README.md

---
license: mit
datasets:
- Helsinki-NLP/tatoeba_mt
- sappho192/Tatoeba-Challenge-jpn-kor
language:
- ja
- ko
pipeline_tag: translation
tags:
- python
- transformer
- pytorch
---
# Japanese to Korean translator for FFXIV

**FINAL FANTASY is a registered trademark of Square Enix Holdings Co., Ltd.**

This project is detailed on the [Github repo](https://github.com/sappho192/ffxiv-ja-ko-translator).

# Demo
[![demo.gif](demo.gif)](https://huggingface.co/spaces/sappho192/ffxiv-ja-ko-translator-demo)
[Click to try demo](https://huggingface.co/spaces/sappho192/ffxiv-ja-ko-translator-demo)

# Usage

## Inference (PyTorch)

```Python
from transformers import(
    EncoderDecoderModel,
    PreTrainedTokenizerFast,
    BertJapaneseTokenizer,
)

import torch

encoder_model_name = "cl-tohoku/bert-base-japanese-v2"
decoder_model_name = "skt/kogpt2-base-v2"

src_tokenizer = BertJapaneseTokenizer.from_pretrained(encoder_model_name)
trg_tokenizer = PreTrainedTokenizerFast.from_pretrained(decoder_model_name)

# You should change following `./best_model` to the path of model **directory**
model = EncoderDecoderModel.from_pretrained("./best_model")

text = "ギルガメッシュ討伐戦"
# text = "ギルガメッシュ討伐戦に行ってきます。一緒に行きましょうか？"

def translate(text_src):
    embeddings = src_tokenizer(text_src, return_attention_mask=False, return_token_type_ids=False, return_tensors='pt')
    embeddings = {k: v for k, v in embeddings.items()}
    output = model.generate(**embeddings)[0, 1:-1]
    text_trg = trg_tokenizer.decode(output.cpu())
    return text_trg

print(translate(text))
```

## Inference (Optimum.OnnxRuntime)
Note that current Optimum.OnnxRuntime still requires PyTorch for backend. [[Issue](https://github.com/huggingface/optimum/issues/526)]
You can use either [[ONNX](https://huggingface.co/sappho192/ffxiv-ja-ko-translator/tree/main/onnx)] or [[quantized ONNX](https://huggingface.co/sappho192/ffxiv-ja-ko-translator/tree/main/onnxq)] model.

```Python
from transformers import BertJapaneseTokenizer,PreTrainedTokenizerFast
from optimum.onnxruntime import ORTModelForSeq2SeqLM
from onnxruntime import SessionOptions
import torch

encoder_model_name = "cl-tohoku/bert-base-japanese-v2"
decoder_model_name = "skt/kogpt2-base-v2"

src_tokenizer = BertJapaneseTokenizer.from_pretrained(encoder_model_name)
trg_tokenizer = PreTrainedTokenizerFast.from_pretrained(decoder_model_name)

sess_options = SessionOptions()
sess_options.log_severity_level = 3 # mute warnings including CleanUnusedInitializersAndNodeArgs
# change subfolder to "onnxq" if you want to use the quantized model
model = ORTModelForSeq2SeqLM.from_pretrained("sappho192/ffxiv-ja-ko-translator",
        sess_options=sess_options, subfolder="onnx") 

texts = [
    "逃げろ!",  # Should be "도망쳐!"
    "初めまして.",  # "반가워요"
    "よろしくお願いします.",  # "잘 부탁드립니다."
    "ギルガメッシュ討伐戦",  # "길가메쉬 토벌전"
    "ギルガメッシュ討伐戦に行ってきます。一緒に行きましょうか？",  # "길가메쉬 토벌전에 갑니다. 같이 가실래요?"
    "夜になりました",  # "밤이 되었습니다"
    "ご飯を食べましょう."  # "음, 이제 식사도 해볼까요"
 ]


def translate(text_src):
    embeddings = src_tokenizer(text_src, return_attention_mask=False, return_token_type_ids=False, return_tensors='pt')
    print(f'Src tokens: {embeddings.data["input_ids"]}')
    embeddings = {k: v for k, v in embeddings.items()}

    output = model.generate(**embeddings, max_length=500)[0, 1:-1]
    print(f'Trg tokens: {output}')
    text_trg = trg_tokenizer.decode(output.cpu())
    return text_trg


for text in texts:
    print(translate(text))
    print()
```

## Training

Check the [training.ipynb](https://huggingface.co/sappho192/ffxiv-ja-ko-translator/blob/main/training.ipynb).