Environment

This project is developed using python 3.7，Pytorch1.4.0, CUDA 10.2 on NVIDIA Titan RTX GPU. You'd better configure the environment as this.

Quick Start

1. Clone the repo:

git clone git@github.com:Badstu/CAKT.git
cd CAKT

2. Install dependencies:

Configure python, pytorch and CUDA enviroments, and then

pip install -r requirements.txt

3. Dataset

You can find dataset at dataset folder, there are five datasets used in this project.

4. Quick run

You can run our CAKT model with main.py.

python main.py

when you get this information, it means that you run it successfully.

• if you want to run our main CAKT model, you can use run_CAKT() function, and you can easily modify some parameters, such as k_frames(k), H and batch_size(b) to do some experiments, for example,

  k_frames: [4, 8, 16, 32]
  H: [11, 13, 15, 17, 19]
  batch_size: [32, 48, 64, 80]
  

• if you want to run our model on different datasets, you can use run_five_dataset() function, we provide five benchmark datasets as follows:

  dataset_name, knowledge_length
  "assist2009", 110,
  "assist2015", 100,
  "assist2017", 102,
  "statics", 1223,
  "synthetic", 50
  

• if you want to run ablation model of CAKT, we provide run_ablation() function, you can set model_name="CAKT_ablation" and set ablation equal to ablation mode:

  ablation_mode: ["LSTM_RECENT", "FC_POOLING", "FC_REAR", "WEIGHT_SUM", "NO_EXP_DECAY"]
  

Please feel free to contact me by email to me or just leave a issue if you have any question.