Evals & Harnesses beyond Code Mode & Pass Rate: in a Math degree, assignments & exams are largely closed book + no fancy calculator this is for good reason tooling can help you get answers, but putting constraints on your tools helps you sharpen innate skills developing reasoning skills on the underlying structure of problem solving is often more valuable to learn than how to use a tool this imo totally applies to LLM RL training & the current code mode paradigm I LOVE code mode and find it amazing how well models can use code to solve any problem, but using code as a tool is a decision on giving a model a very powerful harness to solve problems, Python is Turing Complete ofc this has tradeoffs such as not being able to solve other problems Im seeing this pretty distinctly in visual reasoning tasks that don’t have code mode, models just aren’t great at persistence or visual perception without code Maybe that’s a fine tradeoff, but AGI level should be able to do tasks without code and I bet training in harnesses without code mode will make them generally more intelligent

EVA ADOS SAYS SPACEX HAS CREATED MULTIPLE COMPETITIVE MOATS - Starlink already gives SpaceX $SPCX a major infrastructure advantage, while SpaceX AI could become a leading AI business - Ados argues the real AI moat is infrastructure, not the LLM itself, pointing to SpaceX’s space-based data center ambitions - She says the strongest proof is when competitors become customers, with Amazon $AMZN relying on SpaceX launches and Anthropic using Colossus One

EVAN talks to Teen Vogue about his decision to go solo: “A lot of fans knew I was making music, and I too had been thinking that I wanted to share my music with the fans. Through that music, I honestly thought that I could give them great happiness. So, in order to show them this music, I shared such thoughts with the company, and made this decision.” (