However, the nonlinearity of risk measures makes it challenging to achieve convergence and optimality.

By analyzing the loss landscape of a single Transformer layer using Softmax and Gaussian attention kernels, our work provides concrete answers to these questions.

Convolution is a fundamental operation in the 3D backbone.

However, these methods typically reach superlinear convergence only when the stochastic Hessian noise diminishes, increasing per-iteration costs over time.

FP method and the minorization-maximization theory to verify the convergence.