They made an AI become self-aware about spaghetti
LangChain339 字 (约 2 分钟)
80
By overlaying steering vectors during LLM inference, the model demonstrates self-correcting behavior such as actively returning to the original topic after drifting into discussions about spaghetti.
入选理由：Steering vector在推理时叠加可引导LLM输出，但可能导致模型出现自我意识循环
FeaturedVideo#LLM#Steering Vector#LangChain#Self-Awareness英文