tears lee 绿柚子
916 posts

tears lee 绿柚子
@tears_lee
平平无奇的小透明/0.2%的右派0.8%的自由主义/素食喜辣/沉默观测/让子弹飞一会儿/现实主义/如果我爱国,爱到失业、衣食无着,那就叫不可持续爱国



面白いことに習近平には珍しく激昂して怒りが抑えきれない様子だったという 「怒ってみせた」わけではないらしい 感情を隠せないほど嫌悪が募るなんてことが、習近平にもあるんだね なぜなんだろう

AI price war is getting brutal > deepseek made its v4 Pro model cheaper for good > xiaomi’s mimo v2.5 Pro dropped prices by up to 99% at this point Chinese AI labs are basically giving intelligence away but wait how do these companies even make money when the models are this cheap?




Presented without comment.



China’s dazzling coastal skylines are a magnificent illusion masking a deeply fractured, economically divided nation. A data-driven report from The Economist completely dismantles the myth of China as a unified economic powerhouse, exposing a jaw-dropping internal chasm. The analysis reveals that China’s four richest cities, boasting a combined population of 84 million people, now command a nominal GDP per capita that exceeds Japan's national average. Yet, in stark contrast, its four poorest provinces, home to 140 million citizens, languish with per capita incomes closer to Vietnam. This internal chasm provides Beijing with a highly disruptive global trade advantage. Economists refer to this as a demographic "pterodactyl's downdraft," a massive force crushing traditional economic development models. Unlike historical economies that gradually abandoned low-end manufacturing as their coastal cities grew wealthy, China retains it all. Because of its massive regional wealth disparity, China can simultaneously run world-class, high-tech semiconductor hubs in its coastal megacities while utilizing its deeply impoverished inland provinces as low-wage, state-subsidized labor engines. It behaves like a developed Western nation and a struggling emerging market all at once. However, comparing China's absolute peak performers to Japan’s national average obscures the reality of true economic development. When evaluating peer-to-peer urban performance, Japan’s leading metros continue to decisively outperform China's premier hubs. While Beijing and Shanghai hover around $32,000 to $33,000 in nominal GDP per capita, Tokyo Prefecture towers over them at $48,000 to $52,000, with secondary Japanese centers like Kanagawa and Osaka also comfortably beating out China's top tier. Japan achieved a highly uniform, structurally sound, high-income society. China, conversely, remains a fragile economic paradox, masking hundreds of millions of low-income citizens behind the dazzling, state-funded facades of a few hyper-developed coastal fortresses. #UnveiledChina #ChinaInequality #TheEconomist #EconomicShock #SupertankerSuites #TokyoVsShanghai #GlobalTrade #TwoChinas




Yes, only with an honest conscience can one calmly face any investigation. At the end of 2025, I was subjected to long-term slander by Bob Fu's group(Xiqiu Fu group), labeled a "CCP United Front Work Department spy," and consequently underwent informal investigations by more than ten agents from various US law enforcement agencies. I always cooperated fully and proved my innocence. I will never commit suicide—because I firmly believe in the rule of law system in the United States based on the separation of powers and checks and balances.


国民党中央连夜发文,隔空叫阵赵少康:3800亿+N的军购方案就是“党版”,是党中央开会决定的!(具体内容见评论区) 赵少康假日召集记者,隔空叫阵党中央:季麟连必须开除!幕后指使者必须揪出来!既然是党版,为啥又要开会讨论?讨论了有人不同意你们又不听?以前党版也改过,况且民众党到时候提出来的跟你们的不一样,你们改不改?郑习会累积的一点红利,眼看就这样被消耗了,真的很可惜!

张驰,浙大本科,UCLA 博士(导师朱松纯),2025 年加入字节 Seed 做数学推理,干了一年后离职去北大当助理教授。最近上播客 Into Asia 回顾了在字节的一年,不少判断跟目前的公开叙事直接冲突。 他的核心观点: 1. 字节跑完一轮完整迭代(预训练+后训练)要大约半年,谷歌据传三个月。他认为迭代速度差距是追不上的根本原因。 2. Seed 内部 benchmaxxing 严重。领导按 benchmark 分数考核,大家都在刷榜,但他说纸面追平了不等于真的好用,「实际体验不行」。 3. 2024 年底 Seed 自认追平了 GPT-4o,结果 DeepSeek 一出来才发现差距还在。他入职时全组紧急转强化学习。 4. 他认为中美 AI 差距在扩大不是缩小。原话:「我甚至不同意中国在追赶这个说法,我们仍然远远落后。」同事和学生同意,但智谱、MiniMax 这些上市公司的领导层不会同意。 5. 蒸馏走捷径很普遍。很多公司直接调 Claude/GPT/Gemini 的输出当训练数据。不过他也承认 DeepSeek 在 V3/R1 上有真正的架构创新。 6. 字节主力芯片是 NVIDIA H20,最快的卡留给预训练和后训练团队。国产芯片有但没人用于训练。字节在海外采购新一代 NVIDIA 芯片,但「肯定不在中国大陆」。 7. 美国公司有用户反馈飞轮,模型好用→用户多→反馈好→模型更好。中国模型起步差,没人愿意用在重要工作上,数据拿不到,恶性循环。 8. 他在谷歌实习时觉得基础设施「太好了」,跟字节差距巨大。不只是芯片,训练框架和整个基础设施都差一截。 9. 中国 AI 从业者普遍用美国 Agent 工具。他自己用 Claude Code 和 Copilot,中国模型的编码 Agent 他评价「完全不实用」。字节海外团队直接用 Cursor。 10. Claude Code 好用到让他在想还要不要培养博士生,但又怕不培养下一代,以后没人做研究。 背景补充:张驰在字节只待了约一年,所在的数学组他自己说偏宣传性质,不在核心的预训练/后训练团队。他的观点是个人视角,不代表字节全貌。






Distilled recap of the back-and-forth with Jensen on export controls: Dwarkesh: Wouldn’t selling Nvidia chips to China enable them to train models like Claude Mythos with cyber offensive capabilities that would be threats to American companies and national security? Jensen: First of all, Mythos was trained on fairly mundane capacity and a fairly mundane amount of it by an extraordinary company. The amount of capacity and the type of compute it was trained on is abundantly available in China. Dwarkesh: With that, could they eventually train a model like Mythos? Yes. But the question is, because we have more FLOPs, American labs are able to get to this level of capabilities first. Furthermore, even if they trained a model like this, the ability to deploy it at scale matters. If you had a cyber hacker, it's much more dangerous if they have a million of them versus a thousand of them. Jensen: Your premise is just wrong. The fact of the matter is their AI development is going just fine. The best AI researchers in the world, because they are limited in compute, also come up with extremely smart algorithms. DeepSeek is not an inconsequential advance. The day that DeepSeek comes out on Huawei first, that is a horrible outcome for our nation. Dwarkesh: Currently, you can have a model like DeepSeek that can run on any accelerator if it's open source. Why would that stop being the case in the future? Jensen: Suppose it optimizes for Huawei. Suppose it optimizes for their architecture. It would put others at a disadvantage. As AI diffuses out into the rest of the world, their standards and their tech stack will become superior to ours because their models are open. Dwarkesh: Tesla sold extremely good electric vehicles to China for a long time. iPhones are sold in China. They didn't cause some lock-in. China will still make their version of EVs, and they're dominating, or smartphones, they're dominating. Jensen: We are not a car. The fact that I can buy this car brand one day and use another car brand another day is easy. Computing is not like that. There's a reason why x86 still exists. There's a reason why Arm is so sticky. These ecosystems are hard to replace. Dwarkesh: It's just hard to imagine that there's a long-term lock-in to the Chinese ecosystem, even if they have this slightly better open-source model for a while. American labs port across accelerators constantly. Anthropic's models are run on GPUs, they're run on Trainium, they're run on TPUs. There are so many things you can do, from distilling to a model that's well fit for your chips. Jensen: China is the largest contributor to open source software in the world. China's the largest contributor to open models in the world. Today it's built on the American tech stack, Nvidia’s. Fact. All five layers of the tech stack for AI are important. The United States ought to go win all five of them. in a few years time, I'm making you the prediction that when we want American technology to be diffused around the world—out to India, out to the Middle East, out to Africa, out to Southeast Asia—on that day, I will tell you exactly about today's conversation, about how your policy ... caused the United States to concede the second largest market in the world for no good reason at all.





















