Skip to content

Repository files navigation

Panini: a transformer-based grammatical error correction method for Bangla

Paper @ Neural Computing and Applications and Dataset @ HuggingFace

Get Started

git clone https://github.com/mehedihasanbijoy/BanglaGEC.git

or manually download and extract the github repository of BanglaGEC.

Environment Setup

Create A Virtual Environment

conda env create -f environment.yml

Activate the Environment

conda activate BanglaGEC

Prepare the BGEC Corpus

python -c "from datasets import load_dataset; load_dataset('nahid-hub/BanglaGEC').save_to_disk('./Dataset')"

or manually download the corpus from here and keep the extracted files into ./Dataset/

Training and Evaluation

Panini

python main.py --CORPUS_PATH "./Dataset/corpus.csv" --KNOWLEDGE_PATH "./KnowledgeToBeTransferred/paraphrase.pth" --CHECKPOINT_PATH "./Checkpoints/panini.pth" --MODEL_NAME "panini" --BATCH_SIZE 16 --N_EPOCHS 50

BanglaT5

python main.py --CORPUS_PATH "./Dataset/corpus.csv" --CHECKPOINT_PATH "./Checkpoints/banglat5.pth" --MODEL_NAME "banglat5" --BATCH_SIZE 16 --N_EPOCHS 50

T5-Small

python main.py --CORPUS_PATH "./Dataset/corpus.csv" --CHECKPOINT_PATH "./Checkpoints/t5small.pth" --MODEL_NAME "t5small" --BATCH_SIZE 16 --N_EPOCHS 50

BibTeX Entry and Citation Info

@article{hossain2024panini,
  title={Panini: a transformer-based grammatical error correction method for bangla},
  author={Hossain, Nahid and Bijoy, Mehedi Hasan and Islam, Salekul and Shatabda, Swakkhar},
  journal={Neural Computing and Applications},
  volume={36},
  number={7},
  pages={3463--3477},
  year={2024},
  publisher={Springer}
}

About

[Neural Computing and Applications] Panini: A Transformer Based Grammatical Error Correction Method for Bangla

Topics

Resources

Stars

1 star

Watchers

1 watching

Forks

Contributors

Languages