You can deploy this model and start earning money today!

agnivamaiti/
nllb-200-en-nagamese

Name: nllb-200-en-nagamese
Availability: InStock

agnivamaiti/nagamese-english-mt

cc-by-nc-4.0

Views

No views yet

NLLB-200 English to Nagamese Translation

This model is a fine-tuned version of facebook/nllb-200-distilled-600M for translating English to Nagamese.

How to Use

You can use this model directly with the transformers library. Because Nagamese is not natively supported by NLLB, a surrogate language token (e.g., asm_Latn or ind_Latn) was used during training to represent Nagamese.

Python Inference Script

python

import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# 1. Load the model and tokenizer
model_name = "agnivamaiti/nllb-200-en-nagamese"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

# 2. Prepare your text
text = "Where are you going tomorrow?"
inputs = tokenizer(text, return_tensors="pt")

# 3. Generate translation (Force the surrogate token ID)
# Note: Change 'asm_Latn' if you used a different surrogate token during training (like 'ind_Latn' or 'tur_Latn').
outputs = model.generate(
    **inputs,
    forced_bos_token_id=tokenizer.convert_tokens_to_ids("asm_Latn"),
    max_length=128
)

# 4. Decode the output
translation = tokenizer.decode(outputs[0], skip_special_tokens=True)
print("Translation:", translation)

Training Data

The model was fine-tuned on a custom English-Nagamese parallel corpus. You can find the dataset here: agnivamaiti/nagamese-english-mt.

Additional Scripts

We have included train.py and inference.py in the files of this repository for documentation and reproducibility purposes.