D

Deberta V3 Base Prompt Injection V2

Developed by protectai
A prompt injection detection model fine-tuned on DeBERTa-v3-base, designed to identify malicious prompts that may manipulate language models
Downloads 229.97k
Release Time : 4/20/2024

Model Overview

This model is specifically designed to detect and classify prompt injection attacks that may manipulate language models to produce unintended outputs, enhancing the security of language model applications

Model Features

High Accuracy Detection
Achieves 95.25% accuracy and 99.74% recall on independent test sets
Multi-Dataset Training
Integrates multiple public datasets, covering a wide range of prompt variants
Focus on English Prompts
Specifically optimized for detecting English prompt injections
Community-Driven Improvements
Continuously optimizes model performance based on community feedback

Model Capabilities

Prompt Injection Detection
Text Classification
Security Protection

Use Cases

Large Language Model Security
Chatbot Protection
Detects and blocks malicious prompts attempting to manipulate chatbot outputs
Effectively prevents harmful content generation
API Security Gateway
Integrated into API gateways to filter malicious prompt requests
Enhances the security of language model APIs
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase