Log In Sign Up

The Slingshot Mechanism: An Empirical Study of Adaptive Optimizers and the Grokking Phenomenon

by   Vimal Thilak, et al.

The grokking phenomenon as reported by Power et al. ( arXiv:2201.02177 ) refers to a regime where a long period of overfitting is followed by a seemingly sudden transition to perfect generalization. In this paper, we attempt to reveal the underpinnings of Grokking via a series of empirical studies. Specifically, we uncover an optimization anomaly plaguing adaptive optimizers at extremely late stages of training, referred to as the Slingshot Mechanism. A prominent artifact of the Slingshot Mechanism can be measured by the cyclic phase transitions between stable and unstable training regimes, and can be easily monitored by the cyclic behavior of the norm of the last layers weights. We empirically observe that without explicit regularization, Grokking as reported in ( arXiv:2201.02177 ) almost exclusively happens at the onset of Slingshots, and is absent without it. While common and easily reproduced in more general settings, the Slingshot Mechanism does not follow from any known optimization theories that we are aware of, and can be easily overlooked without an in depth examination. Our work points to a surprising and useful inductive bias of adaptive gradient optimizers at late stages of training, calling for a revised theoretical analysis of their origin.


Phase transition in the knapsack problem

We examine the phase transition phenomenon for the Knapsack problem from...

Deep learning: a statistical viewpoint

The remarkable practical success of deep learning has revealed some majo...

A jamming transition from under- to over-parametrization affects loss landscape and generalization

We argue that in fully-connected networks a phase transition delimits th...

Understanding Decoupled and Early Weight Decay

Weight decay (WD) is a traditional regularization technique in deep lear...

Parameter Norm Growth During Training of Transformers

The capacity of neural networks like the widely adopted transformer is k...

From inexact optimization to learning via gradient concentration

Optimization was recently shown to control the inductive bias in a learn...

Memorization in a neural network with adjustable transfer function and conditional gating

The main problem about replacing LTP as a memory mechanism has been to f...