大模型
x.com ·
@dotey
GPT-6 Astra用Computer Use玩Minecraft,对比Voyager
宝玉指出GPT-6 Astra借助Computer Use可直接玩Minecraft,对比三年前GPT-4需专门脚本的Voyager,并引用Jim Fan的推文回顾Voyager的Skill自进化与记忆功能。
AI推荐理由 对比GPT-4时代Voyager与GPT-6 Astra用Computer Use玩Minecraft,引用Jim Fan的原始推文,提供历史背景和对比,对Agent演进有参考价值。
65
浏览 11,466
点赞 28
评论 24
转发 1
正文
还记得三年前GPT-4玩Minecraft吗?那时候还要借助专门的脚本,还要自己做记忆库技能库。
现在GPT-6 Astra借助Computer Use就可以直接搞定了。
但现在看Voyager依然很了不起,那时候就已经实现了Skill的自进化和记忆功能,算是最早的Harness雏形之一了。
[您的浏览器不支持视频标签。](https://video.twimg.com/amplify_video/2096547508155809792/vid/avc1/1280x672/JPn_C_gHl2IFVcr0.mp4?tag=29)
[🔗 在Twitter上查看](https://x.com/wuyang_zhou/status/2096548032128942370/video/1)

Jim Fan@DrJimFan
如果我们让GPT-4在Minecraft中自由发挥会怎样?⛏️
我很高兴地宣布Voyager,第一个在Minecraft中纯粹在上下文中进行终身学习的代理。Voyager通过编写、完善、提交和检索技能库中的\*代码\*来不断改进自己。
GPT-4开启了一种新的范式:“训练”是代码执行而非梯度下降。“训练好的模型”是Voyager迭代组合的技能代码库,而非浮点矩阵。我们将无梯度架构推向了极限。
Voyager迅速成为一名经验丰富的探险家。在Minecraft中,它获得的独特物品数量是先前方法的3.3倍,旅行距离是2.3倍,解锁关键技术树里程碑的速度最高快15.3倍。
我们开源了一切。让通用代理在Minecraft中涌现!欢迎您今天尝试:[voyager.minedojo.org](https://voyager.minedojo.org/)
论文:[arxiv.org/abs/2305.16291](https://arxiv.org/abs/2305.16291)
代码:[github.com/MineDojo/Voyag…](https://github.com/MineDojo/Voyager)
与我深入探讨:🧵
[您的浏览器不支持视频标签。](https://video.twimg.com/ext_tw_video/1662113301889961986/pu/vid/1280x720/qh4DEjgdq4U86cVZ.mp4?tag=12)
[🔗 在Twitter上查看](https://x.com/DrJimFan/status/1662115266933972993/video/1)
[🔗 查看引用的推文](https://x.com/DrJimFan/status/1662115266933972993)
查看原文(中文正文为 AI 翻译)
还记得三年 GPT-4 玩 Minecraft 吗?那时候还要借助专门的脚本,还要自己做记忆库技能库。
现在 GPT-6 Astra 借助 Computer Use 就可以直接搞定了。
但现在看 Voyager 依然很了不起,那时候就已经实现了 Skill 的自进化和记忆功能,算是最早的 Harness 雏形之一了。
[Your browser does not support the video tag.](https://video.twimg.com/amplify_video/2096547508155809792/vid/avc1/1280x672/JPn_C_gHl2IFVcr0.mp4?tag=29)
[🔗 View on Twitter](https://x.com/wuyang_zhou/status/2096548032128942370/video/1)

Jim Fan@DrJimFan
What if we set GPT-4 free in Minecraft? ⛏️
I’m excited to announce Voyager, the first lifelong learning agent that plays Minecraft purely in-context. Voyager continuously improves itself by writing, refining, committing, and retrieving \*code\* from a skill library.
GPT-4 unlocks a new paradigm: “training” is code execution rather than gradient descent. “Trained model” is a codebase of skills that Voyager iteratively composes, rather than matrices of floats. We are pushing no-gradient architecture to its limit.
Voyager rapidly becomes a seasoned explorer. In Minecraft, it obtains 3.3× more unique items, travels 2.3× longer distances, and unlocks key tech tree milestones up to 15.3× faster than prior methods.
We open-source everything. Let generalist agents emerge in Minecraft! Welcome you all to try today: [voyager.minedojo.org](https://voyager.minedojo.org/)
Paper: [arxiv.org/abs/2305.16291](https://arxiv.org/abs/2305.16291)
Code: [github.com/MineDojo/Voyag…](https://github.com/MineDojo/Voyager)
Deep dive with me: 🧵
[Your browser does not support the video tag.](https://video.twimg.com/ext_tw_video/1662113301889961986/pu/vid/1280x720/qh4DEjgdq4U86cVZ.mp4?tag=12)
[🔗 View on Twitter](https://x.com/DrJimFan/status/1662115266933972993/video/1)
[🔗 View Quoted Tweet](https://x.com/DrJimFan/status/1662115266933972993)