Chinchilla
Chinchilla — derived from over 400 models that parameters and tokens should be scaled in equal proportion
Why it matters
Exposed that the large models of the time were undertrained; the 70B model outperformed the 280B Gopher
Source — Primary sources first; secondary reports are labelled “Press”
Paperarxiv.orghttps://arxiv.org/abs/2203.15556Added in Week 35, 2026 · See that week