curiosity/baseline_reader.py (4 lines):
	- line 191: # TODO: Add in facts
	- line 263: # TODO: These are already space tokenized
	- line 357: # TODO: Add in facts
	- line 429: # TODO: These are already space tokenized


curiosity/reader.py (3 lines):
	- line 192: # TODO: Add in facts
	- line 254: # TODO: These are already space tokenized
	- line 343: # TODO: Add in facts


configs/model.jsonnet (3 lines):
	- line 37: # TODO: add character embeddings, but this is not trivial due to
	- line 91: // TODO: GRU?
	- line 96: // TODO: can be bidirectional, but need to take care


configs/baseline_model.jsonnet (3 lines):
	- line 37: # TODO: add character embeddings, but this is not trivial due to
	- line 91: // TODO: GRU?
	- line 96: // TODO: can be bidirectional, but need to take care


curiosity/paraphrase_models.py (2 lines):
	- line 33: TODO: add dialog & dialog acts history
	- line 150: # TODO (pradeep): Do not hardcode decoder cell type.


curiosity/metrics.py (1 line):
	- line 40: # TODO: This could be batched, but its a pain


curiosity/nn.py (1 line):
	- line 34: # TODO: pass in the prior message


curiosity/models.py (1 line):
	- line 247: # TODO: This could instead of averaged, be attended