We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Mainly Tensorflow (?) but other backends might need investigation as well, restricting I/O access.