language: | |
- hi | |
- en | |
- multilingual | |
license: mit | |
tags: | |
- codeswitching | |
- hindi-english | |
- language-identification | |
datasets: | |
- lince | |
# codeswitch-hineng-lid-lince | |
This is a pretrained model for **language identification** of `hindi-english` code-mixed data used from [LinCE](https://ritual.uh.edu/lince/home) | |
This model is trained for this below repository. | |
[https://github.com/sagorbrur/codeswitch](https://github.com/sagorbrur/codeswitch) | |
To install codeswitch: | |
``` | |
pip install codeswitch | |
``` | |
## Identify Language | |
* **Method-1** | |
```py | |
from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline | |
tokenizer = AutoTokenizer.from_pretrained("sagorsarker/codeswitch-hineng-lid-lince") | |
model = AutoModelForTokenClassification.from_pretrained("sagorsarker/codeswitch-hineng-lid-lince") | |
lid_model = pipeline('ner', model=model, tokenizer=tokenizer) | |
lid_model("put any hindi english code-mixed sentence") | |
``` | |
* **Method-2** | |
```py | |
from codeswitch.codeswitch import LanguageIdentification | |
lid = LanguageIdentification('hin-eng') | |
text = "" # your code-mixed sentence | |
result = lid.identify(text) | |
print(result) | |
``` | |