import tensorflow 提示无法加载 libnvinfer.so.6 libnvinfer_plugin.so.6
tensorflow
一个面向所有人的开源机器学习框架
项目地址:https://gitcode.com/gh_mirrors/te/tensorflow

·
我在阿里云的ECS使用docker
运行tensorflow
的时候,当import tensorflow
的时候提示无法加载 libnvinfer.so.6
libnvinfer_plugin.so.6
的错误信息。
2020-02-25 09:24:32.442272: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libnvinfer.so.6'; dlerror: libnvinfer.so.6: cannot open shared object file: No such file or directory
2020-02-25 09:24:32.442409: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libnvinfer_plugin.so.6'; dlerror: libnvinfer_plugin.so.6: cannot open shared object file: No such file or directory
2020-02-25 09:24:32.442423: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:30] Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.
实际上这只是个警告信息,而不是错误。libnvinfer.so.6
和libnvinfer_plugin.so.6
只在使用NVIDIA TensorRT
的时候才是必须的。没有这两个库也可以运行tensorflow。
关于NVIDIA TensorRT:
NVIDIA TensorRT是一种高性能神经网络推理(Inference)引擎,用于在生产环境中部署深度学习应用程序,应用有图像分类、分割和目标检测等,可提供最大的推理吞吐量和效率。TensorRT是第一款可编程推理加速器,能加速现有和未来的网络架构。TensorRT需要CUDA的支持。TensorRT包含一个为优化生产环境中部署的深度学习模型而创建的库,可获取经过训练的神经网络(通常使用32位或16位数据),并针对降低精度的INT8运算来优化这些网络。借助CUDA的可编程性,TensorRT将能够加速助推深度神经网络日益多样化、复杂的增长趋势。通过TensorRT的大幅度加速,服务提供商能够以经济实惠的成本部署这些计算密集型人工智能工作负载。
参考文献:
python - Could not load dynamic library ‘libnvinfer.so.6’ - Stack Overflow




一个面向所有人的开源机器学习框架
最近提交(Master分支:12 天前 )
4f64a3d5
Instead, check for this case in `ResolveUsers` and `ResolveOperand`, by querying whether the `fused_expression_root` is part of the `HloFusionAdaptor`.
This prevents us from stepping into nested fusions.
PiperOrigin-RevId: 724311958
12 天前
aa7e952e
Fix a bug in handling negative strides, and add a test case that exposes it.
We can have negative strides that are not just -1, e.g. with a combining
reshape.
PiperOrigin-RevId: 724293790
12 天前
更多推荐
所有评论(0)