X · @swyx
· X / Twitter
RT Sander Dieleman @ ICML 2026🇰🇷: Here's a cool piece of LLM lore: the original scaling laws were wrong due to a bug, which probably led to a lo…
RT Sander Dieleman @ ICML 2026🇰🇷Here's a cool piece of LLM lore: the original scaling laws were wrong due to a bug, which probably led to a lot of wasted compute on oversized undertrained models 🫣 (and that was before we even started properly accounting for inference cost!)Diogo Almeida: http://x.com/i/article/20732764