Mingchen Zhuge (@MingchenZhuge) / X

Mingchen Zhuge

354 posts

Mingchen Zhuge

@MingchenZhuge

Founding member @recursive_si; PhD of @SchmidhuberAI; Ex @Meta @Microsoft; Build @MetaGPT_, gptswarm.org, agent-as-a-judge.com, nn.computer, ICLR RSI 2026, etc.

Mountain View, CA

Joined October 2022

Pinned
Mingchen Zhuge
@MingchenZhuge
Apr 10
🫱 Introducing 𝐍𝐞𝐮𝐫𝐚𝐥 𝐂𝐨𝐦𝐩𝐮𝐭𝐞𝐫s: 𝐰𝐡𝐚𝐭 𝐢𝐟 𝐀𝐈 𝐝𝐨𝐞𝐬 𝐧𝐨𝐭 𝐣𝐮𝐬𝐭 𝐮𝐬𝐞 𝐜𝐨𝐦𝐩𝐮𝐭𝐞𝐫𝐬 𝐛𝐞𝐭𝐭𝐞𝐫, 𝐛𝐮𝐭 𝐛𝐞𝐠𝐢𝐧𝐬 𝐭𝐨 𝐛𝐞𝐜𝐨𝐦𝐞 𝐭𝐡𝐞 𝐫𝐮𝐧𝐧𝐢𝐧𝐠 𝐜𝐨𝐦𝐩𝐮𝐭𝐞𝐫 𝐢𝐭𝐬𝐞𝐥𝐟? Beyond today's conventional computers, agents, and
218K
Mingchen Zhuge
@MingchenZhuge
Oct 16, 2024
🔔 new 𝗔𝗴𝗲𝗻𝘁-𝗮𝘀-𝗮-𝗝𝘂𝗱𝗴𝗲 paper: 𝗖𝗮𝗻 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗲 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗮𝘀 𝗲𝗳𝗳𝗲𝗰𝘁𝗶𝘃𝗲𝗹𝘆 𝗮𝘀 𝗵𝘂𝗺𝗮𝗻𝘀? 𝗬𝗲𝘀, 𝘁𝗵𝗲𝘆 𝗰𝗮𝗻! 📄 arxiv.org/abs/2410.10934… 👨‍💻 github.com/metauto-ai/age… Introducing 𝗔𝗴𝗲𝗻𝘁-𝗮𝘀-𝗮-𝗝𝘂𝗱𝗴𝗲, a
198K
Mingchen Zhuge
@MingchenZhuge
May 1, 2025
𝗔𝗴𝗲𝗻𝘁-𝗮𝘀-𝗮-𝗝𝘂𝗱𝗴𝗲 has been accepted at ICML 2025! Many 𝗔𝗴𝗲𝗻𝘁-𝗮𝘀-𝗮-𝗝𝘂𝗱𝗴𝗲 papers will get accepted by different conferences this year—reason is simple, this concept can apply to various domains and can help generate high-quality agentic datasets for
Mingchen Zhuge
@MingchenZhuge
Oct 16, 2024
🔔 new 𝗔𝗴𝗲𝗻𝘁-𝗮𝘀-𝗮-𝗝𝘂𝗱𝗴𝗲 paper: 𝗖𝗮𝗻 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗲 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗮𝘀 𝗲𝗳𝗳𝗲𝗰𝘁𝗶𝘃𝗲𝗹𝘆 𝗮𝘀 𝗵𝘂𝗺𝗮𝗻𝘀? 𝗬𝗲𝘀, 𝘁𝗵𝗲𝘆 𝗰𝗮𝗻! 📄 arxiv.org/abs/2410.10934… 👨‍💻 github.com/metauto-ai/age… Introducing 𝗔𝗴𝗲𝗻𝘁-𝗮𝘀-𝗮-𝗝𝘂𝗱𝗴𝗲, a
29K
Mingchen Zhuge
@MingchenZhuge
May 4, 2025
𝗧𝗼𝗽 𝘀𝗲𝗰𝗿𝗲𝘁 𝗔𝗴𝗲𝗻𝘁-𝗮𝘀-𝗮-𝗝𝘂𝗱𝗴𝗲 can be a great open-source #DeepWiki by just adding 2 code files swap github → openwiki on any repo URL 🫱 github.com/metauto-ai/age…
12K
Mingchen Zhuge
@MingchenZhuge
May 11, 2025
👋 GPTSwarm is all you need (ICML 2024 Oral) @karpathy GPTSwarm.org Built in late 2023, released Feb 2024 ✅First academic project to build agents as graphs ✅ Meta-prompts optimize nodes ✅ RL optimizes edges
Andrej Karpathy
@karpathy
May 11, 2025
We're missing (at least one) major paradigm for LLM learning. Not sure what to call it, possibly it has a name - system prompt learning? Pretraining is for knowledge. Finetuning (SL/RL) is for habitual behavior. Both of these involve a change in parameters but a lot of human
8.7K
Mingchen Zhuge
@MingchenZhuge
Oct 12, 2024
🐝 github.com/metauto-ai/GPT… vs github.com/openai/swarm
shyamal
@shyamalanadkat
Oct 11, 2024
introducing swarm: an experimental framework for building, orchestrating, and deploying multi-agent systems. 🐝 github.com/openai/swarm
6.5K
Mingchen Zhuge
@MingchenZhuge
Oct 16, 2024
Replying to @MingchenZhuge
This is a simple demo of how it collects evidence like humans.
GIF
3.7K
Mingchen Zhuge
@MingchenZhuge
May 12, 2024
MEET @MetaGPT_ @cognition_labs @pyautogen offline (also 20+ outstanding young researchers working on agent.) Super happy!!!🌟 Now back to KSA @AI_KAUST
1.6K
Mingchen Zhuge
@MingchenZhuge
Apr 10, 2025
Another milestone achieved. Thanks!
Mingchen Zhuge
@MingchenZhuge
Sep 23, 2024
It is an important milestone to me✨
5.9K
Mingchen Zhuge
@MingchenZhuge
May 30, 2025
Love Gödel Machine proposed by @SchmidhuberAI 20 years ago Thanks @SakanaAILabs 🎏 and @jeffclune ‘s team (all JS’s fans🥳)
Sakana AI
@SakanaAILabs
May 30, 2025
Introducing The Darwin Gödel Machine: AI that improves itself by rewriting its own code sakana.ai/dgm The Darwin Gödel Machine (DGM) is a self-improving agent that can modify its own code. Inspired by evolution, we maintain an expanding lineage of agent variants,
1.3K
Mingchen Zhuge
@MingchenZhuge
Mar 27, 2025
Happy to see #OpenManus achieving this milestone! Congrats @manna56749964, @didiforx, @SteamedBun18755, @MetaGPT_ (also Jingyu 🐳 - sry cannot find your X aha...) Keep up the journey ahead!
Manna Liang
@manna56749964
Mar 26, 2025
I'm glad that OpenManus 🌟 has exceeded 40k～ Thank you, @AnthropicAI @browser_use @MetaGPT_ @allhands_ai @SWE_agent @didiforx @xingyaow_ @MingchenZhuge @gregpr07
2.7K
Mingchen Zhuge
@MingchenZhuge
Mar 8, 2025
Some viewpoints about #Manus @manusai 1. Their overseas marketing strategy followed the "Devin" approach—essentially a successful "copy-paste" 2. Domestically, they cleverly leveraged keywords like "Chinese" and "first-ever"—a strategy that has proven to yield significant
3.7K
Mingchen Zhuge
@MingchenZhuge
Apr 17, 2025
</> Playing with vibe coding agents to mirror my career path. Which one do you like? 1. genspark.ai 2. mgx.dev 3. v0.dev 4. lovable.dev 5. bolt.new 6. deepsite (human interacts 1, 2, 1, 1, 2, 1)
00:00
7.2K
Mingchen Zhuge
@MingchenZhuge
Apr 3, 2025
👀 𝗜𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗶𝗻𝗴 𝘁𝗵𝗶𝗻𝗴𝘀 𝗳𝗼𝗿 𝗺𝗲: 1. Can have a try to reproduce GPTSwarm: Language Agents as Optimizable Graphs (ICML 2024 Oral). GPTSwarm opened the door to using graph to build agent and combining RL/OPRO to improve agentic systems. In fact, most advanced
OpenAI
@OpenAI
Apr 2, 2025
We’re releasing PaperBench, a benchmark evaluating the ability of AI agents to replicate state-of-the-art AI research, as part of our Preparedness Framework. Agents must replicate top ICML 2024 papers, including understanding the paper, writing code, and executing experiments.
5.6K