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)

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