LMAT-ND: A Meta-Attention-Enhanced Llama-7B Model for AI-Generated News Detection

Abstract excerpt

This study introduces LMAT-ND (Llama-7B Enhanced Meta-Attention Transformer for News Detection) to distinguish AI-generated from human-authored news by integrating a Meta-Attention Mechanism and Dynamic Multi-Head Attention to capture linguistic distinctions, alongside a Dual-Classification Layer for optimized classification. LMAT-ND leverages Meta-Attention to dynamically adjust attention distribution, enhancing...

Topics

Open a Topic to create a Post that cites this publication.

• Algorithms
• COVID-19
• Genetic Variation
• Mesenchymal stem cell research
• Mutation
• Natural Language Processing
• Pluripotent Stem Cells Research
• Tissue Engineering and Regenerative Medicine

Identifiers and source

Literature Corpus work

c89ef176-69b0-54fa-8343-08f792eeb1ff

DOI

10.20944/preprints202505.0684.v1