tawer12

SecAlign-Llama-3.1-8B-injection-at-end-inner10

tawer12/SecAlign-Llama-3.1-8B-injection-at-end-inner10

SecAlign-Llama-3.1-8B-injection-at-end-inner10 is a text generation model from tawer12. Use it when you need the model to write or continue text. It is set up for peft.

LoRA adapter trained with an SFT continuous attacker placed immediately after the text-injection span. Uses up to 10 attack tokens and 10 attacker optimization steps.