Discuz! Board

 找回密码
 立即注册
搜索
热搜: 活动 交友 discuz
查看: 13|回复: 0

14岁奥地利少年创建本地多AI智能体协作桌面应用OpenAutark

[复制链接]

7万

主题

7万

帖子

21万

积分

论坛元老

Rank: 8Rank: 8

积分
219781
发表于 4 天前 | 显示全部楼层 |阅读模式
14岁奥地利少年创建本地多AI智能体协作桌面应用OpenAutark

机器人控制测试发现智能体会误报任务成功

RoboARCade把长序列操作任务放进同一套MuJoCo仿真环境,涉及三种机器人本体。团队扩大了原任务的初始条件范围,让智能体必须在执行过程中利用记忆、判断和视觉信息。综合测试中,GPT-6 Astra得分49.8%,Claude Fable 5.1为43.6%。更丰富的视觉界面带来1.5至6.8个百分点的改善,使用成本同时增至原来的1.7至2.3倍。 Satvik Sharma等人的这项研究收录在11月6日更新的计算机科学预印本列表中。

错误分析显示,主要困难在于选择操作策略,或把正确策略执行得足够精确。智能体在尝试过的子任务中,有24%至39%出现错误宣告成功的情况。在另一个五任务子集上,提示中加入一段成功示范后,Astra从53.5%提高到73.0%。后者属于子集测试,不能当作整套基准上的得分。团队公开了代码、视频与执行轨迹。

https://v.youku.com/v_show/id_XNjU3ODAwNTU2NA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNzExMg==.html
https://v.youku.com/v_show/id_XNjU3ODAwODUwNA==.html
https://v.youku.com/v_show/id_XNjU3ODAwODQ5Mg==.html
https://v.youku.com/v_show/id_XNjU3ODAwODQ4NA==.html
https://v.youku.com/v_show/id_XNjU4NzcwODQ2NA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNzA2NA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNzA1Mg==.html
https://v.youku.com/v_show/id_XNjU4NzcwODQ0OA==.html
https://v.youku.com/v_show/id_XNjU3ODAwNjk1Ng==.html
https://v.youku.com/v_show/id_XNjU4NzcwODQwMA==.html
https://v.youku.com/v_show/id_XNjU4NzcxMTQ1Ng==.html
https://v.youku.com/v_show/id_XNjU4NzcwNzAwMA==.html
https://v.youku.com/v_show/id_XNjU3ODAwODQxNg==.html
https://v.youku.com/v_show/id_XNjU3ODAwODQwNA==.html
https://v.youku.com/v_show/id_XNjU4NzcwODM3Ng==.html
https://v.youku.com/v_show/id_XNjU4NzcwODM2OA==.html
https://v.youku.com/v_show/id_XNjU4NzcwODM2MA==.html
https://v.youku.com/v_show/id_XNjU4NzcxMTQwMA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNTI1Mg==.html
https://v.youku.com/v_show/id_XNjU4NzcxMjkwMA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNjkyMA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNjkwNA==.html
https://v.youku.com/v_show/id_XNjU3ODAwNjgzMg==.html
https://v.youku.com/v_show/id_XNjU4NzcxMTM2MA==.html
https://v.youku.com/v_show/id_XNjU4NzcxMTM0NA==.html
https://v.youku.com/v_show/id_XNjU3ODAwMzg2OA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNjg0MA==.html
https://v.youku.com/v_show/id_XNjU4NzcwODI3Ng==.html
https://v.youku.com/v_show/id_XNjU4NzcxMTMyMA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNTEwNA==.html
https://v.youku.com/v_show/id_XNjU4NzcxMjgxMg==.html
https://v.youku.com/v_show/id_XNjU4NzcwNTA5Mg==.html
https://v.youku.com/v_show/id_XNjU3ODAwMzgwMA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNTA3Mg==.html
https://v.youku.com/v_show/id_XNjU4NzcwNTA1Ng==.html
https://v.youku.com/v_show/id_XNjU4NzcwODIxNg==.html
https://v.youku.com/v_show/id_XNjU3ODAwMjM4MA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNjc2OA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNjc2MA==.html
https://v.youku.com/v_show/id_XNjU3ODAwNjcwMA==.html
https://v.youku.com/v_show/id_XNjU4NzcwOTY2OA==.html
https://v.youku.com/v_show/id_XNjU4NzcwODE1Ng==.html
https://v.youku.com/v_show/id_XNjU3ODAwMjMzMg==.html
https://v.youku.com/v_show/id_XNjU3ODAwNjY1Ng==.html
https://v.youku.com/v_show/id_XNjU4NzcwNDk5Ng==.html
https://v.youku.com/v_show/id_XNjU4NzcwMjM4NA==.html
https://v.youku.com/v_show/id_XNjU4NzcwNDk3Mg==.html
https://v.youku.com/v_show/id_XNjU3ODAwODA4NA==.html
https://v.youku.com/v_show/id_XNjU3ODAwODA3Ng==.html
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

Archiver|手机版|小黑屋|DiscuzX

GMT+8, 2026-11-15 04:45 , Processed in 0.046965 second(s), 18 queries .

Powered by Discuz! X3.4

Copyright © 2001-2020, Tencent Cloud.

快速回复 返回顶部 返回列表