LLMGuard – Prompt Injection Detection
This Streamlit app detects prompt injection attempts in real-time using a fine-tuned Flan-T5 model hosted on Hugging Face.
Features
- Detects malicious prompt injections
- Confidence scoring
- Moderation history panel
- Streamlit-based UI
- Powered by
Tuathe/llmguard-injection-model
Tech Stack
- Hugging Face Transformers
- PyTorch
- Streamlit