@pentagoniac Profile picture

Christopher Nguyen ⽗

@pentagoniac

CEO @aitomatic—Builder, Leader, Professor—ex Panasonic/Google/HKUST/Refugee—https://t.co/cYhm4TOhNk. Industrial Brain Engineering—https://t.co/LlcBWzawTD

Similar User
Mustafa Suleyman photo

@mustafasuleyman

Xinyun Chen photo

@xinyun_chen_

Robert Nishihara photo

@robertnishihara

Naveen Rao photo

@NaveenGRao

Hyung Won Chung photo

@hwchung27

Arjun Bansal photo

@coffeephoenix

Jessica Lessin photo

@Jessicalessin

Zoubin Ghahramani photo

@ZoubinGhahrama1

Quentin Lhoest 🤗 photo

@qlhoest

rohan anil photo

@_arohan_

Rahul Ravindran photo

@23_rahulr

Patrick Wendell photo

@pwendell

John Platt photo

@johnplattml

Wilson Sonsini photo

@wilsonsonsini

Joe Lonsdale photo

@JTLonsdale

Pinned

Geoffrey Hinton just won the Nobel Prize, and while backpropagation has defined much of AI’s past, his latest work—the Forward-Forward (FF) algorithm—may shape its future. What if learning isn’t about minimizing errors, but about exploring the boundaries between reality and…


This is important and generalizable to many domains 👇🏼

"To summarize, I have established three facts. First, AI substantially increases the average rate of materials discovery. Second, it disproportionately benefits researchers with high initial productivity. Third, this heterogeneity is driven almost entirely by differences in…



This. You can’t guarantee consistency *at all*. arxiv.org/abs/2410.02823

It's more that if you're using the LLM for application logic, then you really can't guarantee consistency across models.



Christopher Nguyen ⽗ Reposted

Domain-Specific Agentic AI🚀 How AI Agents Will Shape the Industrial World in the Next 10 Years 🔍 Capture & apply expert knowledge 📈 Higher GDP = Greater AI optimism 🎯 Goal-oriented planning & reasoning 🔓 Open models drive innovation 🔥 Lead the AI revolution for growth!…

Tweet Image 1

Well there’s JEPA, which is like GPT but without all the thé.

This post is unavailable.

Must be a dog person, not that there’s anything wrong with that 🤗

i had no idea cats were this smart



Christopher Nguyen ⽗ Reposted

Each state of their conceptualizing future potentials in a handful of interconnected concurrently considered modalities is all a form of CoT, which cats don't use all the time... just like humans don't, and just like an LLM doesn't, but can. Non-language modalities are baked in.


. @OpenAdaptAI looks interesting!

From "How Domain-Specific AI Agents Will Shape the Industrial World in the Next 10 Years": youtube.com/watch?v=aWEaEg… One way to think about @OpenAdaptAI is as way to automatically record domain knowledge encoded in computer processes and embed that knowledge in LLM prompts.

Tweet Image 1


Loading...

Something went wrong.


Something went wrong.