Securing our frontier model training environment
OpenAI explains how we strengthened containment for frontier model training and evaluation, including offline environments and high-compute red-teaming.
原文：https://www.a8q9.cn/en/article/securing-our-frontier-model-training-environment-c72300.html