great model for translation to indian languages, with restrictive license

#43
by subbur - opened

impressed it was able to tranlsate small english sentence text to indian languages( tried hindi, telugu, kannada). Impressive capabilities.
sample code for loacl use, after downloading this repository, keep this script in that folder.
latset transformers library at 28-06-2025
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained(".")
model = AutoModelForSeq2SeqLM.from_pretrained(".")

article = "paracetmol 1-0-1 after meals"
inputs = tokenizer(article, return_tensors="pt")
#english to telugu conversion
translated_tokens = model.generate(
**inputs, forced_bos_token_id=tokenizer.convert_tokens_to_ids("tel_Telu"), max_length=30
)
#kan_Knda, hin_Deva
tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0]

if any one know if instruction can be added to this model, please give me some direction on that.

subbur changed discussion title from great model for translation to indian languages to great model for translation to indian languages, with restrictive license

Sign up or log in to comment