Maarten V. de Hoop

Universality and Generalization of Causal Transformers Across Context Lengths

This work studies whether one masked transformer can approximate causal token-to-token maps uniformly over sequences of arbitrary length sampling a fixed normalized horizon, and characterizes the causal families admitting uniform approximation on these compact input classes by a single transformer with length-independe...

tomography: annihilators and sensitivity kernels