README.md

metadata

language:
  - vi
  - ba
tags:
  - translation
datasets:
  - custom dataset
metrics:
  - bleu
  - sacrebleu

How to run the model

from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer

model = M2M100ForConditionalGeneration.from_pretrained("transZ/M2M_Vi_Ba")
tokenizer = M2M100Tokenizer.from_pretrained("transZ/M2M_Vi_Ba")
tokenizer.src_lang = "vi"
vi_text = "Hôm nay ba đi chợ."
encoded_vi = tokenizer(vi_text, return_tensors="pt")
generated_tokens = model.generate(**encoded_vi, forced_bos_token_id=tokenizer.get_lang_id("ba"))
translate = tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)[0]
print(translate)