Something wrong in training Clevr

[hjzzju]

It is a really impressive work, however I meet the following problem, it seems that there is something with the size, but I just follow the README.
nohup ./scripts/train_clevr.sh /home/hjz/hjzfolder/ReferringRelationships-master/checkpoints &

return self.build_ssas()
File "/home/hjz/hjzfolder/ReferringRelationships-master/models.py", line 130, in build_ssas
new_image_features = Multiply()([im_features, predicate_att])
File "/home/hjz/anaconda3/envs/tf-keras/lib/python3.6/site-packages/keras/engine/topology.py", line 571, in call
self.build(input_shapes)
File "/home/hjz/anaconda3/envs/tf-keras/lib/python3.6/site-packages/keras/layers/merge.py", line 84, in build
output_shape = self._compute_elemwise_op_output_shape(output_shape, shape)
File "/home/hjz/anaconda3/envs/tf-keras/lib/python3.6/site-packages/keras/layers/merge.py", line 55, in _compute_elemwise_op_output_shape
str(shape1) + ' ' + str(shape2))
ValueError: Operands could not be broadcast together with shapes (14, 14, 1024) (14, 14, 6)

[ranjaykrishna]

predicate_att should be (14, 14, 1) I believe. That's probably what you should check and verify.