Yilin Harry Lu
98 posts
Yilin Harry Lu
@AstropowerDev
UMN csci, math, econ triple honors major | Sensemaking4AI & Human-AI Scientific Discovery | VIS+NLP+AI4Sci Researcher
San Diego, CA Katılım Eylül 2022
879 Takip Edilen86 Takipçiler
刚用了毕业警告，openclaw成功把自己跑崩了🥲🥲🥲
Penny777 🦊@penny777
笑死，GitHub 上面有一个叫 PUA 的项目，目前已经 10k 的 Star 了。😂😂😂 核心功能就是在 AI 开始甩锅的时候，狠狠地 PUA 回去。让 AI 在羞愧之后进而发奋图强继续 Coding。 里面的核心话术味儿太正了，怀疑是阿里人写的。 「你这个 bug 都解决不了，让我怎么给你打绩效？」 「你的底层逻辑是什么？顶层设计在哪？抓手在哪？」 「慎重考虑决定给你 3.25。这个 3.25 是对你的激励。」 「别的模型都能解决。你可能就要毕业了。」 链接：github.com/tanweai/pua
中文
Yilin Harry Lu retweetledi
Breaking: Academic freedom no more.
The NeurIPS Foundation has announced it will no longer accept submissions from US-sanctioned institutions.
English
Yilin Harry Lu retweetledi
Congrats to the @cursor_ai team on the launch of Composer 2!
We are proud to see Kimi-k2.5 provide the foundation. Seeing our model integrated effectively through Cursor's continued pretraining & high-compute RL training is the open model ecosystem we love to support.
Note: Cursor accesses Kimi-k2.5 via @FireworksAI_HQ ' hosted RL and inference platform as part of an authorized commercial partnership.
English
Yilin Harry Lu retweetledi
If you want PhD, sleep.
If you want idea, sleep.
If you want offer, sleep.
If you want paper, sleep.
If you want reward, sleep.
If you want breakthrough, sleep.
If you want collaborator, sleep.
Please sleep well.
Bryan Johnson@bryan_johnson
If you aspire, sleep. If you doubt, sleep. If you create, sleep. If you worry, sleep. If you love, sleep. If you hurt, sleep. Sleep.
English
@peterwildeford Meta above CN labs is laughable, you also forgot the real frontier CN lab - bytedance seed.
English
Based on the data I see, I think:
- Anthropic🇺🇸/Google🇺🇸/OpenAI🇺🇸 all ~tied
- Meta🇺🇸 / xAI🇺🇸 each ~7mo behind
- Moonshot🇨🇳/- Deepseek🇨🇳 / zAI 🇨🇳 / Alibaba🇨🇳each ~9mo behind
- Mistral🇫🇷 ~1.5 years behind
- No other companies competitive
Ethan Mollick@emollick
Both xAI and Meta seem to be falling behind, based on the Grok 4.2 benchmarks and this reporting. Frontier AI models are really a three way race at this point.
English
Atlassian： 全球程序员和项目经理几乎每天都在用的协同开发与代码管理霸主（旗下有 Jira、Trello、Confluence 等）。
Canva（可画）： 用极简在线模板和疯狂叠加的AI功能，把专业设计门槛打成“白菜价”并直接威胁Adobe的视觉巨头。今年要在纳斯达克上市。
Afterpay： 凭借一己之力向全球年轻消费者普及了“先买后付”（BNPL）分期模式的金融科技鼻祖（后被美国 Block 收购）。
Immutable X： 专注于以太坊二层网络扩容，专攻Web3链游和NFT交易的加密货币核心基建。
Synthetix： 在DeFi领域里，极其硬核且处于核心生态位的去中心化衍生品流动性协议。
Envato： 为全球独立开发者和设计师提供代码库、网页主题等数字资产交易的庞大线上集市（旗下拥有 ThemeForest）。
Linktree： 仅仅靠解决“社交主页只能挂一个链接”的微小痛点起家，最终垄断全球大V流量分发的超级链接聚合器。估值数十亿美元。
WiseTech Global： 用一套极其复杂的系统（CargoWise），垄断了全球头部跨国物流和货运代理公司底层数据流的工业级SaaS巨头。
Altium： 全球硬件工程师画PCB电路图时绝对绕不开的底层设计软件王者（已被日本瑞萨电子收购）。
Appen（澳本）： 在AI大模型爆发早期，专门雇佣全球百万众包人员为硅谷巨头们做“喂饭工作”（数据清洗和人工标注）的超级数据包工头。
作为人口仅仅2000多万的国家，能够做出以上这些千亿美元级别的行业领军企业，算没有计算机行业吗？
天天网上离谱言论一个赛一个。
中文
高盛前 CEO 布兰克费因谈招人标准：
名校的平均水平和下限确实更高，但如果只看各校最顶尖那一批，非名校的尖子未必比哈佛差，甚至可能更强，因为他们往往是在更激烈的竞争里一路杀出来的。
中文
Yilin Harry Lu retweetledi
HCI venues need to seriously consider making their review processes more public than they are now. Going forward, less transparency may only hinder our ability to perform similar meta-analyses to diagnose problems as peer review demands expand.
ICLR@iclr_conf
We want to update the community on our response to concerns about low-quality and LLM-generated papers and reviews, and steps we are taking & will be taking blog.iclr.cc/2025/11/19/icl… We will follow up with another blog later on desk rejections and reviewer-related decisions!
English
Talking to your AI glasses is silly, so we built Reality Proxy, a direct manipulation interface that lets you instantly select real-world objects and share context with AI. A step toward JARVIS.
English
Yilin Harry Lu retweetledi
I moved to the US in 2001. Really changed my life. Truly a land of opportunity. With some luck, anyone from anywhere cud grab opportunities, work hard & accomplish things they never dreamed of. May the American dream eternally thrive. Happy 4th.
English
Yilin Harry Lu retweetledi
Ok a few quick things.
Most CS students who get into elite PhD programs in AI especially have already usually published multiple first author papers in "top" conferences. 1/
Suvansh Sanjeev@SuvanshSanjeev
i left my phd before joining openai working in industry demands more rigor – you don’t just need to convince reviewer 2 with a nice graph and an ego-cite, it better actually work if it’s underwriting billions in research investment not saying it always pans out that way in practice. yolo culture is pervasive and “research-quality code” abounds. i certainly don’t have a clean conscience there but some of the best breakthroughs come from people without academic preconceptions, with the discipline to build things that actually work
English
Yilin Harry Lu retweetledi
The reason I’m in America along with so many critical people who built SpaceX, Tesla and hundreds of other companies that made America strong is because of H1B.
Take a big step back and FUCK YOURSELF in the face. I will go to war on this issue the likes of which you cannot possibly comprehend.
English
Yilin Harry Lu retweetledi
Mechanistic interpretability is fascinating - but can it be useful? In particular, can it beat strong baselines like steering and prompting on downstream tasks that people care about?
The answer is, resoundingly, yes.
Our new blog post with @a_karvonen, Sieve, dives into the first counterfactual application of sparse autoencoders to a real downstream task, where we demonstrate Pareto dominance over existing methods.
A 🧵… ~1/8~
tilderesearch.com/blog/sieve
English
This time, the Nobel committee skipped the computer scientist and gave the prize directly to the machine.
Konstantin Sonin@k_sonin
The best Nobel prize ever! Congratulations to @DAcemogluMIT, @baselinescene, and my Harris colleague Jim Robinson!
English
@KexinHuang5 @NatureMedicine @marinkazitnik @payal_chandak @WangQianwenToo @AkhilVaidMD @jure @girish_nadkarni @BenGlicksberg Congrats, great work!
English
📢 Super excited to share our new study @NatureMedicine on developing and validating an explainable graph-based foundation model for drug repurposing, designed specially for rare diseases, which collectively affect 300 million patients globally!
🧵1/10
English