stephen-cheng / Spelling_Correction_with_Soft-Masked_BERT

Soft-Masked Bert Model

• Reference Paper: Spelling Error Correction with Soft-Masked BERT

• Dataset: The data that we will use for this project will be 20 popular books from Project Gutenberg.

Prerequired packages

pip install -r requirements.txt

Parameters:

The length of each sentence is between 4 and 200. So,

• max_len = 32
• min_len = 2

How to run?

• Prepare Data:

python data_prepare.py

• Process Data:

python data_process.py

• Train Models:

python train.py

• Test Models:

python test.py