TensorFlow hashmap with different shaped Tensors as values -
i building input pipeline in tensorflow , need use hashmap.
the hashmap needs map characters (formatted output tf.decode_csv) tensorflow constants. constants mapped not same size.
using hashmap built tensorflow causes issues due values being tensors.
for example trying create hashmap (but tensorflow hashmap):
{"1" : tf.zeros([2,1]), "2" : tf.zeros([2,2])} i cannot use above type of dictionary because inputs operations data pipeline, not values contained within them.
so i'm trying create operations tensorflow input pipeline map set of characters ["b","1","3"] set of tensors[[0,1] , [0,0] , [[0,0],[0,0],[0,0]]
Comments
Post a Comment