Simulating Question-answering Correctness with a Conditional Diffusion

Thanks for your interest! This repository is the implementation of DSim.

Run DSim

1. download the dataset, we upload the one dataset we use (ednet with 500 samples) in https://www.dropbox.com/scl/fo/72604nn56kkhahtl3jny4/AOL6tHJo02ZdRy3k2FiOZYU?rlkey=k577qlem5s6jpz8qgk3gkl7gu&st=pqiq9mzb&dl=0.
2. move the data you download to data/ednet500/
3. run the code with: python main.py

Here are some issues you might encounter, you can address that with the solutions:

1. FileExistsError: [Errno 17] File exists: 'run/xxxx'

   please delete file run/xxx or run with "--run_exist", e.g., "python main.py --run_exist"

2. FileNotFoundError: [Errno 2] No such file or directory: 'data/ednet500/problem_skill_maxSkillOfProblem_number.pkl'

   please check the path of dataset, ensure the path of dataset is correct

3. AssertionError: Torch not compiled with CUDA enabled ensure your cuda is available