Why Adam Works Better with $β_1 = β_2$: The Missing Gradient Scale Invariance Principle
This resolves an open question in deep learning optimization, offering a principle for designing future optimizers, though it is incremental as it explains an existing observation rather than introducing a new method.
The paper tackles the unexplained empirical improvement in Adam optimizer performance when momentum parameters β₁ = β₂, showing that this choice ensures gradient scale invariance, a property aligning with modern optimizer design principles. Experiments across vision and language tasks confirm smoother updates with this setting.
Adam has been at the core of large-scale training for almost a decade, yet a simple empirical fact remains unaccounted for: both validation scores and the qualitative behaviour of the training runs improve when the momentum parameters satisfy $β_{1}=β_{2}$. Some recent studies have reported this pattern, but there is still no explanation for why this choice helps. We show that this choice is closely tied to a structural property that we refer to as \textit{gradient scale invariance}. We formalize this notion and prove that Adam becomes gradient scale invariant of first order if and only if $β_{1}=β_{2}$. This perspective places the balanced regime of Adam in direct alignment with the design principles underlying several recent optimizers that explicitly enforce scale-robust updates. The theory is supported by experiments across vision and language tasks, and across different architectural families, in which rescaling the gradient has a markedly smoother effect on the update when $β_{1}=β_{2}$. Overall, our results offer a coherent explanation for an open question in the behavior of Adam and provide a simple principle that helps guide the design of future optimizers.