Multi-Relational Graph Transformer for Automatic Short Answer Grading (MitiGaTe)

This repository is the official implementation of the paper:

"Multi-Relational Graph Transformer for Automatic Short Answer Grading": Rajat Agarwal, Varun Khurana, Karish Grover, Mukesh Mohania, Vikram Goyal. (NAACL 2022)

Stable version will be released soon. Stay tuned!

Model Architecture

Requirements

Use the environment.yml file to set up the conda environment.

$ conda env create -n ENVNAME --file environment.yml

Training

To train the model, run the following command:

$ python main.py --gpu_id <gpu id> --config 'configs/graph_transformer_sparse.json' --L <layers in graph transformer> --out_dim 32 --hidden_dim 32 --n_heads <attention heads> --epochs <number of epochs for training>

Results