I wish @grok in @Tesla would cue up what I was saying during network failures instead of just bailing out because Starlink switched satellites or whatever.
I made https://t.co/VAqhBRqmum for this. You can record on your Mac with a mini-me bubble, press Done, and you have a shareable link on your clipboard. On the free plan the videos are available for 48 hours.
This is the actual aircraft I'll be flying solo around the world, across all seven continents, including Antarctica.
Currently based in Orlando, Florida.
Thinking of throwing a departure party in Orlando before I fly around the planet.
OPEN INVITE. Come see the plane, meet some cool people, and hang out before takeoff. Date soon TBC.
Drop a comment or DM me if you’d be down to come!
Andrej Karpathy:
“A 1B model can be much more effective than most current LLMs.
90% of current models waste a ton of capacity and money on remembering stuff that doesn’t matter.”
here is his 5 core thesis about Small LLM’s:
1. the intelligence was never the knowledge. Strip the memorized internet, keep the reasoning. What’s left is the cognitive core, and it’s small.
2. the core might fit in ~1B params. Maybe less. “Plenty of room at the bottom.” We’ve been paying for a library when we needed a brain.
3. size stopped being the bottleneck. What makes an agent work is the tools it can call and how tight the feedback loop is, not parameter count.
4. intelligence lives in the system, not the weights. A raw LLM is a CPU with no OS. Small core + good harness > giant model alone.
5. you can own it. A model on your hardware is one no one can rate-limit, price-hike, or deprecate on you.
in 12 months the gap between teams shipping on-device and teams paying per-token frontier bills won’t be capability.
TwIL an 8.8B model at 90%+ of frontier isn’t a benchmark flex.
It’s the thesis Karpathy’s been repeating all year, made real.
study more on this below ↓