https://github.com/nusnlp/ugec

The official code for the "Unsupervised Grammatical Error Correction Rivaling Supervised Methods" paper, published in EMNLP 2023.
https://github.com/nusnlp/ugec

Last synced: 17 days ago
JSON representation

The official code for the "Unsupervised Grammatical Error Correction Rivaling Supervised Methods" paper, published in EMNLP 2023.

• Host: GitHub
• URL: https://github.com/nusnlp/ugec
• Owner: nusnlp
• Created: 2023-10-10T12:39:25.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-12-08T03:09:01.000Z (about 2 months ago)
• Last Synced: 2024-12-08T04:17:41.785Z (about 2 months ago)
• Language: Python
• Homepage: https://aclanthology.org/2023.emnlp-main.185.pdf
• Size: 5.18 MB
• Stars: 2
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Unsupervised Grammatical Error Correction Rivaling Supervised Methods

> Hannan Cao, Liping Yuan, Yuchen Zhang, Hwee Tou Ng. [Unsupervised Grammatical Error Correction Rivaling Supervised Methods](https://aclanthology.org/2023.emnlp-main.185.pdf). In EMNLP 2023. 

## Training Data & Checkpoints

[GEC training data](https://drive.google.com/drive/folders/1c1xNjD7ORGaY9P3vuy1w-q_f60WkGC_D?usp=sharing);

[GEC model checkpoints](https://drive.google.com/drive/folders/1TZNbuEwjifTVqKldfpkXl264CLkHNlXw?usp=sharing);

## English GEC

### Flan-T5-xxl

1. Please store all the downloaded checkpoint and data for Flan-T5-xxl in this folder: en_flan_t5/llm_finetune

2. Install the requirement.txt inside en_flan_t5 folder

Train:

```

bash train.sh

```

Inference: go to en_flan_t5/llm_inference folder

```

bash eval_gec.sh your/ckpt/name

```

### BART-base

1. Please store all the downloaded checkpoint and data for BART-base in this folder: en_fairseq_train

2. Install the requirement.txt inside en_fairseq_train folder

Train:

```

cd gec

bash train.sh path/to/the/model/to/be/restored path/to/data-bin/folder output_path

```

Inference:

```

bash new_generate.sh path/to/model/ckpt testing/input/path

```

## Chinese GEC

1. Please store all the downloaded checkpoint and data for BART-base in this folder: chinese_bart_large

2. Install the requirement.txt inside chinese_bart_large folder

Train:

```

cd gec

bash train_ch.sh

```

Inference:

```

cd gec

bash test_ch.sh

```

## Citation

If you found our paper or code useful, please cite as:

```

@inproceedings{cao-etal-2023-unsupervised,

    title = "Unsupervised Grammatical Error Correction Rivaling Supervised Methods",

    author = "Cao, Hannan  and

      Yuan, Liping  and

      Zhang, Yuchen  and

      Ng, Hwee Tou",

    editor = "Bouamor, Houda  and

      Pino, Juan  and

      Bali, Kalika",

    booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",

    month = dec,

    year = "2023",

    address = "Singapore",

    publisher = "Association for Computational Linguistics",

    url = "https://aclanthology.org/2023.emnlp-main.185",

    doi = "10.18653/v1/2023.emnlp-main.185",

    pages = "3072--3088",

}

```

If you encounter any problem with the code, please contact [email protected] .