onnxruntime produces invalid results due to the wrong shape inference for the clip operator

[coffezhou]

Describe the issue

For the following onnx model,

the shape of result should be 1x3x1x1. But onnxruntime produces wrong shape of results, 1x3x1x3. The outputs are as follows:

ONNXRuntime:
 [array([[[[0.0192995 , 0.0192995 , 0.0192995 ]],

        [[0.00301902, 0.00301902, 0.00301902]],

        [[0.        , 0.        , 0.        ]]]], dtype=float32)]
shape:
 (1, 3, 1, 3)

According to the specification of clip operator, the results' shape of clip should be identical to that of the its inputs. So, in this issue, it is strange that why the results' shape is 1x3x1x3.

To reproduce

Environment

OS: Ubuntu 20.04
onnxruntime: 1.23.0.dev20250515001
CUDA: cuda-12.2.2::cuda-toolkit
CUDNN: 9.1.1.17
NVIDIA GPU: GeForce RTX 3080
NVIDIA Driver Version: 535.183.01
Python Version: 3.12.9

Steps to reproduce

This bug can be reproduced by the following code with the model in the attachment.

import sys
import os

import numpy as np
import onnx
import onnxruntime

import pickle
            
def main():
    onnx_model = onnx.load("111.onnx")
    
    with open("inputs.pkl", "rb") as fp:
        inputs = pickle.load(fp)
    
    try:
        ort_session = onnxruntime.InferenceSession(
            onnx_model.SerializeToString(), providers=["CPUExecutionProvider"]
        )
        ort_output = ort_session.run([], inputs)
    except Exception as e:
        print(e)
        sys.exit(1)
    
    print("ONNXRuntime:\n", ort_output)    
    print("shape:\n", ort_output[0].shape)  
    
if __name__ == "__main__":
    
    main()   

testcase.zip

Urgency

No response

Platform

Linux

OS Version

Ubuntu 20.04

ONNX Runtime Installation

Released Package

ONNX Runtime Version or Commit ID

1.23.0.dev20250515001

ONNX Runtime API

Python

Architecture

X64

Execution Provider

Default CPU

Execution Provider Library Version

No response

[yuslepukhin]

Shape inference logic for standard ops is in ONNX library. It would be better to file your issue there.

[coffezhou]

Shape inference logic for standard ops is in ONNX library. It would be better to file your issue there.

Hi @yuslepukhin ,
Thank you for your suggestion! I have reported a new issue for onnx.