RuntimeError: (NotFound) Operator (preln_residual_bias) is not registered.

[matakk]

问题确认 Search before asking

• [X] 我已经查询历史issue，没有发现相似的bug。I have searched the issues and found no similar bug report.

Bug组件 Bug Component

Inference

Bug描述 Describe the Bug

paddledetection pipeline 推理: 使用paddle模式推理,是可以的,入下图

改用tensorrt:

fp32 fp16 int8都是这个错误. 我搜索也没有相关的说明. 请问如何调试啊?

复现环境 Environment

环境: Windows 10,miniconda,python 3.10 gpu:p40单颗 cuda环境版本: cuda_10.2.89_441.22_win10.exe cudnn-10.2-windows10-x64-v7.6.5.32.zip TensorRT-7.0.0.11.Windows10.x86_64.cuda-10.2.cudnn7.6.zip

Bug描述确认 Bug description confirmation

• [X] 我确认已经提供了Bug复现步骤、代码改动说明、以及环境信息，确认问题是可以复现的。I confirm that the bug replication steps, code change instructions, and environment information have been provided, and the problem can be reproduced.

是否愿意提交PR？ Are you willing to submit a PR?

• [ ] 我愿意提交PR！I'd like to help by submitting a PR!

[matakk]

仔细看了文档: 重新部署环境,安装的官方的这个文件 paddlepaddle_gpu-2.3.2-cp38-cp38-win_amd64.whl 可以推理, 需要调整 trt shape的参数.部分图像可以推理.

参考: https://github.com/PaddlePaddle/PaddleDetection/blob/05ab413e2b5406416733ca1674e954fc80638fed/deploy/TENSOR_RT.md