Eedi Reranker

This project implements a reranking model for misconceptions in educational data using a listwise approach. The model is trained and evaluated on a dataset of misconceptions, leveraging a pre-trained language model with fine-tuning. 

infer notebook synthetic data

Requirements

• Python 3.8+
• Required Python libraries are listed in requirements.txt.

Installation

   git clone <repository-url>
   cd <repository-directory>
   python src/generate_data.py

    # split data into 5 folds
    python src/split_fold.py
    
    # knowledge distillation
    python src/distill.py
    
    # training biencoder model
    python src/train_biencoder.py --fold 4
    python src/train_biencoder.py --fold 3
    python src/train_biencoder.py --fold 2
    python src/train_biencoder.py --fold 1
    python src/train_biencoder.py --fold 0
    
    # run biencoder model and get topk ids
    python src/run_biencoder.py
    
    # training reranking model
    python src/train_reranking.py --fold 4

Authors

• Nguyen Huu Cong
  Email: cong.nh225476@sis.hust.edu.vn

• Vu Hoang Nhat Anh
  Email: anh.vhn20225471@sis.hust.edu.vn

• Nguyen Duc Anh
  Email: anh.nd20225468@sis.hust.edu.vn

• Do Tran Gia Bach
  Email: bach.dtg225472@sis.hust.edu.vn

• Tran Thanh Vinh
  Email: vinh.tt225539@sis.hust.edu.vn