ONNX转MNN出现错误

[12314312]

将ONNX转换成MNN
运行的代码：/workspace/MNN/tools/script# python testMNNFromOnnx.py /workspace/onnxfile/SeaFormer_T_bs32_36.5.onnx
Dir exist
onnx/test.onnx
tensor(float)
['output']
inputs:
input
onnx/
outputs:
onnx/output.txt (1, 1, 512, 512)
onnx/
The device supports: i8sdot:0, fp16:0, i8mm: 0, sve2: 0
Start to Convert Other Model Format To MNN Model..., target version: 2.9
[07:01:53] /workspace/MNN/tools/converter/source/onnx/onnxConverter.cpp:46: ONNX Model ir version: 6
[07:01:53] /workspace/MNN/tools/converter/source/onnx/onnxConverter.cpp:47: ONNX Model opset version: 11
Start to Optimize the MNN Net...
inputTensors : [ input, ]
outputTensors: [ output, ]
Converted Success!
Check convert result by onnx, thredhold is 0.01
input
output: output
output: (1, 1, 512, 512, )
TESTERROR output value error : absMaxV:13.000000 - DiffMax 11.000000
Error for output output
Save mnn result to .error director

出现错误。用DEBUG

Test Node : Div_730 True
onnx/test.onnx
tensor(float)
['1482']
inputs:
input
onnx/
outputs:
onnx/1482.txt (1, 512, 512)
onnx/
The device supports: i8sdot:0, fp16:0, i8mm: 0, sve2: 0
Start to Convert Other Model Format To MNN Model..., target version: 2.9
[07:06:41] /workspace/MNN/tools/converter/source/onnx/onnxConverter.cpp:46: ONNX Model ir version: 6
[07:06:41] /workspace/MNN/tools/converter/source/onnx/onnxConverter.cpp:47: ONNX Model opset version: 11
Start to Optimize the MNN Net...
inputTensors : [ input, ]
outputTensors: [ 1482, ]
Converted Success!
Check convert result by onnx, thredhold is 0.01
input
output: 1482
1482: (1, 512, 512, )
TESTERROR 1482 value error : absMaxV:21.000000 - DiffMax 19.000000
Error for output 1482
Save mnn result to .error director

Test Node : ArgMax_731 False
Error is between Div_730 and ArgMax_731
请问这是什么问题

[jxt1234]

ArgMax 有可能因为数值误差导出取出来的 index 不一样，建议用实际输入测试一下。