Sam🇺🇦

[Ishan Sharma @Ishansharma7390]

Craig Federighi on why apple hasnt yet built ai agents into the iPhone

[Sam🇺🇦 @igenerativeai]

@0xZiye 上个月我刚买完这个，在澳大利亚和新西兰网速不错，回到中国大陆确实不行，经常断联，就是突然网速超级慢。

[Sam🇺🇦 @igenerativeai]

🇦🇺 Quick stop in Sydney 🏃🏼‍♀️‍➡️🏃🏻‍➡️Started the day with an early 5.8km city walk and finally stood in front of the Sydney Opera House — a place I’ve been staring at on world maps since I was a kid 🗺️

#australia #sydney #digitalnomad

[Sam🇺🇦 @igenerativeai]

🇳🇿 Lost in Queenstown — the adventure capital of the world. Is this real life or just a game? 🏞️

#new_zealand #queenstown #digitalnomad

[Sam🇺🇦 @igenerativeai]

🌏 Continent #5. Country #16. 🇦🇺First stop: Melbourne — and wow, this city has soul. Creative, diverse, and endlessly charming. 🐨

#digitalnomads #traveler #melbourne

[Sam🇺🇦 @igenerativeai]

@Stv_Lynn 👍水平很高，不过这个Unsplash的链接，好像已经不能用了，跳转不过去。

[Sam🇺🇦 @igenerativeai]

🤩We missed this course last year, but it's back this year!

[Everton Carneiro @everton_dev]

The wait is over: The best iOS development course is back!

This top-tier iOS development course has received a fresh update. Stanford’s legendary CS193p has published the opening lecture from its Spring 2025 session on YouTube.

[Sam🇺🇦 @igenerativeai]

@waylybaye 🤔你这么一说，我想起来3年前微软组织过一次分享会，当时有个布道的老兄，直接在https://t.co/M0QK2iD4Dw上噼里啪啦的写并且调试，感觉也挺流畅。似乎一直都可以，哈哈

[Claude @claudeai]

New in Xcode 26 Beta 7: Users can now log into their Claude account from Xcode.

Generate code and documentation, automatically fix errors, and create inline playgrounds using coding intelligence features in Xcode, powered by Claude Sonnet 4.

[Sam🇺🇦 @igenerativeai]

@rudrank Can I ask what you used to make this picture? Like what app or website?

[Sam🇺🇦 @igenerativeai]

@JJYing @cursor_ai 差不多10年前就看你写的博客，但是你的博客就很漂亮。这个App同样很美观👍

[Sam🇺🇦 @igenerativeai]

🕯️RIP Bill

[Mr. Macintosh @ClassicII_MrMac]

Today is a very sad day for the Apple community

Bill Atkinson has died at the age of 74😢

He was Apple employee number 51 and the principal designer and developer of the Lisa GUI. He was also among the first 30 members of the original Apple Macintosh development team. RIP Bill

[Sam🇺🇦 @igenerativeai]

@tualatrix 这地方Uber老贵了，坐地铁倒公交可便宜了

[Sam🇺🇦 @igenerativeai]

@Pepperscaling DIA me

[Sam🇺🇦 @igenerativeai]

@evepredict @dusan_panic @Big_Uppy @exolabs Is it a "series" method? Or is there a central node and the other 6 are connected to this central node?

[Epic Maps 🗺️ @theepicmap]

Points Point Nemo is the remotest place in the ocean and home of the Spaceship Cemetery.

[Sam🇺🇦 @igenerativeai]

@yuashwe 🙏先生一路走好🕯️

[宝玉 @dotey]

LangChain：构建与数据对话的聊天机器人5——检索

#LangChain构建与数据对话的聊天机器人

这节课主要内容是关于如何利用LangChain检索数据：

1. 语义检索：该章节开始于对语义检索的讨论，它在很多场景中表现出色，但在处理某些边缘案例时可能存在问题。故而引出了本课程主题 - 高级检索方法。

2. 最大边缘相关性（MMR）：MMR旨在通过增加检索结果的多样性，以防止漏掉重要信息。该方法首先根据语义相似性返回一组初始响应，然后在该小型文档集中选择既相关又具有多样性的文档作为最终呈现给用户的结果。

3. 自我查询：当查询问题包含特定元数据时，自我查询很有用。例如，对于问题“1980年有哪些关于外星人的电影？”，除了需要语义查询外，还需要对元数据（1980年）进行过滤。我们可以让语言模型自动将原始问题分割成过滤器和搜索项。

4. 压缩：压缩可帮助提取检索段落中最相关的部分，而不是返回整个文档。它需要额外的语言模型调用，但可以更好地聚焦在最重要的信息上。

5. MMR、自我查询和压缩可以结合使用以获得最优的检索结果。

6. 其他检索方法：课程还提到了其他不依赖于向量数据库的检索方法，例如SVM检索器和TF-IDF检索器。这些是传统NLP或机器学习中的技术。

课程地址：https://t.co/ptBYvgEqYP

播放列表：
https://t.co/QieycGYOFK
B站：https://t.co/PbiYI6z7Zb

[翁天信 @wengtianxin]

梅里雪山