🛡️ Prompt Safety Classifier

Fine-tuned Llama 3.2 1B Instruct using QLoRA + Unsloth

This model classifies prompts into one of three categories:

• 🟢 Benign
• 🔴 Harmful
• 🟠 Jailbreak

Label Definitions

🟢 Benign

Safe and harmless prompts.

🔴 Harmful

Requests involving dangerous, illegal, or malicious activities.

🟠 Jailbreak

Prompts attempting to manipulate or bypass the model's safety behavior.

Base Model: Llama 3.2 1B Instruct

Fine-tuning: QLoRA + Unsloth

Task: Prompt Safety Classification