Grammar Checker with LSTM

This project is a simple implementation of a Grammar Checker using a Long Short-Term Memory (LSTM) neural network model. It uses PyTorch for deep learning, and the goal is to classify whether the individual words in a sentence are grammatically correct or incorrect. This is compatible with VSCode, or you can directly run in terminal :)

Project Structure

• data/: Contains all training data and example sentences. Play around and edit them to your liking (just make sure they are reasonably similar to the vocabulary in the test set)!
• src/: Contains the core logic including the LSTM model, data preprocessing, training, and saving the model. If you want to train the model, you have to run the train.py
• tests/: Contains a testing script that loads the trained model, runs predictions, and prints the results.
• models/: Contains the model of the Grammar Checker using the LSTM model

Requirements

1. This project requires Python 3. You can install the required dependencies by running:

pip install -r requirements.txt

2. Prepare your text data in the data/sentences.txt file. I have prefilled 30 sentences, and also test sentences.

3. To train the model, run:

python src/train.py

4. The model will be saved to the models/grammar_checker_model.pth directory.

5. In tests/ you can simply run the code after training the model and generating the aforementioned file.