Towards a Low-Delta Oracle Policy for Improved Upper Bound in Sparse Attention

While Diffusion Transformers (DiTs) deliver remarkable visual quality in video generation, the high cost of 3D spatio-temporal attention hinders their scalability. To evaluate and optimize sparse attention, existing studies rely on the Top-$K$ Oracle Policy. However, the rigid truncation of this approach discards the continuous tail, introducing structural errors that accumulate during iterative diffusion. To address these limitations, we introduce a novel Low-Delta Oracle Policy based on a proof that sparse attention achieves zero error when grouping identical attention scores. Building on this, our oracle analysis shows that leveraging low-delta scores preserves the structural integrity of the distribution. Based on this observation, we propose a Unified Tail Aggregation (UTA) method, which aggregates logits where the score variance is bounded by a marginal delta, seamlessly integrating with efficient attention kernels while incurring less than 1% FLOP overhead in our evaluation.Extensive evaluations show our approach achieves up to a 97.4% MSE reduction and a 39.6% improvement under the Top-$K$ oracle and an advanced sparse baseline, respectively.