Meta-Attention: Bayesian Per-Token Routing for Efficient Transformer Inference

Alan Ferrari

Standard transformer architectures apply a single attention mechanism uniformly across all tokens and sequence positions, irrespective of local context or computational budget. We…