pytorch-transformers/pytorch_transformers/modeling_xlnet.py (2 lines):
	- line 750: qlen: TODO Lysandre didn't fill
	- line 751: mlen: TODO Lysandre didn't fill


XLM/src/trainer.py (2 lines):
	- line 455: pred_mask[0] = 0  # TODO: remove
	- line 557: if False:  # AMP checkpoint reloading is buggy, we cannot do that - TODO: fix - https://github.com/NVIDIA/apex/issues/250


pytorch-transformers/pytorch_transformers/tokenization_xlm.py (1 line):
	- line 699: # TODO: make sure we are using `xlm-mlm-enro-1024`, since XLM-100 doesn't have this step


XLM/src/model/transformer.py (1 line):
	- line 417: # TODO: add extra layer norm here?


pytorch-transformers/pytorch_transformers/modeling_transfo_xl.py (1 line):
	- line 1357: # TODO: This is not implemented


XLM/src/model/__init__.py (1 line):
	- line 143: encoder = TransformerModel(params, dico, is_encoder=True, with_output=False)  # TODO: only output when necessary - len(params.clm_steps + params.mlm_steps) > 0