upper biff

follow back💯💯💯

[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 🧵👇

[upper biff @BiffUpper]

@PerleLabs # bullish project

[upper biff @BiffUpper]

# https://t.co/niplmJJQ4n

[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.

[upper biff @BiffUpper]

@PerleLabs # bullish PerleLabs

[upper biff @BiffUpper]

https://t.co/AAjdCO7nOl

[upper biff @BiffUpper]

https://t.co/HoKy1wei8W @opensea

[upper biff @BiffUpper]

https://t.co/igWTlTylXm

[Elon Musk @elonmusk]

Starlink now available in Sri Lanka! 🇱🇰

[upper biff @BiffUpper]

https://t.co/qEx5u5r0ub

[upper biff @BiffUpper]

https://t.co/vC8glvVlH3

[upper biff @BiffUpper]

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/VWJxIatEZd

[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:

[upper biff @BiffUpper]

great pioneer

[Babylon @babylonlabs_io]

https://t.co/kdBgiu12Fq🎇

[upper biff @BiffUpper]

@spaceandtime

手把手教你创建zkSync 的分析仪表板

https://t.co/LCZQZmhxKb

[upper biff @BiffUpper]

@spaceandtime

an infographic about how SxT

[upper biff @BiffUpper]

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/xaCancsYYe

[upper biff @BiffUpper]

@spaceandtime
博客：https://t.co/hMh0QiiPuJ

翻译：https://t.co/8jZABs0PlF

[upper biff @BiffUpper]

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。🔋

[upper biff @BiffUpper]

@spaceandtime
https://t.co/OXeVA9Pnxk
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？

@chiefbuidl 在这篇 SxT 博客中深入探讨了 RAG、其挑战和局限性、为何被作为微调的替代方案采用，以及下一代 LLM 可能的样子：https://t.co/tJu1iNZfw3

[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. 🔋

[upper biff @BiffUpper]

这就是 RAG 的作用。

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)。