1、tf.variable（<variable_name>）与tf.get_variable（<variable_name>）:
When the former creates a variable, it will automatically deal with the same variable when it encounters it. The way to deal with it is to rename it, like 1-1, 1-1-1. When the latter creates a variable, it actually acquires the variable. When the variable exists, it is directly acquired, and when the variable does not exist, it is created.
2、tf.name_scope(<scope_name>)与tf.variable_scope(<scope_name>)：
tf.name_scope(<scope_name>) returns a context manager named scope_name, which is mostly used for graph management. When creating a variable with the same name, an error will be reported.
tf.variable_scope(<scope_name>) A variable that can be shared. If there is a duplicate name, no error will be reported, and it is regarded as a shared variable.