HN user

yoquan

904 karma
Posts63
Comments28
View on HN
phys.org 3mo ago

Human brain operates near, but not at, the critical point

yoquan
5pts0
3dview.tokyo-digitaltwin.metro.tokyo.lg.jp 1y ago

Open data 3D point cloud of Tokyo

yoquan
2pts0
www.all-hands.dev 1y ago

Don't Sleep on Single-Agent Systems

yoquan
1pts0
sleepinyourhat.github.io 1y ago

What Succeeding at AI Safety Will Involve

yoquan
2pts0
arxiv.org 2y ago

Achieving Energetic Superiority Through System-Level Quantum Circuit Simulation

yoquan
2pts0
www3.nhk.or.jp 2y ago

There is a new tunnel under Tokyo

yoquan
3pts0
www.modular.com 2y ago

Mojo Presented at LLVM 2023

yoquan
11pts1
github.com 2y ago

Llama2 in Mojo is 15-20% faster than llama2.c

yoquan
4pts0
www.sydney.edu.au 2y ago

Scientists use quantum device to slow chemical process by factor of 100B

yoquan
155pts82
supabase.com 2y ago

Pgvector 0.4.0 Performance (Vs Qdrant)

yoquan
3pts0
twitter.com 3y ago

OpenLLaMA 7B model trained on RedPajama dataset

yoquan
4pts0
bigthink.com 3y ago

Information is the frontier for the study of life

yoquan
6pts0
old.reddit.com 3y ago

Dr Chad Rigetti to Resign

yoquan
3pts3
asia.nikkei.com 3y ago

China unlikely to become a dominant superpower

yoquan
27pts1
www.wsj.com 3y ago

Bank of Japan Can Stomach Weak Yen

yoquan
2pts0
www.inverse.com 3y ago

Brain study suggests the workday needs to be radically restructured

yoquan
36pts10
neurosciencenews.com 3y ago

Mind Wanders Because Your Brain Whispers

yoquan
6pts0
github.com 3y ago

Easy-to-use image segmentation library with pre-trained model zoo

yoquan
1pts0
old.reddit.com 4y ago

Bert-Large: Prune Once for DistilBERT Inference Performance

yoquan
2pts0
old.reddit.com 4y ago

Reddit AMA from a researcher leaving Google AI after 3 years

yoquan
11pts1
www.japantimes.co.jp 4y ago

What's behind the rapid disappearance of the delta variant in Japan

yoquan
29pts17
www.lesswrong.com 4y ago

Sam Altman Q&A: about GPT-4 and AGI

yoquan
14pts9
www.msn.com 4y ago

Pathways: Google is developing a multipurpose AI

yoquan
2pts0
www.cnet.com 5y ago

Japans daring 10-year mission to visit ancient asteroid Ryugu

yoquan
16pts1
huyenchip.com 5y ago

Machine learning is going real-time

yoquan
6pts0
www.reddit.com 5y ago

Why is the mRNA technology of Pfizer/BioNtech Covid-19 vaccine a big deal

yoquan
1pts0
twitter.com 5y ago

Roger Penrose won the Nobel Prize for showing black holes are almost-unavoidable

yoquan
2pts0
www.schneier.com 6y ago

Vulnerability Finding Using Machine Learning

yoquan
2pts0
www.youtube.com 6y ago

Quantum Machine Learning MOOC

yoquan
1pts0
github.com 6y ago

Tiny Differentiable Simulator is a header-only C++ physics library

yoquan
13pts0

The slides describing the "whys" of Mojo and its early vindication did a good job of clearing suspicion around its marketing hype, imo.

OpenAI 4 years ago

Obviously it will relate to complexity in deep learning but I can't resist thinking about some AI models involving quantum computation stuffs :-)

Actually no. Each layer requires output from previous one, which means sequentially computation. While wider layers can utilize GPU parallel computation better. This is kind of trade-off between less memory (less parameters) vs longer time.

To me learning cause-and-effect is a non trivial process, when recognizing relationship often comes first and intrigues "formal" reasoning process afterwards (if any).

The more formal in the later process, the closer we reach to the real "knowledge". So your suggestion is quite practical, in the sense that we can start from that and figure out how to push the engine toward the more formal spectrum.