---
library_name: transformers
language:
- en
- fr
- it
- es
- ru
- uk
- tt
- ar
- hi
- ja
- zh
- he
- am
- de
license: openrail++
datasets:
- textdetox/multilingual_toxicity_dataset
metrics:
- f1
base_model:
- FacebookAI/xlm-roberta-large
pipeline_tag: text-classification
---

## Multilingual Toxicity Classifier for 15 Languages (2025)

This is an instance of [xlm-roberta-large](https://huggingface.co/FacebookAI/xlm-roberta-large) that was fine-tuned on binary toxicity classification task based on our updated (2025) dataset [textdetox/multilingual_toxicity_dataset](https://huggingface.co/datasets/textdetox/multilingual_toxicity_dataset).

Now, the models covers 15 languages from various language families:

| Language  | Code | F1 Score |
|-----------|------|---------|
| English   | en   | 0.9225  |
| Russian   | ru   | 0.9525  |
| Ukrainian | uk   | 0.96    |
| German    | de   | 0.7325  |
| Spanish   | es   | 0.7125  |
| Arabic    | ar   | 0.6625  |
| Amharic   | am   | 0.5575  |
| Hindi     | hi   | 0.9725  |
| Chinese   | zh   | 0.9175  |
| Italian   | it   | 0.5864  |
| French    | fr   | 0.9235  |
| Hinglish  | hin  | 0.61    |
| Hebrew    | he   | 0.8775  |
| Japanese  | ja   | 0.8773  |
| Tatar     | tt   | 0.5744  |

## How to use

```python
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained('textdetox/xlmr-large-toxicity-classifier-v2')
model = AutoModelForSequenceClassification.from_pretrained('textdetox/xlmr-large-toxicity-classifier-v2')

batch = tokenizer.encode("You are amazing!", return_tensors="pt")

output = model(batch)
# idx 0 for neutral, idx 1 for toxic
```

## Citation
The model is prepared for [TextDetox 2025 Shared Task](https://pan.webis.de/clef25/pan25-web/text-detoxification.html) evaluation.

Citation TBD soon.