tf.experimental.numpy.float16

Half-precision floating-point number type.

Character code: 'e'. Canonical name: np.half. Alias on this platform: np.float16: 16-bit-pr