Code used for the Text Complexity DE Challenge 2022 (https://qulab.github.io/text_complexity_challlenge/).
Here, you will find all the files needed to reproduce our approach followed the for challenge. In particular, complex_train.py is used to train a Gaussian Process Model, as well as run a given model on the dev and test sets. Likewise, roberta_train.py is used to fine-tune a RoBERTa model. Note that the resulting outputs of these two files must be averaged together to follow our approach exactly.
Another note: result.zip must be unzipped before running complex_train.py.