My college roomate and I are fine-tuning an open source LLM to detect prompt injection and other LLM targeted attacks. We made a game you can play here to try to get the LLM to give up a secret code. So far only 1 person has been able to crack it. Checkit it pit here: https://www.integrated.io/
Fine tuning an open source model to detect LLM attacks. Started out with this game where you try to get the LLM to give up the secret key: https://www.integrated.io/
My college roomate and I are fine-tuning an open source LLM to detect prompt injection and other LLM targeted attacks. We made a game you can play here to try to get the LLM to give up a secret code. So far only 1 person has been able to crack it. Checkit it pit here: https://www.integrated.io/