You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Eng-to-Korean dialect translation(5 different dialect)
We trained our model using kor-dialect corpus data downloaded from AIHUB(aihub.or.kr), and paired with their english corpus by translation using hugging face NMT checkpoint.
Number of data pair for training : kangwon 300k, jeju 200k, jd 170k, cc 110k, gs 110k.
How to use :
download checkpoint -> please email me if you want to try(seuyon0101@gmail.com).
python main.py
insert tokenizer type
vocab size small? Yes(8k vocab) , No(16k vocab)
add region tag before input text
Example :
basic transformer model inference :
About
english to multiple korean dialect neural translation model