codegen_sources/preprocessing/obfuscation/bobskater_frameUtils.py (8 lines):
	- line 11: # TODO: After coming back to this a second time, the names aren't really sticking,
	- line 114: TODO: Convert to iterator, not list generator
	- line 125: a StopIteration error (TODO: Make it return None)
	- line 130: # TODO: This could be better an iterator, not a list return
	- line 195: TODO: Properties that are not defined (that are None) just come back as blanks,
	- line 198: TODO: If we need more granularity, we need to edit how this works (would need
	- line 223: # TODO:keyword (in Python <3.3)
	- line 227: # TODO: Annotations (for everything x:)


codegen_sources/preprocessing/obfuscation/bobskater_obfuscator.py (3 lines):
	- line 194: # TODO: Annotations (for everything x:)
	- line 281: # TODO: Name should eventually be unique per scope, as we
	- line 352: # TODO: I have no idea if this functions in not Python 3.5


codegen_sources/model/src/trainer.py (3 lines):
	- line 588: pred_mask[0] = 0  # TODO: remove
	- line 671: # TODO check that sentences are < max_len like in deobfuscate_by_variable
	- line 938: ):  # AMP checkpoint reloading is buggy, we cannot do that - TODO: fix - https://github.com/NVIDIA/apex/issues/250


codegen_sources/preprocessing/obfuscation/javalang_obfuscator.py (2 lines):
	- line 31: # TODO could make it O(log(n)) with binary search for find first token
	- line 37: # TODO: check type for variable definitions?


codegen_sources/preprocessing/bpe_modes/bpe_mode.py (1 line):
	- line 16: # TODO add restore BPE of XLM utils into that class


codegen_sources/preprocessing/lang_processors/tree_sitter_processor.py (1 line):
	- line 109: # TODO make this cleaner with tree sitter AST ?


codegen_sources/wrappers/models.py (1 line):
	- line 32: # TODO -> do it cleaner way for opensourcing


codegen_sources/model/src/model/__init__.py (1 line):
	- line 183: # TODO: only output when necessary - len(params.clm_steps + params.mlm_steps) > 0


codegen_sources/preprocessing/dataset_modes/monolingual_functions_mode.py (1 line):
	- line 25: # TODO make option to go from tokenized version, not necessary to retokenized the whole thing, a fonction "Select starting point"


codegen_sources/preprocessing/dataset_modes/dataset_mode.py (1 line):
	- line 408: # TODO check number of lines


codegen_sources/model/src/evaluation/evaluator.py (1 line):
	- line 1032: # TODO clean lang1


codegen_sources/preprocessing/lang_processors/python_processor.py (1 line):
	- line 176: # TODO raise ValueError(f'Invalid python function \n {code}\n')