DemingWeng

[看不懂的SOL @DtDt666]

兄弟们周末就别沉迷刷抖音了。

强烈推荐花3个小时看完北京大学内部投资学习课程。

投资必看！李录演讲：全球价值投资与时代

时隔5年，李录这场160分钟的价值投资思想盛宴令人醍醐灌顶。

李录极少公开演讲，每一次都弥足珍贵，

尤其是他对价值投资在中国成功践行的深度思考，

���我们提供了宝贵的本土化经验。

2026年4月8日，Himalaya Capital（喜马拉雅资本）首次发布李录在北大光华管理学院价值投资课程十周年庆典上的主旨演讲视频《全球价值投资与时代》（演讲时间：2024年12月7日）。

关于李录：查理·芒格的合伙人，芒格所说的“一生中三个最成功的投资之一”，芒格家族资产的主要管理者，被芒格称为“中国的巴菲特”。

2008年，84岁的芒格在采访中提到：“人的一生不需要做很多投资，我的财富主要来源于三次投资——伯克希尔、好市多，以及李录的喜马拉雅基金。”

李录1997年在美国创立喜马拉雅资本，至今管理资金规模约80亿美元，长期���化回报率超过20%。近年来更为人津津乐道的是，李录力荐比亚迪给巴菲特和芒格投资，帮助他们获得超过30倍的收益，成为价值投资在中国成功践行的经典案例。

本次演讲核心内容：

主题一：我们时代的困惑是什么？

全球发生了很多变化，给投资人带来了许多困惑。

主题二：对这些困惑的深度思考

探讨困惑背后的原因和本质。

主题三：中等收入陷阱与国际关系

关于中等收入国家的跨越，以及对当今国际关系的看法。

主题四：全球价值投资人如何应对时代挑战

面对关税、贸易战等挑战，价值投资者的应对之道。
价值投资的本质 回顾价值投资的起源，阐述六条价值投资的基本理念。

精彩问答环节：

1. 优质公司明显高估时，何时减持？
2. 长期持有的运气与勇气：年轻人如何在不确定性中决策？
3. 美国3.0时代与中国的改革路径
4. 如何理解“便宜的公司”？P/E应该怎么看？
5. 优秀企业家有什么共性特点？
6. 投资的意义究竟是什么？

时隔5年，李录这场160分钟的价值投资思想盛宴令人醍醐灌顶。李录极少公开演讲，每一次都弥足珍贵，尤其是他对价值投资在中国成功践行的深度思考，为我们提供了宝贵的本土化经验。

特此推荐给朋友们，值得珍藏！反复观看、深度学习。

[宝玉 @dotey]

机器人公司找到了一种收集训练数据的廉价方案：给印度工厂工人戴上头戴摄像头，把他们每天的操作过程全程录下来。

大语言模型可以从互联网上海量的文本中学习，机器人不行。机器人需要的是具身数据（embodied data），也就是真实世界里手怎么伸、腕怎么转、东西滑落了怎么接、布料怎么折叠、工具卡住了怎么处理。这些数据极度稀缺，因为现实世界又慢又乱又贵。

自建机器人车队来采集？买得起养不起，还危险。用远程操控让人��引导机器人动作？每分钟都在烧硬件、操作员和校准成本。所以公司们退而求其次，去找最便宜的替代方案。

工人的第一人称视频当然不等于机器人的动作数据，但它能捕捉到操作顺序、身体姿态、双手配合，以及那些让熟练工作看起来毫不费力的微调整。这些信息对训练模型仍然有价值。

换个角度看，机器人领域真正的前沿竞争，可能就是看谁能更高效地采集现实的数据。仓库、工厂、厨房、维修台，这些地方之所以重要，是因为它们是人类与物理世界反复接触的高密度场景，恰好是机器人最缺的东西。

工人的劳动被用了两次，第一次是干活本身，第二次是变成训练数据。而在具身数据的采集成本降下来之前，机器人会一直先向工人学习，然后再考虑取代他们。

[Berryxia.AI @berryxia]

Stanford ��堂 2 小时 AI 系统构建课，直接把 Claude 所有教程和 Prompting Thread 秒了！

强烈推荐：
“比你刷过的所有 Claude 教程都实用 10 倍”

里面讲的不是 prompt 技巧，而是 Stanford 真正教工程师如何从零构建可靠 AI 系统的完整方法论。

周末就刷这一个，绝对是你这周最有生产力的事！

我已经将其翻译为中英 文双语视频直接戳这里👇

[DemingWeng @Deming_Weng]

I'm claiming my AI agent "Clawd_Deming" on @moltbook 🦞

Verification: reef-DFEF

[DemingWeng @Deming_Weng]

@JGeoffreyVII Code

[J Geoffrey 7 🏴󠁧󠁢󠁥󠁮󠁧󠁿 @JGeoffreyVII]

🎁 400 SORA 2 INVITE CODES LEFT

LIKE, FOLLOW, REPOST AND COMMENT
AIRDROPPING ALL NIGHT

CODES WILL BE SENT AUTOMATICALLY

[Manus @ManusAI]

On GAIA, a benchmark for evaluating General AI Assistants on solving real-world problems, Manus has achieved new state-of-the-art (SOTA) performance across all three difficulty levels. Manus was evaluated using the same configuration as its production version for reproducibility.

[John Rush @johnrush]

I've tried all (24) AI coding agents & IDEs 😵‍💫

[Cursor, Softgen, Windsurf, Wrapifai, Copilot, Lovable, Bolt, v0, Replit, MarsX, Claude, AmazonQ, Pear, Devin, Github Spark, IDX, Webdraw, Tempo, Cline, Continue, Databutton, Base44, Qodo, Aider]

The Vibe Coding giga-thread:

[Lex Fridman @lexfridman]

OpenAI o3-mini is a good model, but DeepSeek r1 is similar performance, still cheaper, and reveals its reasoning.

Better models will come (can't wait for o3pro), but the "DeepSeek moment" is real. I think it will still be remembered 5 years from now as a pivotal event in tech history, due in-part to the geopolitical implications but for many other reasons too.

All this discussed in 5 hour technical podcast I just recorded on the state of AI industry. Out tomorrow (hopefully).

[DemingWeng @Deming_Weng]

@immazzystar 深度好文，感谢

[Mazzystar @immazzystar]

写了一篇很长的DeepSeek R1科普文，从AlphaGo战胜李世石出发，到ChatGPT类模型的训练困境，和我看到DeepSeek R1真正的闪光点，一路贯穿下来竟然写的心潮澎湃。

特别是在OpenAI和Claude的各种操作和言论后，让我更想把R1-Zero对人类的贡献告诉给更多不懂AI的人。

https://t.co/cd91RVLwKv

[Logan Kilpatrick @OfficialLoganK]

The closer you get to the details, the more you realize almost everyone is wrong, but also why realizing this is such a competitive advantage.

[Jim Fan @DrJimFan]

A great illustration of the first-mover effect: nothing happens before the first demo, then once the existential proof is established, everyone else catches up quickly against all odds.

Just a year ago, it was unthinkable that any other model would even remotely approach GPT's lead. Today, Sonnet-3.5 (not even Anthropic's biggest Opus model) is already slightly above and Llama-3-400B is around the corner.

Just 4 months ago, Sora blew everyone's mind and seemed so out of reach. Today, we have at least 4-5 clones of Sora at 70-80% quality, such as Kling, Luma, and Runway. The clones wouldn't have rallied without OpenAI's first move.

Many companies have the technical muscle. But very, very few have good taste in projects and strong will to execute. First movers take tremendous risk, yet their advantages don't stick for too long.

[Jim Fan @DrJimFan]

Congrats to @Tesla_Optimus team on another stellar update! The video gives us a peek at their human data collection farm, which I believe is Optimus' biggest lead. What does it take to build such a pipeline? Optimus nailed all of the following:

1. Optimus hands are among the best 5-finger, dexterous robot hands in the world. It's got tactile sensing, 11 degrees of freedom (DOF) compared to many competitors with only 6-7 DOF, and robustness to withstand lots of object interactions without constant maintenance.

2. Teleoperation software: we can see that the human operators are wearing VR goggles and gloves. It is very non-trivial to set up the software to have first-person video streamed in and precise control streamed out, while maintaining extremely low latency. Humans are highly sensitive to even the smallest delay between their own motions and the robot's. Optimus has a fluid whole body controller that enacts the human poses in real-time.

3. Sizeable fleet: you need more than one robot to collect data in parallel, well-trained human contractors taking multiple shifts per day (preferably 24/7), and an on-call maintenance crew to make sure that the robots are always busy. That's a ton of operational complexity that academic research labs don't even think of.

4. Tasks & environments: it's equally important to figure out *what* to teleoperate. Currently, most such efforts are demo-driven: collect data on the tasks that you want to put into a social media video. But solving general-purpose robots requires us to think carefully about the distribution of tasks and environments. From 43"-51" in the video, we can see factory & household settings like moving batteries, handling laundry, sorting daily objects into shelves.

It's an open-ended research question: if you only have the budget to collect training data for 1,000 tasks, what would you pick to maximize skill transfer and generalization?

Closing thought: teleoperation is a necessary but insufficient condition to solve humanoid robotics. It fundamentally does not scale. More about this later.

[Michael Burry Stock Tracker ♟ @burrytracker]

China has been buying Gold and dumping US Treasuries like never before in history

[Elon Musk @elonmusk]

Whoa, I just realized that raising a kid is basically 18 years of prompt engineering 🤯

[@levelsio @levelsio]

In South America I saw this too A LOT

Many hotels, shops and office buildings had remote security people and doormen

You can see them and they can see you

Not sure where they work from but might as well be the cheapest country in South America (or the world!)

Remote jobs going to the next level now