pytorch模型转tensorrt，pad报错

_此心安处是吾乡

于 2023-09-07 22:24:04 发布

阅读量583

点赞数 1

分类专栏： bug日志文章标签： pytorch 人工智能 python

本文链接：https://blog.youkuaiyun.com/u013256285/article/details/132747199

版权

记一次奇葩经历

问题

在convert_to_onnx_and_check_trt时报错

报错信息：

[09/07/2023-20:50:46] [TRT] [E] [shuffleNode.cpp::symbolicExecute::392] Error Code 4: Internal Error (Reshape_276: IShuffleLayer applied to shape tensor must have 0 or 1 reshape dimensions: dimensions were [-1,2])
2023-09-07 20:50:46,481 - trt_inference - ERROR - >>> ERROR: Failed to parse the ONNX file.
[ERROR] [09/07/2023 08:50:46 PM trt_onnx_inference_engine.py:0149] >>> ERROR: Failed to parse the ONNX file.
2023-09-07 20:50:46,481 - trt_inference - ERROR - >>> ERROR: In node 286 (parseGraph): INVALID_NODE: Invalid Node - Pad_286
[shuffleNode.cpp::symbolicExecute::392] Error Code 4: Internal Error (Reshape_276: IShuffleLayer applied to shape tensor must have 0 or 1 reshape dimensions: dimensions were [-1,2])
[ERROR] [09/07/2023 08:50:46 PM trt_onnx_inference_engine.py:0151] >>> ERROR: In node 286 (parseGraph): INVALID_NODE