Duyi Pan

[t.toda @Trtd6Trtd]

https://t.co/jqkxjjFMUn

ハルシネーション対策＆説明性の文脈でナレッジグラフをベースにしたGraphRAGが注目されているが、ナレッジグラフのスキーマや内部構造がブラックボックスだと意味なくね？って問題提起

これを解決する再学習不要のパイプラインを提案

まだ理解できてないのだが、ナレッジグラフブラックボックス問題を部分探索問題として定式化してNP困難であることを証明してるっぽい

[cool ai and ml papers @aimodelsfyi]

BubbleRAG explores unknown knowledge graphs by expanding outward from semantic anchors instead of assuming you know the schema — helps LLMs retrieve the right facts when graph structure is a mystery

https://t.co/kENNTPMsi7

[BangLiu @BangL93]

🧠264 pages and 1416 references chart the future of Foundation Agents.

Our latest survey dives deep into agents—covering brain-inspired cognition, self-evolution, multi-agents, and AI safety.

Discover the #1 Paper of the Day on Hugging Face👇:

https://t.co/CvtqGKyCYb

1/3

[Alexander Wu @AlexanderWu0]

generating tiktok UI by https://t.co/P7f3c8u4l8
prompt is all you need

[Jiayi Zhang @didiforx]

Reasoning models lack atomic thought ⚛️

Unlike humans using independent units, they store full histories🤔

Introducing Atom of Thoughts (AOT): lifts gpt-4o-mini to 80.6% F1 on HotpotQA, surpassing o3-mini and DeepSeek-R1 !

The best part? It's plugs in for ANY framework 🔌
1/5

[MetaGPT @MetaGPT_]

20 Months: 0 → 7 papers (2 ICLR orals) & 40+ institution collabs.
With a clear vision, we're building the open-source foundation for tomorrow's agents.
We also release MGX (https://t.co/AyzVUF9dH3) and commit to open-source its core soon.
Check threads for what we've built!
1/8

[MetaGPT @MetaGPT_]

Introducing MGX (MetaGPT X), The First AI Dev Team.

· Chat with the AI team leader, product manager, architect, engineer, and data analyst 24/7 to create websites, blogs, shops, analytics, games, or anything else you can imagine.
· Build, deploy, share, and remix various creations in a new App World.
· Powered by the world's first multi-agent programming framework, and ranked #1 LLM Agent at ICLR2024.

✨Try it free: https://t.co/n2x3FQzzx9

BTW, we'll open source MetaGPT v1.0 soon.
#MGX #MetaGPT #DevTools #Agent #Coding #Development

[MetaGPT @MetaGPT_]

🚀 Coming Soon: MGX (MetaGPT X).

Swing by our Product Hunt teaser page and click “Notify Me”: https://t.co/YkUSLA7IRl via @kevinwdavid

#ProductHunt #Launch #MetaGPT #MGX

[MetaGPT @MetaGPT_]

🚀 Tired of time-consuming manual agentic workflow design? Let AFLOW take care of everything!

We just open-sourced AFLOW, which automates agentic workflow generation and optimization!

📄 Paper: https://t.co/O11mJwo0ci
💻 Code: https://t.co/fP0cu0aRYT

Try AFLOW now! 🌟

[MetaGPT @MetaGPT_]

🌟 Excited to open-source SELA, a powerful experimentation system integrating MCTS with LLM agents. Across 20 datasets, SELA achieves a 75% win rate against AIDE (OpenAI's top pick in MLE-Bench) and beats traditional AutoML methods developed over years.
💻 Code: https://t.co/dIjSaXnoii
📄 Paper: https://t.co/8j5DLpNr9N

[MetaGPT @MetaGPT_]

Gratefully thank all of our authors! @YizhouChi @Azzzz0Pan @ariaailab @rayhao2024 @BangL93 @AlexanderWu0

[Duyi Pan @Azzzz0Pan]

@MetaGPT_

[MetaGPT @MetaGPT_]

🤖 Can an AI agent design another AI?
🌟 Yes. SELA uses MCTS to design AI, achieving SoTA performance on 20 machine learning datasets. It can learn from past designs and experiments to create better AIs. It's fully open-source.

🚀 Try SELA yourself
📄 Paper: https://t.co/8j5DLpMTkf
💻 Code: https://t.co/pSDAhB4kAJ

[Mingchen Zhuge @MingchenZhuge]

🔔 new 𝗔𝗴𝗲𝗻𝘁-𝗮𝘀-𝗮-𝗝𝘂𝗱𝗴𝗲 paper:

𝗖𝗮𝗻 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗲 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗮𝘀 𝗲𝗳𝗳𝗲𝗰𝘁𝗶𝘃𝗲𝗹𝘆 𝗮𝘀 𝗵𝘂𝗺𝗮𝗻𝘀? 𝗬𝗲𝘀, 𝘁𝗵𝗲𝘆 𝗰𝗮𝗻!

📄 https://t.co/puJUH0uYfI
👨‍💻 https://t.co/W4jjtiXGw7

Introducing 𝗔𝗴𝗲𝗻𝘁-𝗮𝘀-𝗮-𝗝𝘂𝗱𝗴𝗲, a groundbreaking proof-of-concept that reduces costs and time by 97%, while providing rich, intermediate feedback. It precisely captures the natural step-by-step processes of agentic systems. We also developed 𝗗𝗲𝘃𝗔𝗜, a new benchmark featuring 55 automated AI development tasks and 365 requirements. Agent-as-a-Judge not only outperforms LLM-as-a-Judge but also closely mirrors human evaluations with greater efficiency and precision.

The real game-changer? It provides reliable reward signals, paving the way for scalable, self-improving agentic systems.

Thanks my Meta/KAUST mentors/peers/collaborators @SchmidhuberAI @tydsh @zechunliu @vikasc @YoungXiong1 @vikasc @Obs01ete @erniecyc @oneDylanAshley ...

[Jiayi Zhang @didiforx]

🔥🔥🔥Building optimal Workflows with minimize human effort！

[Duyi Pan @Azzzz0Pan]

@MetaGPT_ congrats！

[MetaGPT @MetaGPT_]

🚀 We are thrilled to introduce AFlow: Automating Agentic Workflow Generation

AFLOW autonomously generates & optimizes workflows using Monte Carlo Tree Search:

• Beats human-designed workflows in Coding, Math & QA
• Achieves GPT-4o-level on HumanEval at just 4.55% of its cost
• 96.2% HumanEval accuracy using GPT-4o with AFLOW
• Generate custom workflows in 1.5h with just an eval function

Explore more:
📄 Paper: https://t.co/O11mJwo0ci
💻 Code will be available at https://t.co/zD26C4BtxJ