I have trained a model and want to restore it every time i test it. But I can't use tf.get_variable() to get compiler from the graph and I have to run its initialization each time. I think if I get the compiler.input_tensors, I can just use it in session. It is mutually exclusive with `loom_input_tensor', Is anyone know how to use it?
I have trained a model and want to restore it every time i test it. But I can't use tf.get_variable() to get compiler from the graph and I have to run its initialization each time. I think if I get the compiler.input_tensors, I can just use it in session. It is mutually exclusive with `loom_input_tensor', Is anyone know how to use it?