Lucy X ✨

4.3K posts

Lucy X ✨ banner
Lucy X ✨

Lucy X ✨

@LucyofX

frontier surfer - 推特19年老鸟

Katılım Temmuz 2007
64 Takip Edilen431 Takipçiler
Rob Bensinger ⏹️
Rob Bensinger ⏹️@robbensinger·
Who should I add to this? Also, did I get anyone's view wrong?
Rob Bensinger ⏹️ tweet media
English
87
28
398
63.2K
Lucy X ✨
Lucy X ✨@LucyofX·
2 ways to survive the internet age: 1. Top-down censorship. 2. Retardmaxxing.
English
0
0
0
20
Lucy X ✨
Lucy X ✨@LucyofX·
@antirez i also found Claude is good (maybe slightly better) at frontend, but cannot compete with Codex in backend/technical details.
English
0
0
0
610
antirez
antirez@antirez·
The difference between the two, for serious engineering work, is simply brutal. Claude Code with Opus is, when the task at hand is very complicated, borderline useless, while GPT 5.4 can do a reverse engineering mixing: hardware knowledges, major disassembly skills, and so on.
English
9
52
794
109K
antirez
antirez@antirez·
During the last week I executed very long autonomous sessions of Claude Code Opus 4.6 and Codex GPT 5.4 (both at max thinking budget), in cloned directories (refreshed every time one was behind). I burned a lot of (flat rate, my OSS free account + my PRO account) of tokens...
English
64
178
2.2K
914.9K
Lucy X ✨
Lucy X ✨@LucyofX·
Distribution is EVERYTHING. It already replaced marketing & sales.
Brian Halligan@bhalligan

A lot of founders have pinged me to ask what steps they would need to take to pull of what @jack has done restructuring @blocks this way. He answers it spot on 👇 "look at all the tools you're using. Look at all the information you're generating just by doing your work. Just putting that into an intelligence and being able to query it will give you an understanding of the company that is 2-3x more than you had before, ever." If you want to hear more, let me know and I can share my longer thoughts

English
0
0
0
27
Lucy X ✨
Lucy X ✨@LucyofX·
低技术社会:反扒 高技术社会:反爬
中文
0
0
0
20
Lucy X ✨
Lucy X ✨@LucyofX·
@kepano That text is from a Confucian classic 孝经 on how to behave towards seniors such as parents, elder brothers, or a ruler.
English
0
0
1
132
kepano
kepano@kepano·
Stone tablet of king Sejo's calligraphy "During the wars the royal collection of art and calligraphy was decimated or dispersed. King Hyeonjong (r. 1659-1675) launched a project to collect calligraphy works written by preceding kings and carve them into stone."
kepano tweet media
English
6
11
175
29.9K
Lucy X ✨
Lucy X ✨@LucyofX·
其实比起 articles,我更喜欢 threads。连Meta都照抄了,还取了这个名字,说明这个模式本身就是护城河。 但大家都喜欢,不代表能赚钱。现在创作者经济和盈利才是X活下去的动力。 一个好产品被盈利和 KPI 绑架,可能是科技界最悲哀的事情。但如果不活下去的话,X 可能会撑不到超级 App 的崛起?
中文
0
0
0
55
Lucy X ✨
Lucy X ✨@LucyofX·
江学勤最新的视频 Game Theory 18 - Trump World Order让我震惊了。查了一下维基百科 Technocracy Movement (技术治国运动)词条,发现 Elon Musk 的外祖父是 1940 年 Technocracy Inc. 领导人,原来硅谷技术治国运动是有历史渊源的。
Lucy X ✨@LucyofX

x.com/i/article/2039…

中文
0
0
0
169
Ben James
Ben James@BenJames_____·
I made myself a mini fax machine. Dedicated phone number, instantly prints anything you text it.
Ben James tweet media
English
261
334
6.7K
594.8K
宝玉
宝玉@dotey·
SentrySearch:用自然语言搜索视频内容的开源工具 在几个小时的行车记录仪视频里找到"一辆红色卡车闯了停牌"那个画面,SentrySearch 能让你像搜文字一样搜视频,输入描述,直接导出对应片段。 这个开源命令行工具的原理并不复杂:把视频切成带重叠的片段,用 Google Gemini Embedding API 或本地的 Qwen3-VL 模型把每个片段编码成向量,存进本地向量数据库 ChromaDB。搜索时,文字查询被编码到同一个向量空间里做匹配,命中的片段自动从原文件中裁剪出来。 关键在于,整个过程没有转录、没有逐帧生成文字描述,视频像素直接和文字查询在向量层面比较。这是 Gemini Embedding 2 和 Qwen3-VL-Embedding 这类多模态嵌入模型带来的能力,让对海量视频的语义搜索变得可行。 想用云端 API,一小时视频的索引成本大约 2.84 美元。想完全离线也行,装上本地 Qwen3-VL 模型就不需要任何 API 密钥,24GB 以上显存或内存的 Mac 和 NVIDIA GPU 都能跑。它还专门做了特斯拉行车记录仪的适配,能在裁剪出来的片段上叠加车速、GPS 位置和时间信息。 还有 Skill 可以用:clawhub.ai/ssrajadh/natur… 项目地址:github.com/ssrajadh/sentr…
中文
21
103
583
63.8K
kepano
kepano@kepano·
I want to try this tiny Obsidian travel setup now that iOS supports mouse input
English
41
28
816
109.9K
Lucy X ✨
Lucy X ✨@LucyofX·
@CuiMao @DarioAmodei Anthropic 团队真的强,绝对是世界顶级的舆论和叙事操控手。每次X上稍微有 Codex/OpenAI占上风的趋势,马上就会搞出事情来牢牢把控话题和叙事风口。
中文
0
0
3
1.4K
Lucy X ✨ retweetledi
CuiMao
CuiMao@CuiMao·
哥,你居然都把源码都放出来了,你是真的 OpenAI @DarioAmodei ,你是这个👍🏻,他们都是假的 OpenAI!!!
中文
34
12
317
77K
Lucy X ✨
Lucy X ✨@LucyofX·
Grok 说 Anthropic 的护城河是: 1. 核心护城河:模型本身 + Constitutional AI 的安全哲学 2. 内部执行力和产品做产品的飞轮 3. 企业订阅市场和数据飞轮:用6 个月以上高留存用户会主动把 Claude 用在更高价值任务上,成功率也更高,形成正向学习飞轮
中文
0
0
0
122
Lucy X ✨
Lucy X ✨@LucyofX·
根据Claude Code 泄露源码分析,已有多个已完成但未发布的功能,通过 Bun 的功能开关(feature flags)隐藏。 什么叫牛X?就是给你开源了旗舰产品代码和路线图,照样有深厚的护城河。
中文
1
0
1
148