Views
No views yet
CAMeL-Lab/text-editing-qalb14-nopnx 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 dataset.
This model was introduced in our ACL 2025 paper, Enhancing Text Editing for Grammatical Error Correction: Arabic as a Case Study, where we refer to it as SWEET (Subword Edit Error Tagger).CAMeL-Lab/text-editing-qalb14-nopnx model, you must clone our text editing GitHub repository and follow the installation requirements.
We used this SWEETNoPnx model to report results on the QALB-2014 dev and test sets in our paper.
This model is intended to be used with SWEETPnx (CAMeL-Lab/text-editing-qalb14-pnx) model.1from transformers import BertTokenizer, BertForTokenClassification
2import torch
3import torch.nn.functional as F
4from gec.tag import rewrite
5
6
7nopnx_tokenizer = BertTokenizer.from_pretrained('CAMeL-Lab/text-editing-qalb14-nopnx')
8nopnx_model = BertForTokenClassification.from_pretrained('CAMeL-Lab/text-editing-qalb14-nopnx')
9
10pnx_tokenizer = BertTokenizer.from_pretrained('CAMeL-Lab/text-editing-qalb14-pnx')
11pnx_model = BertForTokenClassification.from_pretrained('CAMeL-Lab/text-editing-qalb14-pnx')
12
13
14def predict(model, tokenizer, text, decode_iter=1):
15 for _ in range(decode_iter):
16 tokenized_text = tokenizer(text, return_tensors="pt", is_split_into_words=True)
17 with torch.no_grad():
18 logits = model(**tokenized_text).logits
19 preds = F.softmax(logits.squeeze(), dim=-1)
20 preds = torch.argmax(preds, dim=-1).cpu().numpy()
21 edits = [model.config.id2label[p] for p in preds[1:-1]]
22
23 assert len(edits) == len(tokenized_text['input_ids'][0][1:-1])
24 subwords = tokenizer.convert_ids_to_tokens(tokenized_text['input_ids'][0][1:-1])
25 text = rewrite(subwords=[subwords], edits=[edits])[0][0]
26 return text
27
28
29text = 'يجب الإهتمام ب الصحه و لا سيما ف ي الصحه النفسيه ياشباب المستقبل،،'.split()
30
31output_sent = predict(nopnx_model, nopnx_tokenizer, text, decode_iter=2)
32output_sent = predict(pnx_model, pnx_tokenizer, output_sent.split(), decode_iter=1)
33print(output_sent) # يجب الاهتمام بالصحة ولا سيما في الصحة النفسية يا شباب المستقبل .
341@inter{alhafni-habash-2025-enhancing,
2 title={Enhancing Text Editing for Grammatical Error Correction: Arabic as a Case Study},
3 author={Bashar Alhafni and Nizar Habash},
4 year={2025},
5 eprint={2503.00985},
6 archivePrefix={arXiv},
7 primaryClass={cs.CL},
8 url={https://arxiv.org/abs/2503.00985},
9}