train.py TypeError: unhashable type

[antopost]

Getting the following error when starting training.py:

[*] Training starts...
Traceback (most recent call last):
File "/home/anton/PycharmProjects/Jmod2clone/J-MOD2/train.py", line 33, in
tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)
File "/home/anton/anaconda3/envs/jmodclone/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 125, in run
_sys.exit(main(argv))
File "/home/anton/PycharmProjects/Jmod2clone/J-MOD2/train.py", line 26, in main
trainer.train()
File "/home/anton/PycharmProjects/Jmod2clone/J-MOD2/lib/trainer.py", line 25, in train
self.model.train()
File "/home/anton/PycharmProjects/Jmod2clone/J-MOD2/models/AbstractModel.py", line 178, in train
tb_x, tb_y = self.tensorboard_data_generator(self.config.max_image_summary)
File "/home/anton/PycharmProjects/Jmod2clone/J-MOD2/models/AbstractModel.py", line 123, in tensorboard_data_generator
curr_train_sample_list = self.training_set[0:num_samples]
TypeError: unhashable type
Samples per epoch: 0

I changed all the parameters in the config file accordingly i.e. correct path, is_train=True, etc.
Working with the Unreal Dataset provided.
I noticed that "self.training_set = {}" in the init of AbstractModel and does not get filled with an actual training set. That should be causing the error. But I don't understand why it stays empty.