1、这边我使用了官网的模型推理：https://www.hiascend.com/document/detail/zh/canncommercial/700/quickstart/quickstart/quickstart_18_0008.html （容器内操作）在执行到编译样例源码和运行样例的时候，出现以下错误，尝试声明环境变量后有如下报错： root@7bd889ecfeb7:/home/ls/samples-7.0.0/inference/modelInference/sampleResnetQuickStart/cpp/scripts# bash sample_run.sh [INFO] The sample starts to run ./main: error while loading shared libraries: libascend_hal.so: cannot open shared object file: No such file or directory [INFO] The program runs failed root@7bd889ecfeb7:/home/ls/samples-7.0.0/inference/modelInference/sampleResnetQuickStart/cpp/scripts# export LD_LIBRARY_PATH=/usr/local/Ascend/ascend-toolkit/latest/aarch64-linux/devlib/aarch64:$LD_LIBRARY_PATHroot@7bd889ecfeb7:/home/ls/samples-7.0.0/inference/modelInference/sampleResnetQuickStart/cpp/scripts# bash sample_run.sh [INFO] The sample starts to run [ERROR]: stub library cannot be used for execution, please check your environment variables and compilation options to make sure you use the correct ACL library. [ERROR] aclInit failed, errorCode is 100039 [ERROR] InitResource failed [ERROR]: stub library cannot be used for execution, please check your environment variables and compilation options to make sure you use the correct ACL library. [ERROR]: stub library cannot be used for execution, please check your environment variables and compilation options to make sure you use the correct ACL library. [ERROR]: stub library cannot be used for execution, please check your environment variables and compilation options to make sure you use the correct ACL library. [ERROR]: stub library cannot be used for execution, please check your environment variables and compilation options to make sure you use the correct ACL library. [ERROR]: stub library cannot be used for execution, please check your environment variables and compilation options to make sure you use the correct ACL library. [ERROR] destroy description failed, errorCode is 100039 [ERROR]: stub library cannot be used for execution, please check your environment variables and compilation options to make sure you use the correct ACL library. [ERROR] unload model failed, errorCode is 100039 [ERROR]: stub library cannot be used for execution, please check your environment variables and compilation options to make sure you use the correct ACL library. [ERROR] aclrtResetDevice failed, errorCode is 100039 [ERROR]: stub library cannot be used for execution, please check your environment variables and compilation options to make sure you use the correct ACL library. [ERROR] aclFinalize failed, errorCode is 100039 [INFO] The program runs failed 2、如果容器只挂载原始的npu，即：可以执行推理嘛？还是必须要升级才行呢？目前容器内是能够回显npu-smi info但是在安装cann包时提示没有安装驱动 docker attach 登录容器，即设置主线程才能生效