Mind Lab

233 posts

Mind Lab banner
Mind Lab

Mind Lab

@Macaron0fficial

Katılım Haziran 2025
48 Takip Edilen1.2K Takipçiler
Mind Lab
Mind Lab@Macaron0fficial·
Mind Lab's 2M-token long-context RL project is now open source. It took only 8 GPUs compared with the thousands-of-GPU scale for prior 1M-context work. Ultra-long-context research is no longer a privilege of giant labs. It’s now open to everyone. 🔗: huggingface.co/papers/2607.14…
English
0
17
70
6.8K
Mind Lab
Mind Lab@Macaron0fficial·
Post-training is entering the era of self-evolution and continual learning. We’re excited to join the AReaL community and support AReaL 2.0. RL-as-a-Micro-Service gives online agent RL the flexible backend interface needed for faster, more efficient post-training loops. Mind Lab also contributed AReaL-MinT: GitHub - areal-project/AReaL-MinT: Open MinT training runtime on AReaL More efficient PEFT work coming.
Mind Lab tweet media
English
0
4
11
728
Mind Lab
Mind Lab@Macaron0fficial·
Mind Lab tweet media
ZXX
0
0
3
1.2K
Mind Lab
Mind Lab@Macaron0fficial·
MinT now fully supports GLM-5.2 post-training, and we’ve open-sourced our Megatron-LM fork with GLM-5.2 support. GLM-5.2 usage on MinT is currently private. If you’re interested in collaborating, please contact sales@mindlab.ltd. Repo：github.com/MindLab-Resear…
English
2
11
86
28.5K
Hugging Models
Hugging Models@HuggingModels·
Meet Macaron V1 Preview 749B, a groundbreaking MoE model that redefines personal agents and tool use. This text generation giant, with 749B parameters, is built for creating Generative UI and A2UI applications, making it a game changer for interactive AI.
Hugging Models tweet media
English
5
19
30
2.7K
Mind Lab
Mind Lab@Macaron0fficial·
Warmly welcome everyone to try our app: Macaron AI!
The Crane@TheCraneWeekly

.@Macaron0fficial just closed a $50M Series A led by @meituan and is open-sourcing a 750 billion-parameter agent model with participation from Yuanhe Puhua, Shokz, and Variable Capital. Co-founded by Kaijie Chen and Andrew Chen (Forbes 30 Under 30 Asia 2026), Mindverse's core R&D team comes out of DeepSeek, ByteDance, and xAI. Mindverse bet on post-training over prompting — using LoRA as the foundation for continuous, parameterized memory. Their Mixture-of-LoRA architecture lets thousands of independent "skill packs" mount on a shared base, enabling rapid activation and privacy isolation. On the consumer side, their Macaron AI lifestyle agent app has 2M users and 100K daily actives. Enterprise partners include Huawei Cloud and Microsoft Cloud. #AI #GenerativeAI #AgentAI #ChinaTech #OpenSource #VentureCapital #Meituan #DeepTech

English
1
0
3
813
Mind Lab
Mind Lab@Macaron0fficial·
@LUXECryptoCH That’s might be something we need to wish for on OpenRouter😉
English
0
0
0
107
Mind Lab
Mind Lab@Macaron0fficial·
@namevinu SWE Verified was also compared, and it scored 2 points higher than GLM-5.1. The detailed results are in the blog. (Link of blog is in the content.)
English
0
0
0
117
cyberfxxker
cyberfxxker@cyberfxxker1·
@Macaron0fficial Your work on multi-LoRA is amazing! I'm looking to join your team and wanted to ask if you have any intern or fresh graduate roles open. I'm finishing up my PhD shortly.🧐
English
1
0
3
522
Mind Lab
Mind Lab@Macaron0fficial·
@latentlocal 👋hi Webpage has better experience, right now mobile not working as well as webpage.
English
1
0
0
225
Latent Local
Latent Local@latentlocal·
@Macaron0fficial Exciting stuff! Two things: 1.) When will this be on providers to try? (Any official partners?) 2.) Webpage does not load well in mobile for me, haha.
Latent Local tweet media
English
2
0
1
755
Adina Yakup
Adina Yakup@AdinaYakup·
Macaron-V1-Preview-749B 👀 a Mixture-of-LoRA personal agent model from MindLab ✨ 744B base + 5 specialist LoRAs ✨ Generative UI as a core skill ✨ Personal agent focused ✨ 202K context ✨ MIT license
Adina Yakup tweet media
English
24
46
512
58K
Mind Lab
Mind Lab@Macaron0fficial·
Listen to a beat loop created by Macaron V1.
English
0
0
5
1.4K
Mind Lab
Mind Lab@Macaron0fficial·
Listen to Macaron V1 perform Twinkle, Twinkle, Little Star. 
English
1
0
4
1.7K
Mind Lab
Mind Lab@Macaron0fficial·
Grateful for the people, the conversations, and the opportunities. Onward.🚀
Mind Lab tweet mediaMind Lab tweet media
English
0
0
4
350
Mind Lab
Mind Lab@Macaron0fficial·
Mind Lab is heading to BEYOND Expo 2026 in Macao. We’ll be there to share our latest thinking on personal AI, agent infrastructure, long-term memory, and generative UI. For us, personal AI is not just another chatbot category. It is a new interface between models, memory, tools, and real human context. 📍 BEYOND Expo 2026 📅 May 27–30, 2026 📌 The Venetian Macao Cotai Expo 🎪 Booth: S-424 #BEYONDExpo2026 #PersonalAI #AIAgents #GenerativeUI
Mind Lab tweet media
English
1
0
1
292
Mind Lab retweetledi
VentureBeat
VentureBeat@VentureBeat·
Delta-mem gives AI agents working memory at 0.12% parameter cost — outperforming systems 600x its size on memory-heavy benchmarks. venturebeat.com/ai/a-0-12-para…
English
0
2
7
3.1K

Keşfet

@HuggingModels @LUXECryptoCH @namevinu @cyberfxxker1 @hhsbsuk30192 @latentlocal @AdinaYakup @elonmusk