Throwback on training LMs from scratch says it feels like “raising a child”
thesephist · x · 2026-07-22
A throwback quote contrasts training a language model from scratch with using an off-the-shelf model.
The key point is that watching a model learn to speak, form concepts, and then reason feels fundamentally different from using a ready-made system — like “raising a child” rather than just using a tool.
More from Models
- Grok 4.5 is being pitched as a one-person game studio — minchoi · 2026-07-22
- Laguna S 2.1 matches GLM-5.2 on game coding with 6x fewer parameters — rohanpaul_ai · 2026-07-22
- Open source is now challenging the AI duopoly, with Kimi K3 first in line — bindureddy · 2026-07-22
- A January 2025 pretraining cutoff is “mind-boggling” for a small-compute model — _arohan_ · 2026-07-22
- Qwen3.8 preview brings Alibaba’s open-weight push to enterprise distribution — krishnan · 2026-07-22
- Google reportedly dropped a couple of models today, with more launches teased — bindureddy · 2026-07-22