s ke

[s ke @ske802286]

I've just registered for the xPoints rewards program and unlocked a permanent 20% points boost. Join me @xStocksFi

[Perle Labs @PerleLabs]

1/ Season 1 of Perle Labs is officially live.

AI doesn’t move forward without human-verified data.

And as AI systems move into high-stakes, real-world deployment, that data has to scale with accountability.

That’s exactlt what Perle Labs was built to do 🧵👇

[s ke @ske802286]

# https://t.co/CgJbtzT5uN

[Perle Labs @PerleLabs]

Embodied intelligence can’t afford bad data.

Perle Labs is building AI data infrastructure designed for this reality:

→ Human-verified, expert-validated work
→ On-chain auditability
→ Sovereign and enterprise-grade design

Built to benefit everyone.

[s ke @ske802286]

@PerleLabs # BULLISH PROJECT

[s ke @ske802286]

https://t.co/mp2x74DJEx

[s ke @ske802286]

https://t.co/xV8YgymvdV @opensea

[s ke @ske802286]

https://t.co/xNy8ZM4Vv2

[王短鸟 (长鸟版) @Wangduanniao]

撸这个CROSS吧，韩国项目，团队特别牛逼
是一个兼容eth的 Layer1 游戏专用区块链​​

目前主网已经上线了

需要下载一下CROSS钱包，可以在appstore下载

下载🔗 https://t.co/qNLtbiTgqy

然后在tg小程序里链接这个钱包
https://t.co/qNLtbiTgqy

还没有快照，现在玩来得及

后面会获得CROSS的空投

韩国人还是牛逼的，这个项目没有接受vc的投资，避免了高 FDV/低流通问题。

通过私募私募 1040万 + 公募 810万，共获得总计 $1850万融资

私募/公售价均为 ​​0.1U

看了一下一共发10亿个币，无增发了。

然后代币解锁跟币价挂钩，受 CZ 之前提到的代币经济学启发

创世团队特别牛逼

大家应该都知道《mir4》吧

就是CROSS ceo Henry Chang​​ 主导的之前

并且还投资过Odin、Nikke胜利女神、Night Crow

然后​​CTO HoEunOh 是前 WEMIX 技术负责人，20+ 年 MMORPG 及区块链架构经验，主导 WEMIX $100亿 FDV 生态搭建

总的来说目前就是下载钱包，链接到tg小程序，等快照，拿空投

并且目前该链支持的游戏Ragnarok: Monster World已经可以玩了

现在可以去APPstore下载玩

我玩了一下，好玩，游戏画面特别精良

体验丝滑，各我们之前玩的区块链游戏完全不同

主要是游戏好玩儿，所以跟大家推荐一下

[Binance @binance]

參與交易就有機會一起去東京，領Macbook Air 等多項大獎！

[s ke @ske802286]

https://t.co/xNy8ZM4Vv2

[s ke @ske802286]

https://t.co/qjfvZj02qC

[s ke @ske802286]

我刚刚在欧易领水中心获得了 0.1 TESTNET_MON！快加入获取更多测试网代币吧！ https://t.co/hnjiky3ZXg

[OP_CAT Layer | Mainnet Live @op_catlayer]

CAT is out of the bag https://t.co/XABOSX0EWe

[Aligned @alignedlayer]

Smashing on-chain verification costs

The current on-chain verification costs in Ethereum are high, reaching 100M USD per year. These costs come from data storage, access, and computation.

Aligned is poised to revolutionize the cost landscape of zero-knowledge proofs, potentially slashing the current high on-chain verification costs in Ethereum.

Aligned allows developers to verify and use the result of their proofs for just 2100 gas.

Curious to know how? Check out this thread:

[s ke @ske802286]

great!

[Babylon @babylonlabs_io]

🎉 Exciting news! Babylon has successfully completed a $70M raise led by Paradigm @paradigm to advance trustless Bitcoin staking. 🚀

1/9

[s ke @ske802286]

@spaceandtime

教你创建zkSync 的分析仪表板

https://t.co/ByqEPcOdnn

[s ke @ske802286]

@spaceandtime

Space and Time 信息图

[s ke @ske802286]

The Space and Time Studio is a comprehensive AI-powered dapp frontend for exploring blockchain data, running queries, and publishing to custom-branded dashboards. @spaceandtime

Check it out: https://t.co/8DHWMnj7RG

[s ke @ske802286]

@spaceandtime
博客链接： https://t.co/VoTBMqC9G0

翻译：https://t.co/Hjn1QLUly3

[s ke @ske802286]

RAG 依赖于一种特殊类型的数据库，称为向量搜索数据库。在 Proof of SQL 之上，Space and Time 还开创了 Proof of Vector Search，这样 LLMs 就可以以可验证的方式检索额外的上下文和新数据 ✅

构建下一代 LLM，向向量搜索成功迈进。由 Space and Time 提供动力。🔋

[s ke @ske802286]

@spaceandtime
https://t.co/JxKGbW72Rx
翻译：
检索增强生成 (Retrieval Augmented Generation, RAG) 是什么，以及为什么所有的 AI 大脑都在谈论它？

[Space and Time @spaceandtime]

Retrieval augmented generation. What is it, and why are all the AI gigabrains talking about it?

@chiefbuidl dives deep into RAG, its challenges and limitations, why it’s being adopted as an alternative to fine-tuning, and what the next gen of LLMs might look like in this SxT blog: https://t.co/Ise8eN7g8V

LLMs generate responses based on the data that they were trained on. When you prompt an LLM, it takes your input, tries to understand the context, and then retrieves what it predicts is the most relevant information to return back to you based on the weights—or levels of importance—assigned to different training datapoints.

Naturally, this means that its response is limited to what’s in its training dataset. If you ask it to quote Shakespeare or the Bible, no problem, but if you ask it to quote the presidential address given last night, it can’t. Fine-tuning doesn’t solve the problem either, because the process is so slow that models become obsolete just as quickly as they’re updated. LLMs need a way to access fresh data in real time.

That’s where RAG comes in.

When a response requires data beyond what the LLM was trained on, it leverages an external database that allows it to grab relevant context or fresh data that will help it generate a better response. This process is known as retrieval augmented generation, or RAG.

RAG relies on a specialized type of database called a vector search database. On top of Proof of SQL, Space and Time also pioneered Proof of Vector Search, so that LLMs can retrieve extra context and fresh data in a verifiable way ✅

Build the next gen of LLMs, and vector search to success. Powered by Space and Time. 🔋

[s ke @ske802286]

当响应需要超出 LLM 训练数据范围的数据时，它会利用一个外部数据库来获取相关的上下文或新数据，以帮助其生成更好的响应。这个过程被称为检索增强生成 (RAG)。