huangjj877/TempParaphraser
The huangjj877/TempParaphraser is a 1 billion parameter model with a 32768 token context length, specifically designed for text paraphrasing. Developed for the EMNLP 2025 paper "TempParaphraser: 'Heating Up' Text to Evade AI-Text Detection through Paraphrasing," its primary function is to generate paraphrases that can bypass AI-text detection. This model is optimized for research and academic applications in text manipulation and AI-text detection evasion.
Loading preview...
Model Overview
The huangjj877/TempParaphraser is a 1 billion parameter model with a 32768 token context length, developed as the official paraphrasing model for the EMNLP 2025 paper, "TempParaphraser: 'Heating Up' Text to Evade AI-Text Detection through Paraphrasing." Its core capability is to generate paraphrased text specifically engineered to evade detection by AI-text detection systems.
Key Capabilities
- AI-Text Detection Evasion: Generates paraphrases designed to alter text sufficiently to bypass common AI-text detection algorithms.
- Text Paraphrasing: Provides a mechanism for rephrasing existing text while maintaining its original meaning.
- Research Tool: Serves as a dedicated tool for academic research into text generation, paraphrasing techniques, and AI-text detection countermeasures.
Intended Use
This model is strictly intended for research and academic purposes only. Commercial use is explicitly prohibited. For detailed usage instructions, examples, and evaluation scripts, users are directed to the associated GitHub repository.