๐ ์ ์ฃผ์ด, ํ์ค์ด ์๋ฐฉํฅ ๋ฒ์ญ ๋ชจ๋ธ (Jeju-Standard Bidirectional Translation Model)
1. Introduction
๐งโ๐คโ๐งMember
- Bitamin 12๊ธฐ : ๊ตฌ์คํ, ์ด์ํ, ์ด์๋ฆฐ
- Bitamin 13๊ธฐ : ๊น์ค์, ๊น์ฌ๊ฒธ, ์ดํ์
Github Link
How to use this Model
- You can use this model with
transformers
to perform inference. - Below is an example of how to load the model and generate translations:
import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
## Set up the device (GPU or CPU)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
## Load the tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("Junhoee/Kobart-Jeju-translation")
model = AutoModelForSeq2SeqLM.from_pretrained("Junhoee/Kobart-Jeju-translation").to(device)
## Set up the input text
## ๋ฌธ์ฅ ์
๋ ฅ ์ ์ ๋ฐฉํฅ์ ๋ง๊ฒ [์ ์ฃผ] or [ํ์ค] ํ ํฐ์ ์
๋ ฅ ํ ๋ฌธ์ฅ ์
๋ ฅ
input_text = "[ํ์ค] ์๋
ํ์ธ์"
## Tokenize the input text
input_ids = tokenizer(input_text, return_tensors="pt", padding=True, truncation=True).input_ids.to(device)
## Generate the translation
outputs = model.generate(input_ids, max_length=64)
## Decode and print the output
decoded_output = tokenizer.decode(outputs[0], skip_special_tokens=True)
print("Model Output:", decoded_output)
Model Output: ์๋
ํ์๊ฝ
## Set up the input text
## ๋ฌธ์ฅ ์
๋ ฅ ์ ์ ๋ฐฉํฅ์ ๋ง๊ฒ [์ ์ฃผ] or [ํ์ค] ํ ํฐ์ ์
๋ ฅ ํ ๋ฌธ์ฅ ์
๋ ฅ
input_text = "[์ ์ฃผ] ์๋
ํ์๊ฝ"
## Tokenize the input text
input_ids = tokenizer(input_text, return_tensors="pt", padding=True, truncation=True).input_ids.to(device)
## Generate the translation
outputs = model.generate(input_ids, max_length=64)
## Decode and print the output
decoded_output = tokenizer.decode(outputs[0], skip_special_tokens=True)
print("Model Output:", decoded_output)
Model Output: ์๋
ํ์ธ์
Parent Model
- gogamza/kobart-base-v2
- https://huggingface.co/gogamza/kobart-base-v2
2. Dataset - ์ฝ 93๋ง ๊ฐ์ ํ
- AI-Hub (์ ์ฃผ์ด ๋ฐํ ๋ฐ์ดํฐ + ์ค๋ ์ธต ๋ฐฉ์ธ ๋ฐํ ๋ฐ์ดํฐ)
- Github (์นด์นด์ค๋ธ๋ ์ธ JIT ๋ฐ์ดํฐ)
- ๊ทธ ์ธ
- ์ ์ฃผ์ด์ฌ์ ๋ฐ์ดํฐ (์ ์ฃผ๋์ฒญ ํํ์ด์ง์์ ํฌ๋กค๋ง)
- ๊ฐ์ฌ ๋ฒ์ญ ๋ฐ์ดํฐ (๋ญ๋ญํ๋งจ ์ ํ๋ธ์์ ์ผ์ผ์ด ์์ง)
- ๋์ ๋ฐ์ดํฐ (์ ์ฃผ๋ฐฉ์ธ ๊ทธ ๋ง๊ณผ ๋ฉ, ๋ถ์๋๋ ์ง๊บผ์ ธ๋ ๋์์์ ์ผ์ผ์ด ์์ง)
- 2018๋ ๋ ์ ์ฃผ์ด ๊ตฌ์ ์๋ฃ์ง (์ผ์ผ์ด ์์ง - ํ๊ฐ์ฉ ๋ฐ์ดํฐ๋ก ์ฌ์ฉ)
3. Hyper Parameters
- Epoch : 3 epochs
- Learning Rate : 2e-5
- Weight Decay=0.01
- Batch Size : 32
4. Bleu Score
2018 ์ ์ฃผ์ด ๊ตฌ์ ์๋ฃ์ง ๋ฐ์ดํฐ ๊ธฐ์ค
- ์ ์ฃผ์ด -> ํ์ค์ด : 0.76
- ํ์ค์ด -> ์ ์ฃผ์ด : 0.5
AI-Hub ์ ์ฃผ์ด ๋ฐํ ๋ฐ์ดํฐ์ validation data ๊ธฐ์ค
- ์ ์ฃผ์ด -> ํ์ค์ด : 0.89
- ํ์ค์ด -> ์ ์ฃผ์ด : 0.77
5. CREDIT
- ๊ตฌ์คํ : kujoon13413@gmail.com
- ๊น์ค์ : 202000872@hufs.ac.kr
- ๊น์ฌ๊ฒธ : worua5667@inha.edu
- ์ด์ํ : rlaorrn0123@sookmyung.ac.kr
- ์ด์๋ฆฐ : i75631928@gmail.com
- ์ดํ์ : gudtjr3638@gmail.com
- Downloads last month
- 76
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.