FuseReluClip Unexpected data type for Clip 'min' input of 11

[Cookiee235]

Describe the issue

The valid model which has been checked by the onnx.checker.check_model(model) statement crash when executing the inference unexpected.
The crash message is shown below:

The Crash stacktrace

 Traceback (most recent call last):
  File "test.py", line 10, in <module>
    original_session = ort.InferenceSession(model_path)
  File "C:\software\conda\envs\OPTFuzz\lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 419, in __init__
    self._create_inference_session(providers, provider_options, disabled_optimizers)
  File "C:\software\conda\envs\OPTFuzz\lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 491, in _create_inference_session
    sess.initialize_session(providers, provider_options, disabled_optimizers)
onnxruntime.capi.onnxruntime_pybind11_state.RuntimeException: [ONNXRuntimeError] : 6 : RUNTIME_EXCEPTION : Exception during initialization: D:\a\_work\1\s\onnxruntime\core\optimizer\relu_clip_fusion.cc:77 onnxruntime::FuseReluClip::Apply Unexpected data type for Clip 'min' input of 11

To reproduce

Step 1: Download the model via this link

Step 2: run the following script:

import onnx
import onnxruntime as ort

model_path = "model_with_clip.onnx"
model = onnx.load(model_path)
onnx.checker.check_model(model)
print(onnx.helper.printable_graph(model.graph))

original_session = ort.InferenceSession(model_path)

Urgency

No response

Platform

Linux

OS Version

Ubuntu 20.04

ONNX Runtime Installation

Built from Source

ONNX Runtime Version or Commit ID

5c1b7cc

ONNX Runtime API

Python

Architecture

X64

Execution Provider

Default CPU

Execution Provider Library Version

No response

[skottmckay]

ORT doesn't support fp64 for the min/max. We typically support common production use-cases. Not sure there's any need to use fp64 for the min/max so I'd suggest changing those to be fp32.

[xadupre]

onnxruntime should support min/max for float64 in onnxruntime but the optimizer FuseReluClip is not (see line https://github.com/microsoft/onnxruntime/blob/main/onnxruntime/core/optimizer/relu_clip_fusion.cc#L61). I wonder if we should fail on this case.

[skottmckay]

If the implementation has limited type support FuseReluClip::SatisfyCondition should be checking the input types.