The Global Empirical NTK: Self-Referential Bias and Dimensionality of Gradient Descent Learning | Read Paper on Bytez