크라잉넛의 첫 시작은 홍대의 작은 라이브클럽 드럭이었습니다. 라이브클럽은 수많은 인디뮤지션이 관객을 처음 만나는 소중한 무대입니다. 낡은 규제를 바꾸고, 소규모 공연과 인디음악 문화가 더 자유롭게 살아날 수 있도록 국민청원에 힘을 보태주세요.
https://t.co/TMhPEr9Fmy
What if AI learned physics the way Newton did – by experiencing it?
We built Sim2Reason: train LLMs inside virtual worlds governed by real physics laws, zero human annotation.
Result: +5–10% improvement on International Physics Olympiad, zero-shot. 🧵
To plan the future, the PFC represents a step-by-step map of actions, and at every step, this plan moves to the past like a conveyor belt.
This proposes a super simple neural-subspace architecture for planning.
A toy environment can clarify it.
Here is my toy model and notes:
This new paper suggests that LLM ‘aha moments’ arise from an emergent planning-vs-execution hierarchy, similar to HRM’s slow-planner/fast-executor idea
So they proposed HICRA which amplifies per-token credit on scarce planning tokens, focusing strategy & often beating GRPO!
omg.. this cant be real
China’s 4DV AI just dropped 4D Gaussian Splatting, you can turn 2D video into 4D with sound..
imagine.. we will be able to change camera angle, zoom in/out while watching movies
5 examples:
Humanity's progress is accelerating insanely fast:
Stone Age→Farming: 100,000 yrs
Farming→Steam: 12,000 yrs
Steam→AI: 200 yrs
2000-2014: 100 years of progress in 14.
Moore's Law predicted 32x. AI chips did 1000x.
Law of Accelerating Returns is getting weird with AI🧵👇🏻
BREAKING: Apple just proved AI "reasoning" models like Claude, DeepSeek-R1, and o3-mini don't actually reason at all.
They just memorize patterns really well.
Here's what Apple discovered:
(hint: we're not as close to AGI as the hype suggests)
If you get overwhelmed with the number of papers that are published in AI here is something that will help you.
The good folks at Latent Space has curated list of papers that you should read in 2025 if you work with AI.
This is very optimal list of papers broken down by topic that you should definitely read before moving on to others.
I've studied intelligence all my long life, yet still I feel I learned important things about intelligence by reading this book.
Thank you, Max Bennett.
Register now for Year 3 of our 5-day course on Machine Learning for Behavior!
Pose estimation, supervised behavior classification, unsupervised clustering (MotionMapper/Moseq), and analysis of behavior dynamics. Bar Harbor in fall is not bad either.
https://t.co/FoZPkRByRi
The next chapter about transformers is up on YouTube, digging into the attention mechanism: https://t.co/TWNXiWM2az
The model works with vectors representing tokens (think words), and this is the mechanism that allows those vectors to take in meaning from context.