🚧 This repository is under development and not yet ready for use 🚧
This is the official implementation of the paper "A Pre-Trained Graph-Based Model for AdaptiveSequencing of Educational Documents".
It contains several modules, mainly for the training of the recommender system, the simulation of the student population and the extraction of keywords from documents.
In order to run the experiments properly, please include the files (pre-trained models) provided in the supplementary materials at the right paths.
To install the required dependencies, please run this in command line:
./install.sh
We use weights&biases for the logging of experimental results.
To pretrain the model on the source tasks, execute this command from the root folder:
python -m src.rl_pretrain
To finetune the model on the target task, execute this command from the root folder:
python -m src.rl_finetune
To run the Bassen et al. baseline, execute this command from the root folder:
python -m src.rl_finetune_bassen
To run the Vassoyan et al. baseline, execute this command from the root folder:
python -m src.rl_finetune_vassoyan
To run the CMAB baseline, execute this command from the root folder:
python -m src.baseline_irt.main