <wschin@outlook.com>

* Update docs/Changelog.md

Co-Authored-By: Wei-Sheng Chin <wschin@outlook.com>

* added print examples to onehot and gather as per comments

* adding modifies model tests for onehot

* updated doc files

* updating unsqueeze test

* typo fix as per comments

* Fix shapeinference function

* Added shapeinference test for cumsum

* update inference test

* fix test

* minor fix -- shape (1) should be (1,)

* Add whitespace after comma to fix flake warning

* Add a helper function update_inputs_outputs_dims to tools

* fix link to doc

* newline at the end

* add test for tools

* doc props

* nit

* ci tests

* ci tests 2

* accept shapes by dictionary inputs and add more error handling

* Update onnx/tools/update_model_dims.py

nit: rephrasing

Co-Authored-By: Wei-Sheng Chin <wschin@outlook.com>

* remove debug line

* fix type annotation

* fix annotation

* fix annotation

* fix annotation

* fix flake8

* sequence related ops

* refine docs

* extend hasInputShape to Sequence

* refining naming and error checking

* refine descriptions

* fix resize shape inference issue in opset10

* include opset10 upsample as well

* nit: const auto*

* rename 'opset7' to 'opset7_to_10'

* Added more test cases for Unsqueeze

* Added a test case for unsqueezing 3 dims

Also renamed the 1 dim test cases slightly.

* Added more test cases for Unsqueeze

* Added a test case for unsqueezing 3 dims

Also renamed the 1 dim test cases slightly.

* Update docs/Operators.md

Feedback from wschin to fix axis bounds.

Co-Authored-By: Wei-Sheng Chin <wschin@outlook.com>

* Re-ran update_doc.sh

* support attribute as empty string

* support attribute as empty string

* code collation

* Add description of default type about y_zero_point

In QuantizeLinear input document.

* Update Changelog.md

* Update Operators.md

…2320)

* 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.  (#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 (#1929)

* fix shape inference and add tests for shape inference

* cosmetic fixes

* plus some formatting

* Fix shape inference for matmul (#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

* change incorrect use of ValueError to TypeError

* change TypeError

* message

…age (#2133)

* Initial commit

* More changes

* Bug fix in ROIAlign shape inf

* Minor name change

* Merge master

* Update defs.cc

* Revamped test cases for Gemm

* Increased number of tests from 2 to 8.

* The tests each test a specific attribute or input case.

* Generated test data.

* Updated the operators doc.

* Updated the test coverage.

* Fixed linting errors causing CI failures

* Updated the TestCoverage and Operators docs for the newly re-linted Gemm tests.

* Tweaks to Gemm test cases:

1. Differentiated between sclar and single element vector biases.

2. Changed the name of 'test_gemm_default_no_bias' to 'test_gemm_default_zero_bias' to be more clear.

3. Added back the test case for specifying all attributes as non-default.

* Added a reference implementation for the Gemm op

* Fixed some mistakes with the types of transA and transB.

Also added the mypy type annotation for the Gemm reference implementation.

* The Gemm scalar bias test actually adds a scalar bias now

* Updated the docs for the strides and dilations attributes to mention the defaults.

* Whoops, forgot to add stride defaults to MaxUnpool.

* Bumped the version numbers after specifying the default values for stride and dilations.

The ops that had their version numbers bumped are AveragePool, MaxPool, MaxUnpool, LpPool, Conv, and ConvTranspose.

Also moved some old definitions from nn/defs.cc to nn/old.cc

* Fixed conflicts and merged master

* Revert "Fixed conflicts and merged master"

This reverts commit b45c6a0.

* Revert "Revert "Fixed conflicts and merged master""

This reverts commit a4ef81b.

* Fixed duplicated operator set registrations

* Revamped test cases for reshape

* Increased the number of test cases from 5 to 7.

    * The new tests cover the 0 dim case

* Changed the existing cases so that, where possible, the 'batch' dim is unchanged.

    * Some frameworks do not allow for the first dim to change, so this makes these tests more generally applicable.

* Generated the new test data, updated the operators doc, and updated the test coverage doc.

* Re-generated Operators.md and TestCoverage.md

* Added a reshape op reference implementation and more tests for reshaping the first dim of a tensor

* Changes to build on AIX Platform

* Update numpy_helper.py

* Update CMakeLists.txt

* Made the 'C' input of Gemm (the bias term) optional.

If missing it defaults to 0.

Also added a test case for no bias.

Updated the Gemm op to version 11.

* Fixed a typo!

* Small tweaks to the Gemm docs.

* Added a shape inference test for Gemm with no bias

* Tweaked coding style slightly by adding braces to single line scopes.

* Fix some node tests

* PR comments and docs

* Update Changelog.md

* Update gen_doc script to validate proto3 files

* Update CMakeLists.txt

* Fix some node tests

* PR comments and docs

* Update Changelog.md

* Fix gemm scalar node test

* Fix the wrong behavior in ConvTranspose

* Address comments

* Relax If's shape inference rule

* Make shape inference tests ok and move code to the right place

* Add document changes

* Update onnx/defs/controlflow/defs.cc

* Update onnx/defs/controlflow/defs.cc

* Address comments

* Address comments

* Address comments

* Fix shape inference test

* Disable a type check

* Address a comment

* Update defs.cc

* Update Changelog.md

* Update Operators.md

* Bump NMS version for avoiding regression in existing models

* Bring old logic back

* Fix spec for Unsqueeze

* Update Changelog.md

* Refine doc

* Refine

* PR comments

* Update Changelog.md

* Update shape inference test file

The requirement is already part of the install_requires so this is
redundant. (It should only be required for python_version < 3.5 in any
case.)

Include 1.6 opset mapping row.

Include 1.6 opset mapping row.

* Incrementing version number to 1.6.0

* Updating ML version to 2

* update shapeInference h file link

* correction

* Fix Unique unit test

Expected data for Y was incorrectly sorted

* Update test coverage documentation

* Update Operators.md

* Clarify example in Loop documentation

* Regenerate op documentation

* generate op documentation

* Address PR review comment

* Address PR feedback

* Update defs.cc

* Update Changelog.md

* Update Operators.md

* add type check for node inputs

* add homogeneous check

* check output type

* split type check function

* cast size to int

* add comment

* infer type during checking

* refactor code

* fix typo

* switch to using context

* extend checker logic

* fail check on undefined inputs

* refactor code

* skip undefined cases

* fix comment

* skip type check on empty constraints

* print out type string

* print data type to debug UT fail

* use string to represent type

* flag on/off type check

* add missing param

* set default arg

* set default arg

* add type string

* fix typo

* fix check error

* add UT

* format code

…ing on what's in memory prior to the constant char string. (#2403)

e.g. you get this:
"[ShapeInferenceError]  , 1]. Its actual value is: -10"
instead of
"[ShapeInferenceError] 'axis' must be in [-2 , 1]. Its actual value is: -10"

)

* Add shape existence check guard in GatherElements shape inference logic

* PR comments

* PR comments

* minor changes to NonZero and Slice

* more updates

* Fix bug in function body verifier

* fix int type

* fix CI compile error

* Improve error message and eliminate unnecessary check

* address PR comments and remove a check

* add int cast

* Exclude Range from tested functions

* Delete duplicate comment word in onnx.in.proto

Also generate the other proto files that are generated from onnx.in.proto

* Use gen_proto.py option "--ml --lite" to generate the onnx-ml version protos.

* fix build issue for windows conda env

* Plus few typo fixes

* add option to turn of use of static runtimes + fix appveyor ci

* Fix Appveyor build for Miniconda35-x64

* address review comments

* updates to build instructions

* update argmin/argmax

* gen doc and tests

* fix typo

* update doc

Updated license to https://github.com/onnx/onnx/blob/master/LICENSE

…2486)

* fix the optimize pass of fuse_consecutive_transposes

* update the bound checker

* make sure the graph and input/output tensors are consistent before and after optimization

* correct typeerror

* value data type correction

* little value

* Edited PythonAPIOverview.md

The example given in the "Creating an ONNX Model Using Helper
Functions" example.

* Edited PythonAPIOverview.md

The example given in the "Creating an ONNX Model Using Helper
Functions" example was not working as expected.

Running the given code would throw a ValidationError() regarding
the node specification (i.e.: "Context: Bad node spec").

This change uses the current node specification, solving the issue.

* add 8 bit support to some reduction ops

* plus updates

* add 8 bit support to maxpool op

* plus review comments

Before defining operator<<(std::ostream&, const Status&).

Co-authored-by: Lu Fang <30275821+houseroad@users.noreply.github.com>

The order of arguments in implementation.cc mismatches with
the header. Since `check_type` is an optional argument, it
should appear after `opset_imports`.

Co-authored-by: bddppq <baidingding7@gmail.com>

…11 (the opset version of the latest stable), but 10 (#2478)

Co-authored-by: Ke Zhang <kezhan@microsoft.com>

Co-authored-by: Ke Zhang <kezhan@microsoft.com>

Draft was shared on Gitter for feedback in December and approved by Steering Committee on 1/9

* Keep symbolic dims in Concat with a single input

When there is only a single input, we need to discard the
symbolic dimension in the specified axis.

* Validate axis of Concat with a single input

Co-authored-by: Ke Zhang <kezhan@microsoft.com>

* Fix shape inference for Split with split attribute

This fixes #1735

* Set rank even when split dimension is unknown

* Let shape inference fail for invalid split

* Fail shape inference for inequal division

* Ignore Python type error in the unittest

* Remove tests for failure scenarios

* Stop using camel case for local variables

Co-authored-by: Ke Zhang <kezhan@microsoft.com>

…nto other markdown rather than versioning.md.

…nto other markdown rather than versioning.md.