About Us
xyLab focuses on AI for Science field, developing efficient research tools to reduce time costs for researchers. xyLab has its own dataset in academic fields and develops dedicated task models, not using GPT or other general model APIs.
Our philosophy is Y = F(X), meaning that all generative tasks are a process of function calculation.
Model Introduction
Detection Model
The detection model is based on 1.2 billion (60 million human-written papers and 60 million AI-generated papers) data, capable of recognizing the styles of various AI models in academic papers with an accuracy of 99.99%.
- Training Data:1.2 billion papers
- Recognition Ability:Various AI model styles
- Accuracy:99.99%
Rewriting Model
A model with 8 billion parameters, accurately learning the styles of human-written papers to transform mechanical AI text into natural and fluent content while maintaining academic rigor.
- Model Parameters:8 billion
- Core Ability:Learning Human Style
- Output Features:Natural and Fluent, but Rigorous
Function Introduction
AI Detection
Paste the content you want to detect into the input box, and click Detect to get the result in 0.1 second.
Note: This tool's algorithm is strict, which may identify texts similar to AI as AI.
Text Rewrite
Paste the content you want to rewrite into the input box, and click Rewrite to get the rewritten content and AI detection result.
Note: It's recommended to input content length between 50-200 words, and input one paragraph at a time.
Strong Rewrite
Get 5 rewriting results at once and get AI detection results for each.
Note: Even if the AI rate of rewritten content may be 100%, the rewritten content is likely to pass detection on other platforms.
Effect Demonstration (Turnitin)
We used paperwrite.ai to generate 10 papers entirely written by AI. In the first round of detection results, all papers had an AI probability of over 80%. After rewriting all the papers and resubmitting them for detection, the results for all papers were "*% detected as AI".
