Chinchilla Ratio

Appears in 1 paper

An improvement on the compute-optimal frontier (from DeepMind's Chinchilla paper, 2022).

As used in Paper 13 — Scaling Laws for Neural Language Models →

An improvement on the compute-optimal frontier (from DeepMind's Chinchilla paper, 2022). Suggests a slightly different optimal allocation: roughly equal exponent contributions from N and D, implying more data relative to parameters than the original paper suggested.