<feed xmlns="http://www.w3.org/2005/Atom"> <id>https://siluy.github.io/</id><title>Siluy's Blog:)</title><subtitle>Where to log my life.</subtitle> <updated>2026-01-29T21:12:53+08:00</updated> <author> <name>siluy</name> <uri>https://siluy.github.io/</uri> </author><link rel="self" type="application/atom+xml" href="https://siluy.github.io/feed.xml"/><link rel="alternate" type="text/html" hreflang="zh-CN" href="https://siluy.github.io/"/> <generator uri="https://jekyllrb.com/" version="4.4.1">Jekyll</generator> <rights> © 2026 siluy </rights> <icon>/assets/img/favicons/favicon.ico</icon> <logo>/assets/img/favicons/favicon-96x96.png</logo> <entry><title>How China’s LLM “Open-Source Boom” Masks a Historical Lag</title><link href="https://siluy.github.io/posts/How-China-s-LLM-Open-Source-Boom-Masks-a-Historical-Lag/" rel="alternate" type="text/html" title="How China’s LLM “Open-Source Boom” Masks a Historical Lag" /><published>2026-01-29T20:00:00+08:00</published> <updated>2026-01-29T20:00:00+08:00</updated> <id>https://siluy.github.io/posts/How-China-s-LLM-Open-Source-Boom-Masks-a-Historical-Lag/</id> <content type="text/html" src="https://siluy.github.io/posts/How-China-s-LLM-Open-Source-Boom-Masks-a-Historical-Lag/" /> <author> <name>siluy</name> </author> <category term="随笔" /> <category term="中文" /> <summary>DeepSeek，Qwen，GLM，Kimi是开源最强的四个模型，全是中国，没有一个是美国的。中国这些大模型公司是远远被低估而不是高估。最主要的原因是市场不同，融资环境不同。这些公司如果在美国，就是第二个OpenAI。 这句话在当下极具煽动性，尤其是在 Llama 4 崩盘、Mark Zuckerberg 问责 Yann LeCun 和田渊栋、GLM-4.7 与 Kimi k2.5 接连发布、OpenAI 被不断指责为 ClosedAI 的 2026 年初，这种论调似乎成了公理。 然而，我却想说：中国模型确实因为地缘与市场因素在商业估值上被低估，但恰恰又因为“开源”这一标签，在技术统治力上被高估了。 简单的地理置换无法复制 OpenAI，因为历史的奇点无法重演。 一、 开源的虚假繁荣与情绪红利 中国开源模型的崛起，首先是一场情绪上的成功。 DeepSeek V3 和 R1 ...</summary> </entry> <entry><title>The New Lesson from Sutton</title><link href="https://siluy.github.io/posts/The-New-Lesson-from-Sutton/" rel="alternate" type="text/html" title="The New Lesson from Sutton" /><published>2025-11-29T22:00:00+08:00</published> <updated>2025-11-29T22:00:00+08:00</updated> <id>https://siluy.github.io/posts/The-New-Lesson-from-Sutton/</id> <content type="text/html" src="https://siluy.github.io/posts/The-New-Lesson-from-Sutton/" /> <author> <name>siluy</name> </author> <category term="随笔" /> <category term="中文" /> <summary>“The biggest lesson that can be read from 70 years of AI research is that general methods that leverage computation are ultimately the most effective, and by a large margin.” — Rich Sutton, The Bitter Lesson 在 70 年的 AI 研究中，那些利用计算能力的通用方法，最终不仅是最有效的，而且以巨大的优势胜出。这是 2019 年 Sutton 在知名的《The Bitter Lesson》中提出的观点。 但时至今日，越来越多的人因为他针对 LLM 释放的消极态度开始询问：这位苦涩教训的提出者本人，是不是开始走向反对教训的方向了呢？座谈时的心理系老师也不例外，向他提出了这个...</summary> </entry> <entry><title>The Future of AI - The Era of Experience and the Age of Design</title><link href="https://siluy.github.io/posts/The-Future-of-AI/" rel="alternate" type="text/html" title="The Future of AI - The Era of Experience and the Age of Design" /><published>2025-11-29T19:00:00+08:00</published> <updated>2025-12-02T19:05:18+08:00</updated> <id>https://siluy.github.io/posts/The-Future-of-AI/</id> <content type="text/html" src="https://siluy.github.io/posts/The-Future-of-AI/" /> <author> <name>siluy</name> </author> <category term="Lecture Note" /> <category term="中文" /> <summary>Lecture Notes: The Future of AI - The Era of Experience and the Age of Design Speaker: Prof. Rich Sutton Subject: Scientific trends in AI, AI Politics, AI Philosophy 一、Scientific trends in AI 1. AI 发展的两个时代 (The Eras of AI) 1. 人类数据时代 (The Era of Human Data) - 当前阶段 现状：2020 年代的 AI 发展非常好，主要依赖人类数据。 原理：训练 AI 预测人类的下一个词或标签，通过人类专家（偏好和示例）进行微调。 本质：将人类现有的知识迁移到静态的（非学习型）AI 中。 局限性： 高质量的...</summary> </entry> <entry><title>Mastering the Art of Sorting without Algorithmic Knowledge</title><link href="https://siluy.github.io/posts/Mastering-the-Art-of-Sorting-without-Algorithmic-Knowledge/" rel="alternate" type="text/html" title="Mastering the Art of Sorting without Algorithmic Knowledge" /><published>2025-11-18T19:00:00+08:00</published> <updated>2025-12-02T19:05:18+08:00</updated> <id>https://siluy.github.io/posts/Mastering-the-Art-of-Sorting-without-Algorithmic-Knowledge/</id> <content type="text/html" src="https://siluy.github.io/posts/Mastering-the-Art-of-Sorting-without-Algorithmic-Knowledge/" /> <author> <name>siluy</name> </author> <category term="spoof" /> <category term="eng" /> <summary></summary> </entry> <entry><title>University Students are Zero-Shot Learners</title><link href="https://siluy.github.io/posts/University-Students-are-Zero-Shot-Learners/" rel="alternate" type="text/html" title="University Students are Zero-Shot Learners" /><published>2025-06-13T04:00:00+08:00</published> <updated>2025-12-02T19:05:18+08:00</updated> <id>https://siluy.github.io/posts/University-Students-are-Zero-Shot-Learners/</id> <content type="text/html" src="https://siluy.github.io/posts/University-Students-are-Zero-Shot-Learners/" /> <author> <name>siluy</name> </author> <category term="spoof" /> <category term="eng" /> <summary></summary> </entry> </feed>
