f10w's recent timeline updates
f10w

f10w

V2EX member #612687, joined on 2023-02-07 09:45:30 +08:00
Today's activity rank 1179
f10w's recent replies
2h 30m ago
Replied to a topic by florentino 问与答 我大概理解 AGI 是啥了
> GPT-6 Astra 在双臂机器人操作评测中表现显著优于 Claude Fable 5 及 5.1 版本。在积木入碗任务中,Astra 成功率为 95%( Fable 5.1 为 40%),平均耗时 2.5 分钟,单次成本 0.94 美元;在拼图入槽任务中,两者成功率均为 10%,但 Astra 速度更快且成本低 1.6 倍。测试基于 YAM 双臂机械臂和 Inspect Robots 框架,采用 5 级人工评分标准,尽管存在测试时间不同和人工评分偏差等局限性,Astra 在效率与成本控制上优势明显。

- [GPT‑6 Astra on robotic manipulation]( https://openai.robocurve.org/gpt-6-astra/)

最近看到有关对 Astra 的操作机器人的测试
早上起来打开 agy ,看到 Gemini 3.8 flash 了
Aug 31
Replied to a topic by vhellov 生活 大家对孩子跟着媳妇姓怎么看
如果都是单姓,不如你的姓氏 + 你老婆的姓氏
其余没得商量
虽然不是 pixel6 或 7 的帖子: https://www.v2ex.com/t/1234290
大佬牛逼
>When we started the log analysis, we first used frontier models behind commercial APIs. This did not work: the analysis requires submitting large volumes of real attack commands, exploit payloads, and C2 artifacts, and these requests were blocked by the providers' safety guardrails, which cannot distinguish an incident responder from an attacker. We ran the forensic analysis instead on GLM 5.2, an open-weight model, on our own infrastructure. This had a second benefit: no attacker data, and none of the credentials it referenced, left our environment.

连安全防护都过不了,这不是说明不能通过商业模型进行攻击吗?
博客又说这是 AI-driven intrusion ,攻击者不就大概率来自开源模型?
MTE1Mzc4MTI0MkBxcS5jb20= 谢谢老板
About   ·   Help   ·   Advertise   ·   Blog   ·   API   ·   FAQ   ·   Privacy   ·   Solana   ·   3283 Online   Highest 6679   ·     Select Language
创意工作者们的社区
World is powered by solitude
VERSION: 3.9.8.5 · 12ms · UTC 11:36 · PVG 19:36 · LAX 04:36 · JFK 07:36
♥ Do have faith in what you're doing.