OpenAI trained GPT-Red, an internal-only attacker model, using self-play reinforcement learning against a population of defender LLMs. It beat human red-teamers 84% to 13% on a replicated indirect ...
2 followers Rua José Dias Coelho 10A, 2855-173, Corroios, Portugal https://badkittysdad.com ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results