Summary: 11 instances, 11 unique

Text	Count
# TODO: Very basic	1
# TODO: Move to utils/training.py	1
# TODO: Take kwargs as input and separate between get_preprocessors kwargs and generate_kwargs	1
# TODO: Quick hack to remove empty directories	1
return np.log(1 + len(get_word2rank()))  # TODO: This is completely arbitrary	1
# TODO: #language_specific	1
# TODO: Default args are not overwritten if provided as args	1
# TODO: Locking a directory should lock all files in that directory	1
# TODO: Do we really need to extract the tokens	1
# TODO: Decrease vocab size or load from smaller file	1
# TODO: Nested with lock_directory() should not be blocking	1