Change Loop op with maximum iterations input M equals to empty string

[hwangdeyu]

while_loop maximum iterations -1 means infinity in tf, but in tf2onnx it's set to max_int to get the same behavior so far.
Refer to Loop spec input M, pass empty string to skip should be better.

fixes #1969
Signed-off-by: Deyu Huang [email protected]

[hwangdeyu]

no need to add new unit tests cause there are a lots of existing tests can cover it.

[fatcat-z]

LGTM, thanks!

[LoicDagnas]

@hwangdeyu I might be wrong but having an empty string for the loop M input make the symbolic shape inference of ONNX runtime failed here. Since M parameter is marked as optional in the official ONNX documentation, it might be fixed on onnxruntime side?

[hwangdeyu]

Yeah, M should pass empty string to skip in ONNX spec. I don't run into the shape inference error. It's ok to create an issue in ONNXRuntime. Or you share your case with us a simple reproduction script code then we will take a look.

@hwangdeyu Deyu Huang FTE I might be wrong but having an empty string for the loop M input make the symbolic shape inference of ONNX runtime failed here. Since M parameter is marked as optional in the official ONNX documentation, it might be fixed on onnxruntime side?

[LoicDagnas]

Running the following command:

python -m onnxruntime.tools.symbolic_shape_infer
    --input path/to/model.onnx
    --output output-model.onnx
    --auto_merge

work using a model converted with tf2onnx==1.11.1 but failed with the same model converted with tf2onnx==1.12.0

Here is the shape inference error:

Traceback (most recent call last):
  File "C:\Users\loic.dagnas\AppData\Local\Programs\Python\Python39\lib\runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "C:\Users\loic.dagnas\AppData\Local\Programs\Python\Python39\lib\runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "C:\dev\ml\debug_arnaud\venv\lib\site-packages\onnxruntime\tools\symbolic_shape_infer.py", line 2424, in <module>
    out_mp = SymbolicShapeInference.infer_shapes(
  File "C:\dev\ml\debug_arnaud\venv\lib\site-packages\onnxruntime\tools\symbolic_shape_infer.py", line 2357, in infer_shapes
    all_shapes_inferred = symbolic_shape_inference._infer_impl()
  File "C:\dev\ml\debug_arnaud\venv\lib\site-packages\onnxruntime\tools\symbolic_shape_infer.py", line 2127, in _infer_impl
    self.dispatcher_[node.op_type](node)
  File "C:\dev\ml\debug_arnaud\venv\lib\site-packages\onnxruntime\tools\symbolic_shape_infer.py", line 1066, in _infer_Loop
    si.CopyFrom(self.known_vi_[node.input[i]])
KeyError: ''

Here is the model in ONNX format
https://drive.google.com/file/d/19jOktShuha40ib2CRGSugfhi_kIjZh_p/view?usp=sharing

If it is preferable, I'll open an issue on ONNXRuntime repo.

[hwangdeyu]

Running the following command:

python -m onnxruntime.tools.symbolic_shape_infer
    --input path/to/model.onnx
    --output output-model.onnx
    --auto_merge

work using a model converted with tf2onnx==1.11.1 but failed with the same model converted with tf2onnx==1.12.0

Here is the shape inference error:

Traceback (most recent call last):
  File "C:\Users\loic.dagnas\AppData\Local\Programs\Python\Python39\lib\runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "C:\Users\loic.dagnas\AppData\Local\Programs\Python\Python39\lib\runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "C:\dev\ml\debug_arnaud\venv\lib\site-packages\onnxruntime\tools\symbolic_shape_infer.py", line 2424, in <module>
    out_mp = SymbolicShapeInference.infer_shapes(
  File "C:\dev\ml\debug_arnaud\venv\lib\site-packages\onnxruntime\tools\symbolic_shape_infer.py", line 2357, in infer_shapes
    all_shapes_inferred = symbolic_shape_inference._infer_impl()
  File "C:\dev\ml\debug_arnaud\venv\lib\site-packages\onnxruntime\tools\symbolic_shape_infer.py", line 2127, in _infer_impl
    self.dispatcher_[node.op_type](node)
  File "C:\dev\ml\debug_arnaud\venv\lib\site-packages\onnxruntime\tools\symbolic_shape_infer.py", line 1066, in _infer_Loop
    si.CopyFrom(self.known_vi_[node.input[i]])
KeyError: ''

Here is the model in ONNX format https://drive.google.com/file/d/19jOktShuha40ib2CRGSugfhi_kIjZh_p/view?usp=sharing

If it is preferable, I'll open an issue on ONNXRuntime repo.

It looks ok for me that M is empty string, I think you can try to create an issue. Thanks!