PROTECT: Parameter-Efficient Tuning for Few-Shot Robust Chinese Text Correction

This repository contains code, model, datasets for PROTECT

Pretrain

dataset

Pretrain dataset available at here and Huggingface

cd scripts
sh pretrain.sh

pretrained model and datasets

To facilitate replication, we provide pre-trained PROTECT model and datasets.

pre-trained PROTECT model available at Huggingface

datasets available at here

cd scripts
sh train.sh