Monthly Roundup #17: April 2024 — LessWrong
Published on April 15, 2024 12:10 PM GMTAs always, a lot to get to. This is...
End-to-end hacking with language models — LessWrong
Published on April 5, 2024 3:06 PM GMTCross-posted from https://tchauvin.com/end-to-end-hacking-with-language-modelsProduced as part of the SERI ML...
AI #57: All the AI News That’s Fit to Print — LessWrong
Published on March 28, 2024 11:40 AM GMTWelcome, new readers! This is my weekly AI post,...
Transformative trustbuilding via advancements in decentralized lie detection — LessWrong
Published on March 16, 2024 5:56 AM GMTAlthough the emergence of functional lie detection would be...
What’s in the box?! – Towards interpretability by distinguishing niches of value within neural networks. — LessWrong
Published on February 29, 2024 6:33 PM GMTAbstractMathematical models can describe neural network architectures and training...
Scientific Method — LessWrong
Published on February 18, 2024 9:06 PM GMTHere I will try to describe the scientific method...
Research Log, RLLMv3 (GPT2-XL, Phi-1.5 and Falcon-RW-1B) — LessWrong
Published on February 15, 2024 3:39 AM GMTThis lengthy log detailing near-zero temperature results has evolved...
AI #49: Bioweapon Testing Begins — LessWrong
Published on February 1, 2024 3:30 PM GMTTwo studies came out on the question of whether...
A starter guide for evals — LessWrong
Published on January 8, 2024 6:24 PM GMTThis is a linkpost for https://www.apolloresearch.ai/blog/a-starter-guide-for-evalsThis is a starter...
AI #45: To Be Determined — LessWrong
Published on January 4, 2024 3:00 PM GMTThe first half of the week was filled with...
A hermeneutic net for agency — LessWrong
Published on January 1, 2024 8:06 AM GMT[Metadata: crossposted from https://tsvibt.blogspot.com/2023/09/a-hermeneutic-net-for-agency.html. First completed September 4, 2023.]...
AI Alignment Metastrategy — LessWrong
Published on December 31, 2023 12:06 PM GMTI call "alignment strategy" the high-level approach to solving...
If Clarity Seems Like Death to Them — LessWrong
Published on December 30, 2023 5:40 PM GMT "—but if one hundred thousand [normies] can turn...