t5 efficient tiny grammar correction

Introduction

The T5-Efficient-TINY model for grammar correction is designed to correct grammatical errors in English text. It is based on the T5-Efficient-TINY architecture and was fine-tuned using a subset of the C4_200M dataset. The model introduces random typos in the input sentences to enhance error correction capabilities.

Architecture

This model is a variant of the T5 architecture, optimized for efficiency, and suitable for grammar correction tasks. It uses the PyTorch and ONNX frameworks, making it compatible with various deployment environments, including inference endpoints.

Training

The model was trained using the C4_200M dataset, which includes a wide range of English text data. Random typos were added using the nlpaug library to simulate real-world typing errors. The training was conducted as part of a project for the Full Stack Deep Learning course.

Guide: Running Locally

1. Installation: Ensure you have Python and PyTorch installed. Install the Hugging Face Transformers library with pip install transformers.
2. Download the Model: Use the Hugging Face Model Hub to get the model files.
3. Run Inference: Load the model and tokenizer using the Transformers library to perform grammar correction on input text.

• Cloud GPUs: For faster inference, consider using cloud GPU services like AWS EC2, Google Cloud Platform, or Azure.

License

The T5-Efficient-TINY Grammar Correction model is released under the MIT License, allowing for flexible use and modification.