Et5 Typos Corrector

🚀 Korean Typos Corrector

This is a Korean colloquialism-specific typos corrector fine-tuned based on the ETRI-et5 model. It effectively corrects typos in Korean colloquial language, providing accurate text correction services.

🚀 Quick Start

Base on PLM model(ET5)

• ETRI(https://aiopen.etri.re.kr/et5Model)

Base on Dataset

• Modu 말뭉치(https://corpus.korean.go.kr/request/reausetMain.do?lang=ko) typos correction data

Data Preprocessing

• 1. Remove special characters (comma) and period.
• 2. Remove null values("").
• 3. Remove sentences that are too short (length less than or equal to 2).
• 4. Remove words containing name tags such as &name&, name1 in the sentence (only remove the words and keep the sentence).
• Total: 318,882 pairs

💻 Usage Examples

Basic Usage

from transformers import T5ForConditionalGeneration, T5Tokenizer

# Load the T5 model
model = T5ForConditionalGeneration.from_pretrained("j5ng/et5-typos-corrector")
tokenizer = T5Tokenizer.from_pretrained("j5ng/et5-typos-corrector")

device = "cuda:0" if torch.cuda.is_available() else "cpu"
# device = "mps:0" if torch.cuda.is_available() else "cpu" # for mac m1

model = model.to(device) 

# Example input sentence
input_text = "아늬 진짜 무ㅓ하냐고"

# Encode the input sentence
input_encoding = tokenizer("맞춤법을 고쳐주세요: " + input_text, return_tensors="pt")

input_ids = input_encoding.input_ids.to(device)
attention_mask = input_encoding.attention_mask.to(device)

# Generate the output of the T5 model
output_encoding = model.generate(
    input_ids=input_ids,
    attention_mask=attention_mask,
    max_length=128,
    num_beams=5,
    early_stopping=True,
)

# Decode the output sentence
output_text = tokenizer.decode(output_encoding[0], skip_special_tokens=True)

# Print the result
print(output_text) # 아니 진짜 뭐 하냐고.

Advanced Usage

from transformers import T5ForConditionalGeneration, T5Tokenizer, pipeline

model = T5ForConditionalGeneration.from_pretrained('j5ng/et5-typos-corrector')
tokenizer = T5Tokenizer.from_pretrained('j5ng/et5-typos-corrector')

typos_corrector = pipeline(
    "text2text-generation",
    model=model,
    tokenizer=tokenizer,
    device=0 if torch.cuda.is_available() else -1,
    framework="pt",
)

input_text = "완죤 어이업ㅅ네진쨬ㅋㅋㅋ"
output_text = typos_corrector("맞춤법을 고쳐주세요: " + input_text,
            max_length=128,
            num_beams=5,
            early_stopping=True)[0]['generated_text']

print(output_text) # 완전 어이없네 진짜 ᄏᄏᄏᄏ.

📄 License

This project is licensed under the Apache-2.0 license.