---
license: cc-by-nc-4.0
language:
- en
- si
pipeline_tag: translation
datasets:
- zaanind/sinhala_englsih_parrel_corpus
- zaanind/sinhala_englsih_nmt

inference:
  parameters:
    src_lang: "eng_Latn"
    tgt_lang: "sin_Sinh"


widget:
- text: "you will receive a notification when your order is ready for pickup"
  example_title: "example 1"

- text: "you will receive a response to your inquiry within 24 hours"
  example_title: "example 2"
  
- text: "i'm glad i could make it to your birthday event it was such a memorable experience"
  example_title: "example 3"



---


# What Is This?
It is a [NLLB-200-600M](https://huggingface.co/facebook/nllb-200-distilled-600M) model 
fine-tuned for translating between englih and sinhala languages

Training & Inference Codes - [https://github.com/zaanind/NLLB-200-Sinhala](https://github.com/zaanind/NLLB-200-Sinhala)


# Try It
- [Try it at our website](https://zaanind.pythonanywhere.com)

# How to use the model:

1.Install necessary libraries

```
pip install  requests sentencepiece transformers==4.33 sacremoses  sacrebleu
```

2.Translate! 

```
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
import torch


model_name = "zaanind/nllb-ensi-v1-tuning"  #download model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)


def dotranslate(text): #define function for generation
  inputs = tokenizer(text, return_tensors="pt")
  translated_tokens = model.generate(**inputs, forced_bos_token_id=tokenizer.lang_code_to_id["sin_Sinh"])
  out = tokenizer.decode(translated_tokens[0], skip_special_tokens=True)

  return out


dotranslate("hello how are you?") #translate
```


# Contact  - [https://t.me/zaanind](https://t.me/zaanind)