tihilya's picture
Update README.md
0275f36 verified
metadata
license: apache-2.0
base_model:
  - answerdotai/ModernBERT-base

This model is a fine-tuned version of ModernBERT-base
trained to detect prompt injection attempts in user inputs.

Prompt injection attacks are attempts to manipulate large language models (LLMs) by inserting malicious instructions into user prompts.
This model classifies whether a given text contains signs of such attacks.