xlm/trainer.py (2 lines):
	- line 444: pred_mask[0] = 0  # TODO: remove
	- line 546: if False:  # AMP checkpoint reloading is buggy, we cannot do that - TODO: fix - https://github.com/NVIDIA/apex/issues/250


src/trainer.py (2 lines):
	- line 444: pred_mask[0] = 0  # TODO: remove
	- line 546: if False:  # AMP checkpoint reloading is buggy, we cannot do that - TODO: fix - https://github.com/NVIDIA/apex/issues/250


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


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


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


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