---
license: cc
language:
- ve
- tn
- sot
- nso
metrics:
- perplexity
library_name: transformers
tags:
- tshivenda
- sotho
- south africa
- low-resource
- bantu
- xlm-roberta
widget:
- text: Rabulasi wa <mask> u khou bvelela nga u lima
- text: >-
    Vhana vhane vha kha ḓi bva u bebwa vha kha khombo ya u <mask> nga
    Listeriosis
---

# Zabantu - Tshivenda & Sotho family

This is a variant of [Zabantu](https://huggingface.co/dsfsi/zabantu-bantu-250m) pre-trained on a multilingual dataset of Tshivenda(ven) and Sotho family(Northern Sotho, Southern Sotho, Setswana) sentences on a 
transformer network with 170 million traininable parameters.


# Usage Example(s)

```python
from transformers import pipeline
# Initialize the pipeline for masked language model
unmasker = pipeline('fill-mask', model='dsfsi/zabantu-sot-ven-170m')

sample_sentences = ["Rabulasi wa <mask> u khou bvelela nga u lima",
                    "Vhana vhane vha kha ḓi bva u bebwa vha kha khombo ya u <mask> nga Listeriosis"]

# Perform the fill-mask task
results = unmasker(sentence)
# Display the results
for result in results:
    print(f"Predicted word: {result['token_str']} - Score: {result['score']}")
    print(f"Full sentence: {result['sequence']}\n")
    print("=" * 80)

```