# MHKA

**Repository Path**: hawksilent/MHKA

## Basic Information

- **Project Name**: MHKA
- **Description**: The corresponding code from our paper "Social Commonsense Reasoning with Multi-Head Knowledge Attention (EMNLP 2020)". Do not hesitate to open an issue if you run into any trouble!
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-12-26
- **Last Updated**: 2021-12-26

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Social-Commonsense-Reasoning-Multi-Head-Knowledge-Attention
![](https://img.shields.io/github/last-commit/Heidelberg-NLP/MHKA?color=blue) 


Paper Link: https://www.aclweb.org/anthology/2020.findings-emnlp.267.pdf

This directory contains the following parts of the 'Social Commonsense Reasoning Multi-Head Knowledge Attention' experiment. 

<p align="center">
  <img src="model_emlp.png" alt="MHKA">
</p>

## Reference

If you make use of the contents of this repository, please cite [the following paper](https://www.aclweb.org/anthology/N19-1368):

```bib
@inproceedings{paul-frank-2020-mhka,
    title = "Social Commonsense Reasoning with Multi-Head Knowledge Attention",
    author = "Paul, Debjit  and Frank, Anette",
    booktitle = In Findings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP),
    month = November,
    year = 2020,
    publisher = "Association for Computational Linguistics"
}
```
## Requirements 
~~~~
- python3.8+
- pytorch
~~~~
Install the library and dependencies
~~~~
git clone https://github.com/huggingface/transformers
cd transformers
pip install .
pip install -r ./examples/requirements.txt
pip install tensorflow
pip install ftfy==5.1
conda install -c conda-forge spacy
python -m spacy download en
pip install tensorboardX
pip install tqdm
pip install pandas
pip install ipython
~~~~

## Recreating Results
<p align="center">
  <img src="results_mhka.png" alt="Results">
</p>