Run-time Steering Can Surpass Post-Training: Reasoning Task Performance
lesswrong.com·4h
📊Bayesian Inference
What I'm reading (#2): More on Kimi K2, how to build a bad research center, Pretraining with RL, and sporks of AGI
interconnects.ai·15h
📊Bayesian Inference
AI Safety at the Frontier: Paper Highlights, July '25
lesswrong.com·15h
📊Bayesian Inference
Why Racket? Why Lisp?
practicaltypography.com·1d·
Discuss: Lobsters, Hacker News
🐍python
Accessibility and the Agentic Web
tetralogical.com·1d·
Discuss: Lobsters, Hacker News
📊Bayesian Inference
Zoom Out: Distributions in Semantic Spaces
lesswrong.com·5d
📊Bayesian Inference
Janet Minipbt Translation
github.com·22h·
Discuss: Lobsters, Hacker News
📊Bayesian Inference
Pianolizer Helps Your Musical Projects Distinguish Notes (2022)
hackaday.com·1d·
Discuss: Lobsters, Hacker News
🐍python
Why do we even need SIMD instructions ?
lemire.me·1d·
Discuss: Lobsters, Hacker News
📊Bayesian Inference
Measuring intelligence and reverse-engineering goals
lesswrong.com·2h
🤖LLM
Statistical suggestions for mech interp research and beyond
lesswrong.com·4d
📊Bayesian Inference
The Necessity of Studying Emergent Machine Ethics Now
lesswrong.com·3h
📊Bayesian Inference
Improving Geographical Resilience For Distributed Open Source Teams with FREON
soatok.blog·1d·
Discuss: Lobsters, Hacker News
📊Bayesian Inference
So you think you can heritable?
dynomight.net·4d·
Discuss: Hacker News
📊Bayesian Inference
AI must RTFM: Why tech writers are becoming context curators
passo.uno·2d·
Discuss: Lobsters, Hacker News
🤖LLM
Statistical takes for mech interp research and beyond
lesswrong.com·4d
📊Bayesian Inference
What even is distributed systems
notes.eatonphil.com·1d·
Discuss: Lobsters, Hacker News
📊Bayesian Inference
Concept Poisoning: Probing LLMs without probes
lesswrong.com·5d
📊Bayesian Inference
Creating the Longest Possible Ski Jump in "The Games: Winter Challenge"
mrwint.github.io·3d·
Discuss: Lobsters, Hacker News
📊Bayesian Inference
We can still sleep peacefully — or so we thought. Steganography via internal activations is already possible in small language models — a potential first step t...
lesswrong.com·1d
🤖LLM