Creating Tensors: Setting DataTypes

In [1]:

import tensorflow as tf

In [2]:

floatArr = [1.7, 7.4]
tensorFloat = tf.constant(floatArr)
tensorFloat.dtype

Out [2]:

tf.float32

In [4]:

# 
# using tf.cast
# 
castedAs16 = tf.cast(tensorFloat, dtype=tf.float16)
castedAs16.dtype

Out [4]:

tf.float16