Hugging Face's logo

clarin-pl
/
combo-nlp-xlm-roberta-base-arabic-padt-ud2.17

Token Classification
PyTorch
Arabic
dependency-parsing
combo
universal-dependencies
Model card Files
xet
 Community
combo-nlp-xlm-roberta-base-arabic-padt-ud2.17
File size: 3,916 Bytes
6cc5b12
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
---
license: cc-by-sa-4.0
language:
  - ar
tags:
  - dependency-parsing
  - combo
  - universal-dependencies
datasets:
  - universal_dependencies
model-name: Combo Nlp Xlm Roberta Base Arabic Padt Ud2.17
pipeline_tag: token-classification
---

# COMBO-NLP Model for Arabic

## Model Description

This is a Arabic-language model based on [COMBO-NLP](https://gitlab.clarin-pl.eu/syntactic-tools/combo-nlp), an open-source natural language preprocessing system. It performs:

- sentence segmentation (via [LAMBO](https://gitlab.clarin-pl.eu/syntactic-tools/lambo))
- tokenisation (via [LAMBO](https://gitlab.clarin-pl.eu/syntactic-tools/lambo))
- part-of-speech tagging
- morphological analysis
- lemmatisation
- dependency parsing

The Arabic model uses ``FacebookAI/xlm-roberta-base`` as its base encoder and is trained on [UD_Arabic-PADT](https://github.com/UniversalDependencies/UD_Arabic-PADT) (UD v2.17).

## Evaluation

Evaluation was performed on the UD_Arabic-PADT test split using the standard [CoNLL 2018 eval script](https://universaldependencies.org/conll18/conll18_ud_eval.py).

Two evaluation rows are reported:
- **Full-text (F1)**: raw text is segmented by [LAMBO](https://gitlab.clarin-pl.eu/syntactic-tools/lambo), then parsed and compared against gold — measures end-to-end pipeline performance including segmentation quality.
- **Aligned accuracy**: accuracy on correctly segmented (aligned) tokens — measures parsing quality on tokens that were correctly identified by the segmenter.

### Morphosyntactic Tagging

| Metric | Tokens | Sentences | Words | UPOS | XPOS | UFeats | AllTags | Lemmas |
| ------ | ------ | --------- | ----- | ---- | ---- | ------ | ------- | ------ |
| Full-text (F1) | 99.78 | 73.33 | 97.39 | 94.75 | 92.50 | 92.59 | 92.01 | 93.03 |
| Aligned accuracy | 0.00 | 0.00 | 0.00 | 97.29 | 94.98 | 95.06 | 94.47 | 95.51 |

### Dependency Parsing

| Metric | UAS | LAS | CLAS | MLAS | BLEX |
| ------ | --- | --- | ---- | ---- | ---- |
| Full-text (F1) | 83.86 | 80.76 | 78.77 | 73.01 | 74.72 |
| Aligned accuracy | 86.11 | 82.92 | 81.19 | 75.26 | 77.02 |


## Usage

Install the library from PyPI (assuming you have a virtual environment created):

```bash
pip install combo-nlp
```

Install the Lambo segmenter - only needed when passing raw text strings to COMBO:

```bash
pip install --index-url https://pypi.clarin-pl.eu/ lambo
```

```python
from combo import COMBO

# Load a pre-trained model with corresponding Lambo segmenter
nlp = COMBO("Arabic")

# Parse raw text (handles sentence splitting + tokenization)
result = nlp("الثعلب البني السريع يقفز فوق الكلب الكسول.")

# Inspect results
for sentence in result:
    for token in sentence:
        print(f"{token.form:<15} {token.lemma:<15} {token.upos:<8} head={token.head}  {token.deprel}")
```

Refer to the COMBO-NLP documentation for installation and usage instructions:

- [https://gitlab.clarin-pl.eu/syntactic-tools/combo-nlp](https://gitlab.clarin-pl.eu/syntactic-tools/combo-nlp)
- [https://gitlab.clarin-pl.eu/syntactic-tools/lambo](https://gitlab.clarin-pl.eu/syntactic-tools/lambo)

## Citation

<!--
```bibtex
@misc{combo_nlp_2026,
  title = {COMBO-NLP Models Trained on UD v2.17},
  author = {Ulewicz, Michał and Jabłońska, Maja and Klimaszewski, Mateusz and Przybyła, Piotr and Pszenny, Łukasz and Rybak, Piotr and Wiącek, Martyna and Wróblewska, Alina},
  year = {2026},
  howpublished = {\url{https://huggingface.co/collections/clarin-pl/combo-ud-217-models}}
}
```
-->

## Resources

- COMBO-NLP: [https://gitlab.clarin-pl.eu/syntactic-tools/combo-nlp](https://gitlab.clarin-pl.eu/syntactic-tools/combo-nlp)
- LAMBO: [https://gitlab.clarin-pl.eu/syntactic-tools/lambo](https://gitlab.clarin-pl.eu/syntactic-tools/lambo)
- UD_Arabic-PADT: [https://github.com/UniversalDependencies/UD_Arabic-PADT](https://github.com/UniversalDependencies/UD_Arabic-PADT)