How Well Can Transformers Emulate In-context Newton's Method?

arXiv:2403.03183v1 Announce Type: new
Abstract: Transformer-based models have demonstrated remarkable in-context learning capabilities, prompting extensive research into its underlying mechanisms. Recent studies have suggested that Transformers can implement first-order optimization algorithms for in-context learning and even second order ones for the case of linear regression. In this work, we study whether Transformers can perform higher order optimization methods, beyond the case of linear regression. We establish that linear attention Transformers with ReLU layers can approximate second order optimization algorithms for …

abstract algorithms arxiv capabilities case context cs.ai cs.lg in-context learning linear linear regression math.oc optimization prompting regression research stat.ml studies study transformer transformers type work