---
library_name: transformers
tags:
- toxicity
- hindi
license: cc
datasets:
- Polygl0t/hindi-toxicity-qwen-annotations
language:
- hi
metrics:
- precision
- recall
- accuracy
model-index:
- name: hindi-roberta-toxicity-classifier
  results: []
pipeline_tag: text-classification
base_model:
- l3cube-pune/hindi-roberta
---
# hindi-roberta Toxicity Classifier


hindi-roberta-toxicity-classifier is an [HindRoBERTa](https://huggingface.co/l3cube-pune/hindi-roberta) based model that can be used for judging the toxicity level of a given Hindi text string. This model was trained on the [Polygl0t/hindi-toxicity-qwen-annotations](https://huggingface.co/datasets/Polygl0t/hindi-toxicity-qwen-annotations) dataset.

## Details

For training,  we added a classification head with a single regression output to [l3cube-pune/hindi-roberta](https://huggingface.co/l3cube-pune/hindi-roberta). Only the classification head was trained, i.e., the rest of the model was frozen.

- **Dataset:** [hindi-toxicity-qwen-annotations](https://huggingface.co/datasets/Polygl0t/hindi-toxicity-qwen-annotations)
- **Language:** Hindi
- **Number of Training Epochs:** 20
- **Batch size:** 256
- **Optimizer:** `torch.optim.AdamW`
- **Learning Rate:** 3e-4
- **Eval Metric:** `f1-score`

This repository has the [source code](https://github.com/Polygl0t/llm-foundry) used to train this model.

### Evaluation Results

#### Confusion Matrix

|       | **1** | **2** | **3** | **4** | **5** |
|-------|-------|-------|-------|-------|-------|
| **1** | 11526 | 2601  | 134   | 7     | 0     |
| **2** | 722   | 1713  | 281   | 10    | 0     |
| **3** | 240   | 1092  | 590   | 7     | 2     |
| **4** | 21    | 242   | 308   | 104   | 13    |
| **5** | 5     | 46    | 78    | 68    | 123   |

- Precision: 0.58656
- Recall: 0.45341
- F1 Macro: 0.47433
- Accuracy: 0.7028

## Usage


Here's an example of how to use the Toxicity Classifier:

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

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

tokenizer = AutoTokenizer.from_pretrained("Polygl0t/hindi-roberta-toxicity-classifier")
model = AutoModelForSequenceClassification.from_pretrained("Polygl0t/hindi-roberta-toxicity-classifier")
model.to(device)

text = "यह एक उदाहरण है।"
encoded_input  =  tokenizer(text, return_tensors="pt", padding="longest", truncation=True).to(device)

with  torch.no_grad():
	model_output  =  model(**encoded_input)
	logits  =  model_output.logits.squeeze(-1).float().cpu().numpy()

# scores are produced in the range [0, 4]. To convert to the range [1, 5], we can simply add 1 to the score.
score = [x + 1 for x in logits.tolist()][0]

print({
 "text": text,
 "score": score,
 "int_score": [int(round(max(0, min(score, 4)))) + 1 for score in logits][0],
})
```


## Cite as 🤗

```latex
@misc{shiza2026lilmoo,
      title={{Raising Bars, Not Parameters: LilMoo Compact Language Model for Hindi}}, 
      author={Shiza Fatimah and Aniket Sen and Sophia Falk and Florian Mai and Lucie Flek and Nicholas Kluge Corr{\^e}a},
      year={2026},
      eprint={2603.03508},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2603.03508}, 
}
```

## Aknowlegments

Polyglot is a project funded by the Federal Ministry of Education and Research (BMBF) and the Ministry of Culture and Science of the State of North Rhine-Westphalia (MWK) as part of TRA Sustainable Futures (University of Bonn) and the Excellence Strategy of the federal and state governments.

We also gratefully acknowledge the granted access to the [Marvin cluster](https://www.hpc.uni-bonn.de/en/systems/marvin) hosted by [University of Bonn](https://www.uni-bonn.de/en) along with the support provided by its High Performance Computing & Analytics Lab.

## License

According to [l3cube-pune/hindi-roberta](https://huggingface.co/l3cube-pune/hindi-roberta), the model is released under [cc-by-4.0](https://spdx.org/licenses/CC-BY-4.0). For any queries, please get in touch with the authors of the original paper tied to [hindi-roberta](https://huggingface.co/l3cube-pune).