Question about train model function

In this module
[https://github.com/allenai/scicite/blob/master/scicite/training/train_multitask.py#L316](url)
in this line

https://github.com/allenai/scicite/blob/b33a8cd0df5743c9d483b1d0fdce8ec62bc8f135/scicite/training/train_multitask.py#L316

datasets_for_vocab_creation as I understand we got it from the configuration file, but neither

aclarc-experiment.json
nor

scicite-experiment.json

has this parameter

Could you please explain what is wrong here?