使用paddlex导出的onnx模型进行推理，在Ascend 310B arch64板卡上报错，cpu推理失败。

[younland]

描述问题

语义分割模型在310B板卡上cpu推理失败。

复现

1、使用paddlex训练自定义训练语义分割模型，训练完成后导出onnx

2、使用paddleseg仓库下的infer_onnx.py进行推理

3. 您使用的模型和数据集是？

4. 请提供您出现的报错信息及相关log

`
2025-04-27 15:14:46.723595316 [E:onnxruntime:, inference_session.cc:1644 operator()] Exception during initialization: /onnxruntime_src/onnxruntime/core/providers/cpu/tensor/upsamplebase.h:201 onnxruntime::UpsampleMode onnxruntime::UpsampleBase::StringToUpsampleMode(const string&) mode attribute is . It can only be nearest(default) or linear or cubic.

Traceback (most recent call last):
File "/home/HwHiAiUser/paddle/PaddleSeg/deploy/python/infer_onnx.py", line 69, in
main(args)
File "/home/HwHiAiUser/paddle/PaddleSeg/deploy/python/infer_onnx.py", line 58, in main
sess = InferenceSession(args.onnx_file)
File "/usr/local/miniconda3/lib/python3.9/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 383, in init
self._create_inference_session(providers, provider_options, disabled_optimizers)
File "/usr/local/miniconda3/lib/python3.9/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 435, 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: /onnxruntime_src/onnxruntime/core/providers/cpu/tensor/upsamplebase.h:201 onnxruntime::UpsampleMode onnxruntime::UpsampleBase::StringToUpsampleMode(const string&) mode attribute is . It can only be nearest(default) or linear or cubic.
`

[younland]

在ascend 310B上进行模型转换失败：
(base) root@davinci-mini:/home/HwHiAiUser/paddle/PaddleX# paddlex --paddle2onnx --paddle_model_dir ./models/seg --onnx_model_dir ./models/seg --opset_version 11
Input dir: models/seg
Output dir: models/seg
Paddle2ONNX conversion starting...
2025-04-27 17:21:19 [WARNING] The .pdmodel file is deprecated in paddlepaddle 3.0 and will be removed in the future. Try to convert from .pdmodel file to json file.
I0427 17:21:19.828433 1023507 program_interpreter.cc:243] New Executor is Running.
[Paddle2ONNX] Start parsing the Paddle model file...
[ERROR][Paddle2ONNX] [OP: pd_op.pool2d] Adaptive only support static input shape.
[ERROR][Paddle2ONNX] There are unsupported features.

---

C++ Traceback (most recent call last):

0 paddle2onnx::Export(char const*, char const*, char**, int*, int, bool, bool, bool, bool, bool, paddle2onnx::CustomOp*, int, char const*, char**, int*, char const*, bool*, bool, char**, int)

---

Error Message Summary:

FatalError: Process abort signal is detected by the operating system.
[TimeInfo: *** Aborted at 1745745680 (unix time) try "date -d @1745745680" if you are using GNU date ***]
[SignalInfo: *** SIGABRT (@0xf9e13) received by PID 1023507 (TID 0xe7ffd9073020) from PID 1023507 ***]

Paddle2ONNX conversion failed with exit code -6
(base) root@davinci-mini:/home/HwHiAiUser/paddle/PaddleX# pip list | grep paddle
paddle2onnx 2.0.1
paddlepaddle 3.0.0rc1
paddleseg 0.0.0.dev0 /home/HwHiAiUser/paddle/PaddleSeg
paddlex 3.0.0rc1 /home/HwHiAiUser/paddle/PaddleX

[a31413510]

pp-lite-seg目前不支持导出onnx模型，可以尝试使用其它的分割模型，如segformer、seaformer、OCRNet_HRNet系列