For modular addition at each reported modulus (79, 97, and 127), EGD reaches high test accuracy after only a few epochs, vanilla SGD remains on a long plateau before eventually grokking, and column normalization also groks faster than vanilla SGD. · CiteArk
For modular addition at each reported modulus (79, 97, and 127), EGD reaches high test accuracy after only a few epochs, vanilla SGD remains on a long plateau before eventually grokking, and column normalization also groks faster than vanilla SGD.
来源：paper:PDF pp. 2 and 10, Figure 1 and Section 5.2
