Henry Hu
About
Work
Blog
Links
Reading
Links
Interesting reads I want to keep track of.
The World Inside Neural Networks
goodfire.ai
Steering Along Manifolds to Control Neural Networks
goodfire.ai
Six (and a half) intuitions for KL divergence
lesswrong.com
Dynalist note
dynalist.io
Toy Models of Superposition
arxiv.org
How GPT, Claude, and Gemini are actually trained and served - Reiner Pope
youtube.com
Modular Manifolds
thinkingmachines.ai
Six (and a half) intuitions for SVD
lesswrong.com
Richard Dawkins is NOT Listening to Jordan Peterson - Bret Weinstein
youtube.com
Linear Mechanisms for Spatiotemporal Reasoning in Vision Language Models
arxiv.org
Situational Awareness: From AGI to Superintelligence
situational-awareness.ai
Universal Paperclips
decisionproblem.com