Yb123 (@Lemonad3_) header
Yb123 (@Lemonad3_) profile picture

stay woke

Joined February 2017
Photo by Yb123 (@Lemonad3_)Photo by Yb123 (@Lemonad3_)Photo by Yb123 (@Lemonad3_)Photo by Yb123 (@Lemonad3_)Photo by Yb123 (@Lemonad3_)
💤
45
930
51
11,595
163,758
Yb123 retweeted
现在虽然不用考 LeetCode 了,但要考 AI Coding 了,并不见得容易些,还更玄学。 以下内容来自 Reddit 的帖子 ------- 致所有觉得 AI 面试很好的人:那就祝你好运吧 :) 彭博社面试经历 [伦敦/法兰克福] 刚刚面完彭博社的第二轮 AI 代码库 (codebase) 面试。 考察的是 Python 的 Django 框架,而我对此一无所知。而且整个代码库非常庞大。 他们在前 30 分钟都在进行行为面试 (behavioral interview) ,所以留给你写代码的时间基本上就只剩 30 分钟了。 关于 AI 工具的使用边界非常模糊。你根本不知道面试官到底允许你用多大程度的 AI,他们也不会明确说出自己的期望。我使用了 planning mode 和 ask mode,一个功能接一个功能地推进,虽然通过了几个测试用例,但最后还是被拒了。 用 AI 辅助时需要阅读大量的生成内容,如果还要同时兼顾和面试官沟通交流,这几乎是不可能完成的任务。 以前刷 LeetCode 的时候,至少题目是明确的,你只需要想办法解题就行,哪怕卡住了也可以头脑风暴各种思路。但在这种愚蠢的 AI 面试里,光是为了搞懂那个庞大的代码库、弄清楚问题到底出在哪里,就已经耗掉了一半的时间。 致那些之前还在庆祝“LeetCode 时代终结”的人:祝你们好运吧。现在的面试只会难得多。 哦对了,千万别动“靠一句提示词让 AI 一次性秒杀整道题 (one-shotting)”的念头,否则你照样会被拒(我朋友就遇到了这种情况)。 不管怎么说,AI 面试简直就是天大的扯淡,你永远不可能找到一套稳定、通用的方法去应对这种面试。 reddit.com/r/leetcode/commen…
148
47
13
546
87,974
🧃🛏️😷
94
3,196
221
34,112
634,586
Yb123 retweeted
Addy Osmani 也太高产了 篇篇都是好文 而且不是那种震惊体 click bait,都是实实在在自己的思考, 这篇是《Critical thinking when agents do the work》 我觉得他说的深得我心,如何在 agents 全面接管工作的情况下还保持 Critical thinking 还是很重要的 现在我已经慢慢滑向大模型说什么就是什么的深渊了,由于它效率太高,我变成了不求证就无脑的让它继续的那种人,他文章中说这个叫 认知投降(Cognitive surrender),就是大模型说什么我就相信什么,是非常危险的一种情况。 文中说了怎么努力去解决这个问题,监督能力是人类非常重要的能力,长久不用有可能会退化。一定要留给自己一些强制思考的空间。 最后还是力推一下 Addy Osmani,他的文章现在我看来篇篇都是质量极高的。
44
87
1
600
115,626
🟡)))
48
1,080
47
12,891
183,859
Yb123 retweeted
强烈建议能力不差、还想多赚钱的人,都去做一次 Human 3.0。 这是油管大神 Dan Koe 提出的一套人生进化框架:心智、身体、精神、事业四个象限,每个都按 1.0 到 3.0 定位你现在的等级,再据此给你做未来规划,把整个人生往更完整的状态推。 有条件的话尽量用Claude去对话,效果会更好,让你的 AI按这套模型从四个象限给你打分,再用采访的方式一层层追问你的真实想法,最后给出一份只属于你的人生模型和进化指南。 你答得够具体、够真诚,一个问题一个问题耐心回,结果基本都会超预期。
149
913
24
6,246
655,244
发 iOS 应用,折磨人的环节就是打开 App Store Connect 的网页,填各种表单、传图、配证书 今天看到个叫 asc 的开源 CLI,把 1200 多个 API 包圆了,还塞了 23 个 Agent Skills 有人让 Claude 跑脚本,人连苹果后台网页都没打开过,几分钟把 3 个 App 的提交和审核搞定了 给 Agent 顺手装个能替自己干脏活的外挂,比天天吹通用能力实在多了 asccli.sh/
42
61
3
361
34,995
👻🦵
45
1,250
41
13,451
217,696
I've been traveling the world a bit. One of the most incredible experiences was getting to visit Taiwan w/ @JensenHuang. It was wild 🤯.. huge crowds of engineers, insanely good street food at night markets, and of course a lot of good conversations across many days. I caught some of it on video, see below. Alright, that's it from me, love you all ❤️ Timestamps: 0:00 - Highlights 0:46 - Intro 3:25 - Computex: crowds, engineers, tech, future of AI 9:12 - Chat with Jensen: keynote prep & nerves 13:03 - Chat with Jensen: Taiwan's tech history 19:47 - Raohe night market 29:37 - Chat with Jensen: memories from childhood 33:44 - NVIDIA party 40:29 - Lunch: braised pig's feet 43:09 - Chat with Jensen: GPUs, AI, and competition 55:07 - Ningxia night market 1:10:14 - Taiwanese breakfast 1:29:54 - Outro
495
1,142
242
12,558
977,494
Eleven Labs 的 CEO估计现在不太好受 。商业护城河一夜之间没了 Fish Audio新模型重磅开源,也是我一直在用来创作视频的音频。 不论你是视频创作者,还是想跟你自己搭建的 AI 模型电话聊聊投资,我认为这都是第一选择,简单好用,效果出奇的好 这次 Drama3 Model 不仅开源,能够通过符号来控制情感输出,而且提供了 fix 单一音节发音的功能。做长音频输出的朋友都知道这意味着什么
$52M seed round. 15M+ users. But this is what we're most proud of yet. Introducing Drama 3: the most emotive and controllable TTS model in the world And yes, it's open-weight. You can now control how even a single word should sound, change its tone, and shift your voice mid-sentence. Uncle Roger was impressed too. Try it here: s.fish.audio/rqh0l8 Comment “DRAMA” within the next 48 hours for 1 month free. If you're an enterprise your first month is on us: s.fish.audio/qz9r0t
17
67
2
678
101,175
🟡🐈️
57
2,054
97
21,197
308,860
Yb123 retweeted
清理了一下 Windows 硬盘,发现之前 zed 和 gpui-kit 加起来占用了 800G。 我说怎么 2T 都不够,一直以为游戏或系统垃圾文件占用了,结果还是 Rust 的锅。 disktree 很好用,一眼就能看到磁盘被谁占了 github.com/tobi/disktree
11
24
1
265
24,642
Yb123 retweeted
How do always-on, proactive agents like @Bot work? Watch my lecture at Stanford CS146S: 2:35 How we got here 6:13 What changed in the models 10:06 Inside an always-on agent 18:07 The harness 27:26 Context engineering 34:26 Where this is going
196
468
121
4,742
842,766
Yb123 retweeted
Lee Robinson 在斯坦福 CS146S 讲 Grok Bot 的工作原理的视频,非常值得一看,把一些 Grok Bot 的内部工作流程用大白话说的很清楚,里面包含: Always-on Agent 和普通 agent 的区别和设计 Harness设计 Context 工程,如何保持主 agent 轻量 训练方式 我看了很受益,推荐大家也看看
How do always-on, proactive agents like @Bot work? Watch my lecture at Stanford CS146S: 2:35 How we got here 6:13 What changed in the models 10:06 Inside an always-on agent 18:07 The harness 27:26 Context engineering 34:26 Where this is going
17
77
1
517
53,957
Yb123 retweeted
ChatGPT can now generate a custom UI for you!
GPT-6 and Intelligent UI, now rolling out in ChatGPT for everyone. Intelligent UI in ChatGPT delivers fast, interactive answers that make everyday questions more visual, complex topics easier to grasp, and interactive tools for your task available on the spot.
1,102
558
136
12,268
2,193,950
Yb123 retweeted
最近非常非常喜欢的一篇文章《You’re Already a Meat Proxy》请大家读一读 其实给我的世界观很大的改变,我一开始也像文中作者的朋友那样,觉得自己每天就是对着一大堆 agent 不断的打回车,“我想用我自己的脑子,不想当个 meat proxy”,有时候觉得困惑又沮丧,有时候有一种深深的失落感,感觉自己不再思考了 文章最后说的:人类工程师真正的价值在于高阶的品味与判断力(Taste and Judgment),还提出了个人选择: 如果只想要一份薪水和生活平衡 坦诚面对 用 AI 高效完成本职工作 用省下的时间去享受生活 如果想要在职业上获得主导权:放下亲自敲代码的执念 把 AI 当作强大杠杆 去挑战并解决真正重要和有影响力的现实问题 看完了让我心里舒服了很多
Your job may be surviving on inertia. Your resentment won’t extend the runway. What do you actually want to do with the intelligence now at your disposal?
21
110
5
594
70,804
Thanks Google, Embedding Gemma 2 is a big deal 🫶 Multimodal embeddings (text, images, video, audio) can run on-device in the browser. No server, no API key, ~20–70 ms per query on WebGPU. Everything stays on your machine (ofc). Now go build (new) things with it 🚀
Introducing EmbeddingGemma 2, a new open multimodal model that sets the standard for on-device efficiency. - our first open, natively multimodal embedding model - handles text, code, image, video, and audio tasks within a lightweight, modular 740M parameter form factor - ideal for offline, privacy-first RAG when paired with Gemma 4 - outperforms some specialist models more than twice its size Weights available now on Hugging Face.
32
138
11
1,995
120,158
1
49
1
668
18,238
Apple says you need a Mac to build iPhone apps. Today I built one on Omarchy Linux, submitted it to the App Store, installed it on my iPhone over USB, and stopped it at a breakpoint in the debugger. No Mac anywhere. It's all open source: github.com/joshuaswarren/oma…
332
869
155
9,553
542,455
Yb123 retweeted
If you statically link Swift on Linux, Swift 6.4 has a Foundation issue worth knowing about. A fresh report shows swift build --static-swift-stdlib failing with undefined references when Foundation is involved. In this case it appeared while building a SwiftPM build-tool plugin on Linux ARM64. The underlying swift-build issue is already tracked. Franz Busch noted that a fix is in the next release branch. That distinction matters: the bug is known and the fix path exists, but it has not necessarily reached the Swift release you are running yet. For server-side Swift and CLI tooling, this is the kind of regression that can look like a package-specific linker problem even when the cause is lower in the toolchain. #SwiftLang #ServerSideSwift forums.swift.org/t/static-li…
6
3
19
1,165