DevRel @Solana • ex @AztecNetwork & @Aave • On a mission to make privacy normal • Exploring verifiably safe AI • Humane technology 🫶 • artificialarcade.com
- Last week I trained a probe to see how LLMs process a human's mood. The results were interesting I'm taking @BlueDotImpact's technical AI safety course and we are learning all about probes, SAEs, and other mech interp techniques I've been really interested in learning how LLMs



