Grammar Guruji is an interactive web application powered by Streamlit, designed to correct grammatical errors in text using state-of-the-art NLP models. It leverages the Hugging Face Transformers library and Optimum Intel OpenVINO integration for efficient model inference.
- Sentence Splitting: Breaks down text into individual sentences for better context understanding.
- Grammar Checking: Utilizes a robust model to identify grammatical errors.
- Grammar Correction: Offers suggestions to improve the text's grammar.
- Streamlit Interface: Provides an easy-to-use web interface for real-time grammar correction.
The application uses two primary models:
- Grammar Checker:
textattack/roberta-base-CoLA- A model trained on the Corpus of Linguistic Acceptability (CoLA) for grammatical error detection. - Grammar Corrector:
pszemraj/flan-t5-large-grammar-synthesis- A model that generates grammatically correct sentences.
Both models are optimized with Intel OpenVINO for faster inference on CPU.
To set up Grammar Guruji, follow these steps:
-
Clone the repository:
git clone [https://github.com/your-username/grammar-guruji.git](https://github.com/TABREZ-96/Grammer_Guruji/)
-
Install the required packages:
To start the application, run:
streamlit run main2.py
Navigate to http://localhost:8501 in your web browser to access the Grammar Guruji interface.
- Hugging Face for the Transformers library.
- Intel for the Optimum library and OpenVINO toolkit.
- The developers and researchers behind the CoLA dataset and T5 models.
For more information on the models and their usage, please refer to the official documentation of Transformers and Optimum.