“Just dropping this here since it looks like a lot of people haven't heard of it.”
Voyager: An Open-Ended Embodied Agent with Large Language Models
Minecraft 里由 GPT-4 驱动、用代码技能库终身学习的智能体
提出 Voyager,一个在 Minecraft 中由 LLM 驱动的终身学习智能体,包含最大化探索的自动课程、存放可执行代码的不断增长的技能库,以及结合环境反馈、执行错误和自我验证的迭代提示机制。它通过黑盒方式调用 GPT-4,无需微调;相比此前最优,获得 3.3 倍的独特物品、行进距离多 2.3 倍,关键科技树里程碑解锁最多快 15.3 倍,并能把技能库用到新世界。
HN 上的人反复用它说明让 LLM 把技能写成可复用代码并存起来、让智能体随时间自我改进的思路。
HN 书评
推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及“you could potentially do what they did in the Voyager paper Voyager”
“Any lesson learnt in the library could be used for fine-tuning or the next training run for a base model.”
“Their implementation uses actual code, JS scripts in their case, as the stored trajectories, which has the neat feature of parameterization built in so trajectories are more reusable.”
“I think LLM as a toolsmith like demonstrated in the Voyager paper (1) is another interesting approach to creating a system that can learn to do a task better over time.”
“there was a well-known paper called Voyager by NASA in which an agent was able to write its own skills in the form of code and improve them over time.”