README.md

---
datasets:
- grammarly/coedit
- Owishiboo/grammar-correction
language:
- en
base_model:
- google-t5/t5-base
pipeline_tag: text-generation
---
# English Grammar Correction model

# Quick Start (Python)

### Installation
```bash
pip install transformers torch
```

### Basic Usage
```python
from transformers import T5Tokenizer, T5ForConditionalGeneration
import torch

# Load model and tokenizer
model_name = "yoon-eunbin/t5-gec-model"
tokenizer = T5Tokenizer.from_pretrained(model_name)
model = T5ForConditionalGeneration.from_pretrained(model_name)

# Set device
device = 'cuda' if torch.cuda.is_available() else 'cpu'
model = model.to(device)

# Prepare input
text = "He has left the room when I came into the room."
inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True, max_length=64).to(device)

# Generate correction
outputs = model.generate(**inputs, max_length=64)

# Decode output
corrected_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(f"Original: {text}")
print(f"Corrected: {corrected_text}")