Daniel Dippold retweeted
Frontier pretraining is said to be a big-lab-only game. We don’t have 100k chips yet, so there’s only one way: algorithmic efficiency. Our new recipe matches DeepSeek V4 Pro’s pretrain using 50x less compute – that’s roughly half the FLOPs used for GPT3, or ~$0.5M on GB200.
magic.dev/blog/pretraining
🚀 Big day for EWOR! We’re committing €60M to back the world’s most exceptional founders in building game-changing companies. Empirically, EWOR Fellows raise €1M - €11M before even finishing the fellowship. (1/4)
What do Fellows get? 1:1 mentorship (up to 5 hours/week with unicorn founders), bespoke modules, and a 2000+ global network of VCs and experts. Our virtual-first, borderless model helps founders build from Nairobi, hire in Europe, and raise in Silicon Valley. 3/4
Our 2025 cohort is already stacked with incredible founders tackling huge challenges, from first-time founders to people with multiple 8-figure exits. For anyone who's excited about applying, go to ewor.com! 4/4


