The level of optimization you are achieving is incredible. Same as the optimization of the new Embeddings in December. Congratulations to the team!
RESEARCH
-
AGI Development Speed and Existential Risk Analysis
By
–
Or, you need to get to a similar ish spot to figure out each, but bits are easier than atoms so we always figure out AGI a little faster and it always kills everyone.
-
AGI, Fermi Paradox, and Existential Risk from Advanced AI
By
–
What if the answer to the Fermi paradox is that you need AGI to figure out interstellar travel but the AGI keeps killing everyone before it tells them how to do interstellar travel?
-
ChatGPT: What The New York Times Gets Wrong
By
–
ChatGPT: What The New York Times and others are getting terribly wrong about it (It really helps to explain complex technology, not to sensationalize it.) https://
zdnet.com/article/chatgp
t-what-the-new-york-times-and-others-are-getting-terribly-wrong-about-it/
… @OpenAI @Microsoft @nytimes #AI #ArtificialIntelligence #deeplearning #ChatGPT -
Google ROSIE Scales Robot Learning With Semantic Data Augmentation
By
–
Google’s ROSIE Data Augmentation Strategy Scales Robot Learning With Semantically Imagined Experience
-
Tim Roughgarden’s Inspiring Talk Captivates Full Audience
By
–
Such an inspiring and captivating talk by Tim Roughgarden. Full room, standing room only. I didn't see a single person on their phone or computer.
-
MIT’s Algorithms Textbook Remains Most Cited CS Work
By
–
MIT’s “Introduction to Algorithms,” published #otd in 1990, is the world’s most cited CS text, with 9K citations & half a million copies sold. https://
bit.ly/3y1yMPR (
@mitpress
) -
KTN Model Enables Knowledge Transfer in Heterogeneous Graphs
By
–
Introducing KTN, a new model that transfers knowledge from label-abundant node types in a heterogeneous graph to zero-labeled node types using existing relational information. Learn how it outperforms state-of-the-art transfer learning baselines → https://t.co/IqcqD80Rlk pic.twitter.com/Chxk4NgUZc
— Google AI (@GoogleAI) 1 mars 2023Introducing KTN, a new model that transfers knowledge from label-abundant node types in a heterogeneous graph to zero-labeled node types using existing relational information. Learn how it outperforms state-of-the-art transfer learning baselines → https://
goo.gle/3Zd3bGX -
Specialized Hardware Architectures Mitigate Error Challenges
By
–
Specialized hardware architectures enable new levels of performance but also contribute to challenging hardware errors. Our CEO, @andrewdfeldman
, shares with @semiengineering how we mitigate errors with our design process. Read the full article here: -
Dr. Tanya Berger-Wolf on AI Solutions for Wildlife Conservation
By
–
Dr. Tanya Berger-Wolf, Co-Founder/Director of #AI for conservation nonprofit @WildMeORG spoke at #H2OWorldSydney on the vital topic of AI solutions for #wildlifeconservation. Revolutionizing scientific inquiry, conservation, and policy decisions. http://
ow.ly/kbzz50N6oQC