README.md

---
license: apache-2.0
datasets:
- CodeTed/CGEDit_dataset
language:
- zh
metrics:
- accuracy
library_name: transformers
tags:
- CGED
- CSC
pipeline_tag: text2text-generation
---
# CGEDit - Chinese Grammatical Error Diagnosis by Task-Specific Instruction Tuning

Try the model from this space "[Chinese Grammarly](https://huggingface.co/spaces/CodeTed/Chinese-Grammarly)".

This model was obtained by fine-tuning the corresponding `ClueAI/PromptCLUE-base-v1-5` model on the CoEdIT dataset. 
![CGEDit_model.png](https://cdn-uploads.huggingface.co/production/uploads/64c7473f513a7fa7c32e153b/AtlsZUWz86rKyb_9EWlDa.png)


## Model Details
### Model Description
- Language(s) (NLP): `Chinese`
- Finetuned from model: `ClueAI/PromptCLUE-base-v1-5`
### Model Sources
- Repository: [https://github.com/TedYeh/Chinese_spelling_Correction](https://github.com/TedYeh/Chinese_spelling_Correction)

## Usage
```python
from transformers import AutoTokenizer, T5ForConditionalGeneration

tokenizer = AutoTokenizer.from_pretrained("CodeTed/Chinese_Grammarly")
model = T5ForConditionalGeneration.from_pretrained("CodeTed/Chinese_Grammarly")
input_text = '糾正句子裡的錯字: 看完那段文張，我是反對的！'
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids, max_length=256)
edited_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
```