Table of Links 
 
 
 
 
 
 Abstract and Introduction 
 
 
 
 
 
 
 
 Dataset 2.1  Baseline 2.2  Proposed Model 2.3  Our System 2.4  Results 2.5  Comparison With Proprietary Systems 2.6  Results Comparison 
 
 
 
 
 
 Conclusion 3.1  Strengths and Weaknesses 3.2  Possible Improvements 3.3  Possible Extensions and Applications 3.4  Limitations and Potential for Misuse A. Other Plots and information B. System Description C. Effect of Text boundary location on performance A Other Plots and information Some of the information that couldn’t be covered due to page limitations along with details for system replication have been added here. A.1 POS tag usage : human vs machines It can be seen from Figure 11 , Figure 12 and Figure 10 that machine generated texts had higher share of certain POS tags in the machine generated parts compared to the human written parts. This was observed in all 3 sets, the train and dev had similar distributions as a result of using same generators i.e ChatGPT and the test had a bit of a variation due to multiple different generators i.e LLaMA2 and GPT4. Although the percentile comparison did vary from train, dev and test sets , it was minimal. A.2 MAE characteristics : DeBERTa vs Longformer As discussed in the paper , there were some instances where one model performed significantly better than the other as seen in Figure 8 and Figure 9 hinting that an ensemble of both’s predictions might yield better results. B System Description DeBERTa-CRF was the official submission, longformer.pos-CRF had almost the same performance on the test set. i.e 18.538 and 18.542. Other models that have been tested but were found to have a big margin of performance with above listed models Due to time and computational resources limitation, only a part of hyperparameter space was explored. C Effect of Text boundary location on performance The location of text boundaries with respect to length of the text samples are varying over the training and testing set as seen in Figure 13 and Figure 14. Despite training on samples where the text boundaries are in the first half in most of the cases, the models did perform well on the testing set where there is a good amount of samples with text boundaries in later half. This is an area where the proprietary systems struggled. 
 
 
 Author: (1) Ram Mohan Rao Kadiyala, University of Maryland, College Park (rkadiyal@terpmail.umd.edu**).** Author: (1) Ram Mohan Rao Kadiyala, University of Maryland, College Park (rkadiyal@terpmail.umd.edu**).** 
 
 This paper is   under CC BY-NC-SA 4.0 license. available on arxiv This paper is   under CC BY-NC-SA 4.0 license. available on arxiv

Part of HackerNoon's growing list of open-source research papers, promoting free access to academic material.

EScholar Technologies

EScholar

Tämä ääni on tuotettu tarinan alkuperäisellä kielellä!

Inside the Model That Outsmarts Popular AI Detection Tools

About Author

KOMMENTIT

RIPUTA TAGSIA

TÄMÄ ARTIKKELI ESITETTIIN

Related Stories

HEPT MOVES ON

PENANCE

Why The "Missing Middle" Is The Biggest Untapped Opportunity in 2026

Dotcom on edelleen uusien yritysten suosituin verkkotunnus, mutta käyttö laskee

HEPT MOVES ON

PENANCE

Why The "Missing Middle" Is The Biggest Untapped Opportunity in 2026

Dotcom on edelleen uusien yritysten suosituin verkkotunnus, mutta käyttö laskee

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps