To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`

To disable TensorFlow's usage of the OneDNN optimization library (which is designed to enhance performance on modern hardware), you would need to set an environmental variable called `TF_ENABLE_ONEDNN_OPTS`. By setting this variable to `0`, you're telling TensorFlow not to use these optimizations. This can be useful if you encounter issues with OneDNN or want to test your code without its performance boost.

通常，设置环境变量是在命令行启动你的程序之前完成的。例如，在Unix-like systems, 你可以在终端输入：

export TF_ENABLE_ONEDNN_OPTS=0

然后运行你的TensorFlow脚本。在Windows系统中，则可以添加到系统的环境变量设置中。

相关问题

I tensorflow/core/util/util.cc:169] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.

这是一个 TensorFlow 的提示信息。它告诉用户，TensorFlow 正在使用 oneDNN（DNN 库）的自定义操作，因此可能会因不同的计算顺序而导致略微不同的数值结果，这是由于浮点数舍入误差所造成的。如果您想禁用 oneDNN，您可以设置环境变量 `TF_ENABLE_ONEDNN_OPTS=0`。

2023-07-14 07:37:44.816686: I tensorflow/core/util/port.cc:110] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.

您好！关于您提到的信息，这是关于 TensorFlow 的一条提示消息，指出您的环境中启用了一种叫做 oneDNN 的自定义操作。由于不同计算顺序可能会导致浮点数舍入误差，因此可能会在数值上稍有不同。如果您希望关闭该功能，可以设置环境变量 `TF_ENABLE_ONEDNN_OPTS=0`。这样就可以避免由此引起的结果差异。