Aobo Cheng (@12oboo) 's Twitter Profile
Aobo Cheng

@12oboo

AI Engineer @TikTok_us

ID: 4557659304

linkhttps://github.com/obooman calendar_today14-12-2015 15:50:15

463 Tweet

54 Followers

1,1K Following

Aobo Cheng (@12oboo) 's Twitter Profile Photo

Before : user + application Now: agent + context + evaluator Context includes methods and informations. Will evaluator keep existing or will gone in the future?

Aobo Cheng (@12oboo) 's Twitter Profile Photo

Since we all agree that agent can do more things with longer time and agentic process, the communication between human and ai should be async now.

Andrej Karpathy (@karpathy) 's Twitter Profile Photo

Good post from Balaji on the "verification gap". You could see it as there being two modes in creation. Borrowing GAN terminology: 1) generation and 2) discrimination. e.g. painting - you make a brush stroke (1) and then you look for a while to see if you improved the

Bas van der Ploeg (@basvanderploeg) 's Twitter Profile Photo

I saw a YouTube video about a physical media cartridge player made by The Stock Pot. 🔥 After a few small adjustments, I made my own. 🤓 The cartridges contain NFC chips that trigger automations when inserted into the player. #3DPrinting #Apple

I saw a YouTube video about a physical media cartridge player made by The Stock Pot. 🔥

After a few small adjustments, I made my own. 🤓

The cartridges contain NFC chips that trigger automations when inserted into the player.

#3DPrinting #Apple
swyx (@swyx) 's Twitter Profile Photo

"everything that makes agents good is context engineering" excited to release dex's talk at AI Engineer, coiner of Context Engineering which has captured the zeitgeist of some of the most important problems in AI Engineering today!

"everything that makes agents good is context engineering" 

excited to release <a href="/dexhorthy/">dex</a>'s talk at <a href="/aiDotEngineer/">AI Engineer</a>, coiner of Context Engineering which has captured the zeitgeist of some of the most important problems in AI Engineering today!
Aobo Cheng (@12oboo) 's Twitter Profile Photo

PLAN is highly related with context condensation, the way make an agent doing its best is a good design on designing of PLAN and context condensing in the thread

Aobo Cheng (@12oboo) 's Twitter Profile Photo

Code is the bridge between LLM generation randomness and expected accuracy - not the same content, but the same result, what do you think

Aobo Cheng (@12oboo) 's Twitter Profile Photo

It’s getting pretty clear now, you are building models, or you are calling apis to build product, made a choice

Anthropic (@anthropicai) 's Twitter Profile Photo

New on the Anthropic Engineering blog: tips on how to build more efficient agents that handle more tools while using fewer tokens. Code execution with the Model Context Protocol (MCP): anthropic.com/engineering/co…