小米公布澎湃HyperOS 4修复进展，涉及小米17系列
小米公布澎湃HyperOS 4修复进展，涉及小米17系列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://lightroom.adobe.com/shares/7387bf03bccfbc41a8d8cc7bdbf472f2
https://lightroom.adobe.com/shares/6a400c7516cb3445b22c86569150a7d1
https://lightroom.adobe.com/shares/af6fda5abf8c474584eb4903d3c1b45b
https://lightroom.adobe.com/shares/9aba8f299ae7b94db2270695ce7d6973
https://lightroom.adobe.com/shares/97d7066a63530646b3809622017989b1
https://lightroom.adobe.com/shares/6fe90945fcabfb46b5687b841968424a
https://lightroom.adobe.com/shares/6494d968974eda41b868bb62afefd73a
https://lightroom.adobe.com/shares/57bfd598983ea34f830f5aa1b3c080b0
https://lightroom.adobe.com/shares/cc852ba3f8e6d848b7c1e22633a79819
https://lightroom.adobe.com/shares/894b98f4f9079940a46ec5719cdb9e0e
https://lightroom.adobe.com/shares/b1b50702d14f1d46a166d89d628721ef
https://lightroom.adobe.com/shares/4aafb863d880f148a438ba294b5f5882
https://lightroom.adobe.com/shares/958d9915c8edcf4090f65d1852fe64a3
https://lightroom.adobe.com/shares/a3f9316a63f4eb4496459e3c697f0854
https://lightroom.adobe.com/shares/e12e7150757445419cb2a90e57af3dd1
https://lightroom.adobe.com/shares/b44c844846797e46a4f204c87565c24c
https://lightroom.adobe.com/shares/f880bd503456094886d2151eab8bf92b
https://lightroom.adobe.com/shares/24539f6350753549bb800bd6f596805e
https://lightroom.adobe.com/shares/125aa3d41e737042b20ebf88d1224a85
https://lightroom.adobe.com/shares/8c2908ef4cce7d49b898644e6fe695cb
https://lightroom.adobe.com/shares/81e9069f86c9a140a924673b92fa9829
https://lightroom.adobe.com/shares/6d3436451fca7645a82fa4decc5a3ea0
https://lightroom.adobe.com/shares/7d8d4185f10ecd42b3e71e6cc0b31ea8
https://lightroom.adobe.com/shares/7f1b0ef9cb3dec48b82b04624d4bb41b
https://lightroom.adobe.com/shares/22fd8de4c7c7554486f8ba96c0d72b6f
https://lightroom.adobe.com/shares/353f7e3e8213a544b21c17b858935e50
https://lightroom.adobe.com/shares/5162398fc37b2849bcfd2741e77ca0da
https://lightroom.adobe.com/shares/be89d6e2094c5647b35d827a25f91086
https://lightroom.adobe.com/shares/619346d52816624488c20c8e65dc6bcf
https://lightroom.adobe.com/shares/3c3176e7e8ded84cb56868c566049a6e
https://lightroom.adobe.com/shares/73bbc673a3607e4b813cb09a2b50cfc7
https://lightroom.adobe.com/shares/eac68260deceff41aa4cdb44a7ecdbcc
https://lightroom.adobe.com/shares/adf1dace31ade543ac865c15f8ae20c5
https://lightroom.adobe.com/shares/ef62f43c25086b4dbe8060a731af3028
https://lightroom.adobe.com/shares/6967902e8cd56449b66b9c570627624e
https://lightroom.adobe.com/shares/b73a0df69e006d4dae580ccb67fff090
https://lightroom.adobe.com/shares/fad44d4805b180489223299df1c478ab
https://lightroom.adobe.com/shares/0c315f761683f74fbfca237a4f96394e
https://lightroom.adobe.com/shares/a93c6592f370ba4dab25c3d8968dd6d0
https://lightroom.adobe.com/shares/5168d37941f1cc4c925a6e4959bf124f
https://lightroom.adobe.com/shares/363a002eba52944cbcb86ade9ee5a15f
https://lightroom.adobe.com/shares/a6f3e74e1d737e428586427657be1c62
https://lightroom.adobe.com/shares/30bee18415f8734e8d23d8be8e8b82f8
https://lightroom.adobe.com/shares/3a610f2923459d4d890c85c9d4d3c309
https://lightroom.adobe.com/shares/0b5130406103da4ea5b5fc86e5283229
https://lightroom.adobe.com/shares/2e2ecfd83c996846bb53e2c130b57119
https://lightroom.adobe.com/shares/9e0d6f42dd551c429e590cd39ba14d2c
https://lightroom.adobe.com/shares/4857c358a8180b42b44d36e21cd8a7ba
https://lightroom.adobe.com/shares/719adceb5a96a74bbe67f0dac8cb00f1
https://lightroom.adobe.com/shares/d4579717da3557478fa7f5680449bec4
沙发！沙发!
页:
[1]