File size: 1,705 Bytes
dc46a0d ea8dfae dc46a0d ea8dfae cbbc4b6 ea8dfae dc46a0d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 |
---
title: TrueWrite Scan:- AI-Powered Writing Assistant
emoji: 🚀
colorFrom: red
colorTo: pink
sdk: static
pinned: false
license: mit
---
TrueWrite Scan is a full-stack AI application designed to enhance writing integrity. It integrates state-of-the-art NLP models to provide real-time Grammar Checking, Plagiarism Detection, and AI Content Classification.
🌟 Key Features
📝 Advanced Grammar Checker: Utilizes language-tool-python and custom rule sets to detect syntax and stylistic errors.
🔍 Semantic Plagiarism Detection: Powered by sentence-transformers to detect paraphrased content, not just exact matches.
🤖 AI Content Detector: Implements Hugging Face Transformers to analyze perplexity and burstiness, identifying machine-generated text.
📄 Multi-Format Support: Parses text from .txt, .docx, and .pdf files.
🛠️ Tech Stack
Frontend
Framework: Next.js 13+ (React)
Styling: Tailwind CSS
Deployment: Cloudfare Pages
Backend
Framework: FastAPI (Python)
ML Libraries: PyTorch, Transformers, Scikit-learn, NumPy
Deployment: Hugging Face Spaces (Dockerized)
🏗️ Architecture & Deployment Strategy
Running heavy ML models (PyTorch + JVM for Grammar) requires significant RAM. I implemented a Split Deployment Strategy to optimize performance on free-tier infrastructure:
Frontend: Handles UI/UX and client-side logic.
Backend: A custom Docker container running FastAPI.
Why Docker? To install System-level dependencies (OpenJDK 21 for Grammar check, build-essential for Auth).
Optimization: Configured to use CPU-optimized PyTorch builds to reduce slug size and memory footprint.
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|