Chinchilla Scaling: A replication attempt
This work critically examines a key scaling law in AI, highlighting potential flaws that could mislead researchers and practitioners in model development.
The paper attempts to replicate Hoffmann et al.'s third method for estimating compute-optimal scaling laws, finding inconsistencies with their other methods, poor data fit, and implausibly narrow confidence intervals, while the authors' rederivation yields compatible results.
Hoffmann et al. (2022) propose three methods for estimating a compute-optimal scaling law. We attempt to replicate their third estimation procedure, which involves fitting a parametric loss function to a reconstruction of data from their plots. We find that the reported estimates are inconsistent with their first two estimation methods, fail at fitting the extracted data, and report implausibly narrow confidence intervals--intervals this narrow would require over 600,000 experiments, while they likely only ran fewer than 500. In contrast, our rederivation of the scaling law using the third approach yields results that are compatible with the findings from the first two estimation procedures described by Hoffmann et al.