Which method is not supported in TensorFlow to define variables?

Question

Options

• ARandom number
• BConstant
• CCalculated from the initial values of other variables
• Dnull

• A
  15% (5)
• B
  9% (3)
• C
  3% (1)
• D
  73% (24)

Explanation

Why D is correct: TensorFlow does not support initializing variables with null - variables must have a defined initial value at creation time, as TensorFlow's computation graph requires concrete tensors with known types and shapes.

Why the distractors are wrong:

• A (Random number): Valid - tf.Variable(tf.random.normal([3,3])) is a standard pattern for initializing weights.
• B (Constant): Valid - tf.Variable(tf.constant(1.0)) or simply tf.Variable(1.0) works fine.
• C (Calculated from other variables): Valid - you can derive an initial value from existing variables using TensorFlow ops before passing the result to tf.Variable().

Memory tip: Think of TensorFlow variables like strongly-typed class fields in Java or C# - they need a concrete initial value (random, constant, or computed), not a null/undefined placeholder. If it's a real value, TF takes it; if it's nothing, TF won't accept it.

No community discussion yet for this question.