README.md

---
license: apache-2.0
language:
- pt
- vmw
datasets:
- LIACC/Emakhuwa-Portuguese-News-MT
base_model:
- facebook/nllb-200-distilled-600M
pipeline_tag: translation
---

# CTranslate2 NLLB-200 Translation Example

This guide demonstrates how to use NLLB-finetuned model for bilingual translation between Portuguese (`por_Latn`) and a target language (`vmw_Latn`).

## Prerequisites

- Install required packages:
```bash
pip install transformers torch
```

## Inference
```python
from transformers import AutoModelForSeq2SeqLM, NllbTokenizer, AutoTokenizer
import torch

src_lang="por_Latn"
tgt_lang="vmw_Latn"
text="Olá mundo das língua!"

device = "cuda:0" if torch.cuda.is_available() else "cpu"

model_name="felerminoali/nllb200_pt_vmw_bilingual_ver1"
model = AutoModelForSeq2SeqLM.from_pretrained(model_name).to(device)
tokenizer = NllbTokenizer.from_pretrained(model_name)

tokenizer.src_lang = src_lang
tokenizer.tgt_lang = tgt_lang

inputs = tokenizer(
    text, return_tensors='pt', padding=True, truncation=True, 
    max_length=1024
)
model.eval() # turn off training mode
result = model.generate(
    **inputs.to(model.device),
    forced_bos_token_id=tokenizer.convert_tokens_to_ids(tgt_lang)
)

print(tokenizer.batch_decode(result, skip_special_tokens=True)[0])

```