Discuz! Board

 找回密码
 立即注册
搜索
热搜: 活动 交友 discuz
查看: 1|回复: 0

黄河流域重点河湖生态流量达标率超过90%

[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
黄河流域重点河湖生态流量达标率超过90%

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_XNjU4ODA2MzI5Mg==.html
https://v.youku.com/v_show/id_XNjU3ODM2NjU4MA==.html
https://v.youku.com/v_show/id_XNjU4ODA2MzI3Ng==.html
https://v.youku.com/v_show/id_XNjU3ODM2NjU2MA==.html
https://v.youku.com/v_show/id_XNjU4ODA2MDE1Ng==.html
https://v.youku.com/v_show/id_XNjU4ODA2MDE1Mg==.html
https://v.youku.com/v_show/id_XNjU4ODA1OTE5Mg==.html
https://v.youku.com/v_show/id_XNjU3ODM2MTk4OA==.html
https://v.youku.com/v_show/id_XNjU3ODM2MzU0MA==.html
https://v.youku.com/v_show/id_XNjU3ODM2NDkxMg==.html
https://v.youku.com/v_show/id_XNjU3ODM2NDg4OA==.html
https://v.youku.com/v_show/id_XNjU3ODM2MzQ5Mg==.html
https://v.youku.com/v_show/id_XNjU3ODM1OTExMg==.html
https://v.youku.com/v_show/id_XNjU3ODM1OTEwNA==.html
https://v.youku.com/v_show/id_XNjU3ODM1OTA3Ng==.html
https://v.youku.com/v_show/id_XNjU3ODM2MDYyNA==.html
https://v.youku.com/v_show/id_XNjU3ODM2MDU2NA==.html
https://v.youku.com/v_show/id_XNjU4ODA1NzUxNg==.html
https://v.youku.com/v_show/id_XNjU3ODM2MDUyMA==.html
https://v.youku.com/v_show/id_XNjU3ODM1ODk1Mg==.html
https://v.youku.com/v_show/id_XNjU3ODM1NTk3Mg==.html
https://v.youku.com/v_show/id_XNjU3ODM2MDQ4OA==.html
https://v.youku.com/v_show/id_XNjU3ODM2MzMxNg==.html
https://v.youku.com/v_show/id_XNjU4ODA1NzQwOA==.html
https://v.youku.com/v_show/id_XNjU3ODM2MzMwNA==.html
https://v.youku.com/v_show/id_XNjU3ODM1NTkwNA==.html
https://v.youku.com/v_show/id_XNjU3ODM1ODg4NA==.html
https://v.youku.com/v_show/id_XNjU3ODM2MzIzMg==.html
https://v.youku.com/v_show/id_XNjU3ODM1ODg1Ng==.html
https://v.youku.com/v_show/id_XNjU4ODA1NzMwOA==.html
https://v.youku.com/v_show/id_XNjU3ODM1NTgwNA==.html
https://v.youku.com/v_show/id_XNjU4ODA1Mjc1Mg==.html
https://v.youku.com/v_show/id_XNjU4ODA1Mjc0OA==.html
https://v.youku.com/v_show/id_XNjU3ODM1ODc5Mg==.html
https://v.youku.com/v_show/id_XNjU3ODM1ODc4MA==.html
https://v.youku.com/v_show/id_XNjU4ODA1MjczMg==.html
https://v.youku.com/v_show/id_XNjU4ODA1NzI0OA==.html
https://v.youku.com/v_show/id_XNjU3ODM1NDMwNA==.html
https://v.youku.com/v_show/id_XNjU4ODA1NDAwMA==.html
https://v.youku.com/v_show/id_XNjU4ODA1Mzk2NA==.html
https://v.youku.com/v_show/id_XNjU4ODA1MjYzMg==.html
https://v.youku.com/v_show/id_XNjU3ODM2MDIyOA==.html
https://v.youku.com/v_show/id_XNjU3ODM1NTYyOA==.html
https://v.youku.com/v_show/id_XNjU3ODM1NTYxMg==.html
https://v.youku.com/v_show/id_XNjU3ODM1NTU5Mg==.html
https://v.youku.com/v_show/id_XNjU3ODM1ODU4MA==.html
https://v.youku.com/v_show/id_XNjU4ODA1MjU2OA==.html
https://v.youku.com/v_show/id_XNjU4ODA1MjUwNA==.html
https://v.youku.com/v_show/id_XNjU3ODM1ODQ4MA==.html
https://v.youku.com/v_show/id_XNjU3ODM1NDA1Ng==.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 05:55 , Processed in 0.043526 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