ValueError: This ORT build has ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider'] enabled. Since ORT 1.9, you are required to explicitly set the providers parameter when instantiating InferenceSession. For example, onnxruntime.InferenceSession(..., providers=['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider'], ...)

Question

ValueError: This ORT build has ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider'] enabled. Since ORT 1.9, you are required to explicitly set the providers parameter when instantiating InferenceSession. For example, onnxruntime.InferenceSession(..., providers=['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider'], ...)

onui05211 opened this issue a year ago · comments

我在互联网搜索解决方法，尝试了很多，但是始终无法解决修改过
model = onnxruntime.InferenceSession(self._model_fp)
onnxruntime.InferenceSession(..., providers=['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider']
请问各位大佬有遇到过这种情况吗？

onui05211 · Answer 1 · Sun Jun 25 2023 10:13:14 GMT+0800 (China Standard Time)

此问题已经得到解决，根据他的提示，我在recognizer.py第187行，更改model = onnxruntime.InferenceSession(self._model_fp)为model = onnxruntime.InferenceSession(self._model_fp,providers=['CUDAExecutionProvider', 'CPUExecutionProvider'])
在utility.py157行更改sess = ort.InferenceSession(model_file_path)为sess = ort.InferenceSession(model_file_path,providers=['TensorrtExecutionProvider,CUDAExecutionProvider','CPUExecutionProvider'])

ZOOU Qinn · Answer 2 · Tue Sep 26 2023 10:59:38 GMT+0800 (China Standard Time)

相同安装步骤，有的PC上没有问题，另一些则报这种错误。请问是什么原因造成？
针对这种情况，会不会对代码作修改？

Breezedeus · Answer 3 · Tue Sep 26 2023 13:55:20 GMT+0800 (China Standard Time)

应该把onnxruntime版本降到 <1.16 就可以。onnxruntime新版接口做了调整，过几天会发布新版cnocr