Model Description
CAMeL-Lab/text-editing-qalb14-15-zaebuc-no-insert is a text editing model tailored for grammatical error correction (GEC) in Modern Standard Arabic (MSA).
The model is based on AraBERTv02, which we fine-tuned using the QALB-2014, QALB-2015, and ZAEBUC datasets.
The concept of this model was introduced in our ACL 2025 paper, Enhancing Text Editing for Grammatical Error Correction: Arabic as a Case Study.
The model was fine-tuned to fix non-insertion (i.e., NoInsert) errors and non-punctuation (i.e., NoPnx) errors. The fine-tuning code and associated resources are publicly available on our GitHub repository: https://github.com/CAMeL-Lab/text-editing.
Intended uses
To use the CAMeL-Lab/text-editing-qalb14-15-zaebuc-no-insert model, you must clone our text editing GitHub repository and follow the installation requirements.
This model is intended to be used with Insert (CAMeL-Lab/text-editing-qalb14-15-zaebuc-insert) and Pnx (CAMeL-Lab/text-editing-qalb14-15-zaebuc-pnx) models.
How to use
Clone our text editing GitHub repository and follow the installation requirements
from transformers import BertTokenizer, BertForTokenClassification
import torch
import torch.nn.functional as F
from gec.tag import rewrite
noinsert_tokenizer = BertTokenizer.from_pretrained('CAMeL-Lab/text-editing-qalb14-15-zaebuc-no-insert')
noinsert_model = BertForTokenClassification.from_pretrained('CAMeL-Lab/text-editing-qalb14-15-zaebuc-no-insert')
insert_tokenizer = BertTokenizer.from_pretrained('CAMeL-Lab/text-editing-qalb14-15-zaebuc-insert')
insert_model = BertForTokenClassification.from_pretrained('CAMeL-Lab/text-editing-qalb14-15-zaebuc-insert')
pnx_tokenizer = BertTokenizer.from_pretrained('CAMeL-Lab/text-editing-qalb14-15-zaebuc-pnx')
pnx_model = BertForTokenClassification.from_pretrained('CAMeL-Lab/text-editing-qalb14-15-zaebuc-pnx')
def predict(model, tokenizer, text, decode_iter=1):
for _ in range(decode_iter):
tokenized_text = tokenizer(text, return_tensors="pt", is_split_into_words=True)
with torch.no_grad():
logits = model(**tokenized_text).logits
preds = F.softmax(logits.squeeze(), dim=-1)
preds = torch.argmax(preds, dim=-1).cpu().numpy()
edits = [model.config.id2label[p] for p in preds[1:-1]]
assert len(edits) == len(tokenized_text['input_ids'][0][1:-1])
subwords = tokenizer.convert_ids_to_tokens(tokenized_text['input_ids'][0][1:-1])
text = rewrite(subwords=[subwords], edits=[edits])[0][0]
return text
text = 'يجب الإهتمام ب الصحه و لا سيما ف ي الصحه النفسيه ياشباب المستقبل،،'.split()
output_sent = predict(noinsert_model, nopnx_tokenizer, text, decode_iter=2)
output_sent = predict(insert_model, nopnx_tokenizer, text, decode_iter=1)
output_sent = predict(pnx_model, pnx_tokenizer, output_sent.split(), decode_iter=1)
print(output_sent) # يجب الاهتمام بالصحة ولا سيما في الصحة النفسية يا شباب المستقبل .
Citation
@inter{alhafni-habash-2025-enhancing,
title={Enhancing Text Editing for Grammatical Error Correction: Arabic as a Case Study},
author={Bashar Alhafni and Nizar Habash},
year={2025},
eprint={2503.00985},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2503.00985},
}