=

The passage around this formula

while what governs whether any individual deployment is safe to ship is closer to the worst-case regression $\Delta_{\max} = \max_{t}\left(\mathrm{Acc}_t^{FP} - \mathrm{Acc}_t^{Q}\right)$. A quantization scheme can post a $\bar{\Delta}$ close to zero while $\Delta_{\max}$ is large, provided the loss concentrates on one or a few tasks that are a small share of the suite. Nothing about $\bar{\Delta}$ being small implies $\Delta_{\max}$ is small; the two only converge if degradation is spread evenly, and the studies below find that it is not.

Read this part in the article →