02 Types of Tensors

Tensorflow's tensors come in three different flavors:

  • Constant: For storing tensors that you do not expect to change in value.
  • Variable: For storing trainable values that will be updated as part of the learning process. A constant tensor can actually be passed to it as an argument to initialize its values.
  • Placeholder: Tensors that act as an interface between Tensorflow and regular Python variables. Is useful for passing your input and output data.

Each of these different types of tensors will be discussed in their own separate lesson.