Showing 19–27 of 48

Oct 17PROMPTSFew-shot examples that don’t backfire6 min ↗Oct 15PROMPTSThe "red-team yourself" prompt: self-critique that works, and when it backfires6 min ↗Oct 13AGENTSDesigning tools that AI agents can actually use6 min ↗Oct 11PAPERSSpeculative decoding: faster answers without changing them6 min ↗Oct 11AGENTSTool-use evals are broken. Here's how to fix them.6 min ↗Oct 09MODELSContext windows are not memory6 min ↗Oct 07TESTSWhat do reasoning tokens actually cost you? How to measure and control them6 min ↗Oct 02PAPERSScaling laws, in plain English7 min ↗Oct 01BENCHMARKSYour 2-point improvement might be noise: confidence intervals for LLM evals6 min ↗