Skip to content
View SimonAKing's full-sized avatar
🍅
PRESS START
🍅
PRESS START

Block or report SimonAKing

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
SimonAKing/README.md

Hey 👋, I'm Simon AKing!

- 👨‍💻 Front | Back | Left | Right | End Engineer

- ✨ Indie Hacker | Building What People Want

- 🤖 AI Enthusiast | Exploring AI-powered Products

Pinned Loading

  1. Context engineering 解读 - 07/08/2025 Context engineering 解读 - 07/08/2025
    1
    分享一个最近比较火的概念 “context engineering”(由 Andrej Karpathy 提出),一句话解释:在模型推理时 提供出 任务所需的所有完整信息的工程。
    2
    > 以下内容参考自 [Awesome-Context-Engineering](https://github.com/Meirtz/Awesome-Context-Engineering)
    3
    4
    为什么近期 context engineering 被广泛讨论?原因是 prompt engineering 不足以代表了现在生产级 LLMs 应用的复杂度。
    5
    首先 LLMs 的概率函数是:
  2. AI 时代的分发策略:速度即护城河 - 06/30/2025 AI 时代的分发策略:速度即护城河 - 06/30/2025
    1
    分享下 a16z 的近期一篇文章,对产品早期增长很有启发:[In Consumer AI, Momentum Is the Moat](https://a16z.com/momentum-as-ai-moat/)
    2
    > 标题:AI 时代的分发策略:速度即护城河,对应的[中文版](https://mp.weixin.qq.com/s?__biz=MzA3NzUxMzM5MQ%3D%3D&mid=2453892038&idx=1&sn=095208fd8d4e5bd20ac5a2f84ae4e0b8&chksm=8931266811d6e574943d635df64d85dc473e8fee0a390b062c2cf05f930b3c4d1808093d35da&mpshare=1&scene=1&sharer_shareinfo=1f779427ddbc399e68d7b53ab2eb5dda&sharer_shareinfo_first=1f779427ddbc399e68d7b53ab2eb5dda)
    3
    4
    文章的主旨是:LLMs 同质化趋势下,AI 产品 的护城河不再是技术壁垒 为主(技术的差异越来越小),而是增长的速度与产品认知差。
    5
    
                  
  3. 行业观察:模型即产品趋势 - 06/26/2025 行业观察:模型即产品趋势 - 06/26/2025
    1
    分享下最近的一个行业观察:
    2
    1. 国内 ai 模型厂商 minmax、kimi 最近有一些新动态:
    3
      1. minmax 推出了 通用 agent 产品 https://agent.minimax.io/share/281535490056355
    4
      2. kimi 推出了 researcher 模型与产品 https://www.163.com/dy/article/K30342N605566Y1D.html
    5
    2. 国外 claude 今天则进一步 artifacts 部署、发布等端到端流程 内置
  4. Andrej - AI 时代下的软件演进趋势分享 - 06/24/2025 Andrej - AI 时代下的软件演进趋势分享 - 06/24/2025
    1
    分享下最近 [Andrej](https://x.com/karpathy) 在 YC 上的一次分享,主题是 AI 时代下软件演进的趋势,其中有一些深刻的洞察:
    2
    该分享原内容:[视频](https://youtu.be/LCEmiRjPEtQ)[PPT](https://docs.google.com/presentation/d/1sZqMAoIJDxz79cbC5ap5v9jknYH4Aa9cFFaWL8Rids4/edit?usp=sharing)[中文版](https://docs.qq.com/document/DVmpCdVRSbFFycmhr)
    3
    
                  
    4
    1. 软件演进的三个时代:从传统的代码编程到 训练神经网络 再到自然语言编程,突然间 每个人都是程序员。
    5
        > 对应了现在 vibe coding 兴起,ai coding 产品风靡(该概念也是 Andrej 提出)
  5. 多模态研究(张祥雨)播客分享 - 06/09/2025 多模态研究(张祥雨)播客分享 - 06/09/2025
    1
    分享一期 最近被疯狂推荐的播客 - [102。 和张祥雨聊,多模态研究的挣扎史和未来两年的 2 个“GPT-4 时刻”](https://www.xiaoyuzhoufm.com/episode/683d2ceb38dcc57c641a7d0f?s=eyJ1IjoiNjU1ZDhiY2NlZGNlNjcxMDRhNDUyODcxIn0%3D)[文字版](https://mp.weixin.qq.com/s/892QuRPH9uP6zN6dS-HZMw))。
    2
    该播客是拾象 CEO 李广密对大模型公司阶跃星辰首席科学家张祥雨的访谈,其中张祥雨阐述了他近些年对 LLMs、MultiModal 领域的一些思考,我和 claude 总结了下收获比较大的点,欢迎讨论:
    3
    
                  
    4
    1. 自监督学习 在 NLP 与 CV 领域的差异
    5
      NLP 领域中 自监督训练 配合 decoder-only 架构取得了惊人成功。相比之下,CV 领域试图复制这种成功的尝试却遇到了瓶颈。无论是对比学习还是掩码图像建模(MIM),都没能展现出类似的 scaling 效果。这背后可能的深层原因:
  6. Memory 的几种类型 - 05/25/2025 Memory 的几种类型 - 05/25/2025
    1
    最近刷到了一张 agent 记忆的分类图,简单概括下:
    2
    短期记忆:记住刚刚说过的话
    3
    比如你问"我女儿多大了",它记得你刚说过是5岁,就能接着聊"哦,那该上幼儿园了"
    4
    
                  
    5
    长期记忆:存储重要信息、用户偏好