AI RESEARCH
Approximation Error Upper and Lower Bounds for H\"{o}lder Class with Transformers
arXiv CS.LG
•
ArXi:2605.07463v1 Announce Type: new We explore the expressive power of Transformers by establishing precise approximation error upper and lower bounds for H\"{o}lder class. Specifically, a new approximation upper bound is derived for the standard Transformer architecture equipped with Softmax operators, ReLU activation functions, and residual connections.