Metadata-Version: 2.4
Name: ne-lid
Version: 1.0.0
Summary: Language identification for Northeast Indian languages
Author-email: MWire Labs <connect@mwirelabs.com>
Project-URL: Homepage, https://mwirelabs.com
Project-URL: HuggingFace, https://huggingface.co/MWirelabs/ne-lid
Keywords: language-identification,nlp,northeast-india,low-resource,fasttext
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Classifier: License :: Free To Use But Restricted
Requires-Python: >=3.8
Description-Content-Type: text/markdown
Requires-Dist: fasttext-wheel>=0.9.2
Requires-Dist: huggingface-hub>=0.20.0

# ne-lid

Language identification for Northeast Indian languages.

**11 languages · 99.09% accuracy · fastText · CC-BY-4.0**

## Install

```bash
pip install ne-lid
```

## Usage

```python
from ne_lid import NELID

model = NELID()  # downloads from HuggingFace on first run

# Single prediction
result = model.predict('Ki paidbah shnong ki la ia shim bynta')
print(result)  # {'lang': 'kha', 'score': 0.9999}

# Top-3 predictions
results = model.predict('Ka sngi ka lieh', k=3)
print(results)

# List supported languages
model.languages()
```

## Supported Languages

| Code | Language | Accuracy |
|------|----------|----------|
| asm | Assamese | 100% |
| brx | Bodo | 99% |
| eng | English | 98% |
| grt | Garo | 100% |
| hin | Hindi | 97% |
| kha | Khasi | 99% |
| trp | Kokborok | 100% |
| mni | Meitei | 100% |
| lus | Mizo | 99% |
| nag | Nagamese | 100% |
| njz | Nyishi | 99% |

Overall test accuracy: **99.09%**

## Links
- [HuggingFace](https://huggingface.co/MWirelabs/ne-lid)
- [MWire Labs](https://mwirelabs.com)
