https://github.com/lixinsu/RCZoo

question answering, reading comprehension toolkit
https://github.com/lixinsu/RCZoo

deep-learning question-answering reading-comprehension

Last synced: 3 months ago
JSON representation

question answering, reading comprehension toolkit

• Host: GitHub
• URL: https://github.com/lixinsu/RCZoo
• Owner: lixinsu
• License: mit
• Created: 2018-07-24T03:14:18.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2022-10-16T05:37:21.000Z (about 2 years ago)
• Last Synced: 2024-05-13T00:04:39.833Z (6 months ago)
• Topics: deep-learning, question-answering, reading-comprehension
• Language: Python
• Homepage:
• Size: 12.7 MB
• Stars: 167
• Watchers: 15
• Forks: 43
• Open Issues: 12
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• MRCPapers - lixinsu / RCZoo

README

        

## Introduction

The RCZoo project is a toolkit for reading comprehension model. It contains the [PyTorch](https://pytorch.org/) reimplement of multiple reading comprehension models  

## Usage

 - run `sh download.sh` to download the dataset and the glove embeddings. 

 - run `sh runs/train_squad.sh [bidaf|drqa|slqa|fusionnet|docqa]` to start the train process. (Check the xxx.sh scripts before run, as the preprocessing only need to be executed once)

 

## Dependencies

python 3.5  

Pytorch 0.4  

tqdm  

## performance

We train each model on train set for 40 epoch, and report the best performance on dev set.  

Model | Exact Match | F1 | EM(+ELMo) | F1(+ELMo)

---- | --- | --- | --- | --- 

Rnet | 69.25 | 78.97 |

BiDAF | 70.47 | 79.90 | 73.04 | 81.48

documentqa | 71.47 | 80.84 | 

DrQA | 68.39 | 77.90 |

QAnet | ... | ... |

SLQA | 67.09 | 76.67 | 

FusionNet | 68.27 | 77.79 |  

## Current progress

### [Rnet](https://www.microsoft.com/en-us/research/wp-content/uploads/2017/05/r-net.pdf)

### [BiDAF](https://arxiv.org/abs/1611.01603)

### [documentqa](https://arxiv.org/abs/1710.10723)

### [DrQA](https://arxiv.org/abs/1704.00051)

### [QAnet](https://arxiv.org/abs/1804.09541)

### [SLQA](http://aclweb.org/anthology/P18-1158)

### [FusionNet](https://openreview.net/forum?id=BJIgi_eCZ&noteId=BJIgi_eCZ)

 ## acknowledgement

 some code are borrowed from [DrQA](https://github.com/facebookresearch/DrQA.git), a cool project about reading comprehension.