cibfaye/nllb-fr-wo

Model Card for Model ID

Model Description

This model is a fine-tuned version of facebook/nllb-200-distilled-600M on the galsenai/french-wolof-translation dataset. It is designed to perform translation from French to Wolof.

Evaluation

The model was evaluated on a subset of 50 lines from the test split of the galsenai/french-wolof-translation dataset. The evaluation metric used was BLEU score, computed using the sacrebleu library.

Evaluation Results

BLEU score: 9.17

How to Use

from transformers import AutoModelForSeq2SeqLM, AutoTokenizer

model_name = "cibfaye/nllb-fr-wo"
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

def translate(text, src_lang='fra_Latn', tgt_lang='wol_Latn', a=32, b=3, max_input_length=1024, num_beams=5, **kwargs):
    tokenizer.src_lang = src_lang
    tokenizer.tgt_lang = tgt_lang
    inputs = tokenizer(text, return_tensors='pt', padding=True, truncation=True, max_length=max_input_length)
    result = model.generate(
        **inputs.to(model.device),
        forced_bos_token_id=tokenizer.convert_tokens_to_ids(tgt_lang),
        max_new_tokens=int(a + b * inputs.input_ids.shape[1]),
        num_beams=num_beams,
        **kwargs
    )
    return tokenizer.batch_decode(result, skip_special_tokens=True)

text = "Votre texte en français ici."
translation = translate(text)
print(translation)