Guan Wang (@makingAGI) / X

Guan Wang

33 posts

Guan Wang

@makingAGI

CEO of Sapient Intelligence. Building efficient & powerful general intelligence through brain-inspired architecture.

Singapore

Joined July 2025

Pinned
Guan Wang
@makingAGI
May 20
The HRM-Text paper is now available 🎉 HRM-Text explores a different approach to language model pretraining: hierarchical recurrent computation, task-completion training, and latent-space reasoning. At just 1B parameters, HRM-Text achieves competitive performance with
88K
Guan Wang
@makingAGI
Jul 21, 2025
🚀Introducing Hierarchical Reasoning Model🧠🤖 Inspired by brain's hierarchical processing, HRM delivers unprecedented reasoning power on complex tasks like ARC-AGI and expert-level Sudoku using just 1k examples, no pretraining or CoT! Unlock next AI breakthrough with
1.3M
Guan Wang
@makingAGI
Jul 21, 2025
Replying to @ai_for_success
Only ~2 GPU hours for pro Sudoku. 50~200 for ARC-AGI 😀
60K
Guan Wang
@makingAGI
Aug 17, 2025
Thanks to @arcprize for reproducing and verifying the results! ARC-AGI-1: public 41% pass@2 - semi private 32% pass@2 ARC-AGI-2: public 4% pass@2 - semi private 2% pass@2 Due to differences in testing environments, a certain amount of variance in results is acceptable.
39K
Guan Wang
@makingAGI
Jul 21, 2025
Will Sudoku become the MNIST for reasoning? Simple rules, clear structure, unique solutions—yet surprisingly challenging for modern LLMs, often requiring explicit trial-and-error to solve.
sapientinc/sudoku-extreme · Datasets at Hugging Face
From huggingface.co
8.4K
Guan Wang
@makingAGI
Jul 23, 2025
🌟Exactly my thoughts on the next-gen of AI reasoning. Leveraging insights from neuroscience, our Hierarchical Reasoning Model offers practical, efficient scaling of depth. Not every problem can be solved faster with more processors; sometimes, the key is adding depth.
Konpat Ta Preechakul
@konpatp
Jul 21, 2025
Some problems can’t be rushed—they can only be done step by step, no matter how many people or processors you throw at them. We’ve scaled AI by making everything bigger and more parallel: Our models are parallel. Our scaling is parallel. Our GPUs are parallel. But what if the
00:00
9.9K
Guan Wang
@makingAGI
Aug 4, 2025
❤️ Thanks for highlighting our HRM paper! Apologies for any confusion—we're working on clarifying this thoroughly. Stay tuned for updates!
commotum
@commotum
Aug 3, 2025
Ok, so this paragraph in isolation looks pretty bad, but based on the code, THEY DIDN'T TRAIN ON THE TEST SET. In fact, THEY DIDN'T PRETRAIN AT ALL. And that's the point of the paper! 1/
7.3K
Guan Wang
@makingAGI
Jul 31, 2025
Replying to @itsAlexGuerrero and @grok
LLMs training = auto-regressive paradigm. HRM training = RNN thinker with a built-in “stop” signal (ACT) so it knows when to stop. Different playbooks.
2.3K
Guan Wang
@makingAGI
Jul 29, 2025
Thanks for featuring us!😃
Ben Dickson
@bendee983
Jul 28, 2025
Replying to @bendee983
My story on HRM with comments from @makingAGI, CEO of Sapient Intelligence x.com/VentureBeat/st…
8.2K
Guan Wang
@makingAGI
Jul 31, 2025
Replying to @ADarmouni
No. Eval data stayed separate. Details here:
github.com
my mistake! · Issue #1 · sapientinc/HRM
2K
Guan Wang
@makingAGI
Jul 31, 2025
Replying to @notrajivpoddar
Nope. LLMs are awesome but architecture-capped. We’re trying to push past those limits.
2.6K
Guan Wang
@makingAGI
Jul 31, 2025
Replying to @JonathanRoseD
No plan yet. But code, checkpoints, and demo data are open - feel free to roll your own 🛠️
4.4K
Guan Wang
@makingAGI
Jul 31, 2025
Replying to @codeslubber
Fair point! We’re inspired by brains, not handmade expert-system hierarchies. We let the hierarchy learn and grow from data, not hand-made rules.
1.5K
Guan Wang
@makingAGI
Jul 31, 2025
Replying to @mikeboysen
HRM is a way smaller, focused model: A) needs less data, B) answers niche questions without a giant LLM, C) runs cheap.
1.4K