Grammar Correction Web App

Overview

This project implements a web-based tool for correcting grammatical errors in text using the pretrained vennify/t5-base-grammar-correction model. The tool allows users to input text, which is then processed by the model to correct grammar and provide a count of corrected words.

Uses of the Model

The primary use case for this model is to enhance the grammatical correctness of input text. It serves as a valuable tool for content creators, writers, and individuals seeking to improve the quality of written content. The model is particularly useful in applications where clear and error-free communication is essential, such as in document preparation, content editing, and educational materials.

Project Structure

• app.py: Flask application for the web interface.
• model.py: Contains the correct_grammar function for grammar correction.
• templates/index.html: HTML template for the input form.
• templates/result.html: HTML template for displaying the correction result.

How to use

1. Clone the repository git clone https://github.com/your-username/grammar-correction-tool.git.
2. Install the dependencies pip install -r requirements.txt.
3. Run the Flask application python app.py.

Functionality

• The correct_grammar function in model.py takes input text and uses the vennify/t5-base-grammar-correction model to correct grammar.It returns the corrected text, a list of corrected words, and the count of corrected words.
• The main functionality of the model.py includes:
  • Initializing the T5 grammar correction model (vennify/t5-base-grammar-correction) using HappyTransformer.
  • Defining the correct_grammar function, which takes input text, the initialized model, and beam settings as arguments.
  • Generating corrected text from the grammar model using the generate_text method.
  • Extracting original and corrected text, calculating differences between them, and extracting corrected words.
  • Returning the corrected text, corrected words, and count of corrected words.

Usage

from transformers import pipeline
# Load the Grammar Correction T5 Model from Hugging Face
grammar_correction_model = pipeline(task="text2text-generation", model="hassaanik/grammar-correction-model")
# Input text with grammatical errors
input_text = "They is going to spent time together."
# Get corrected output and details
result = grammar_correction_model(input_text, max_length=200, num_beams=5, no_repeat_ngram_size=2)
# Print the corrected output
print(result)