pith. sign in

Real-time video generation with pyramid attention broadcast

2 Pith papers cite this work. Polarity classification is still indexing.

2 Pith papers citing it

fields

cs.CV 2

years

2026 2

verdicts

UNVERDICTED 2

representative citing papers

Attention Sinks in Diffusion Transformers: A Causal Analysis

cs.CV · 2026-05-10 · unverdicted · novelty 6.0 · 2 refs

Suppressing attention sinks in diffusion transformers does not degrade CLIP-T alignment at moderate levels but induces sink-specific perceptual shifts six times larger than equal-budget random masking.

citing papers explorer

Showing 2 of 2 citing papers.

  • DFSAttn: Dynamic Fine-grained Sparse Attention for Efficient Video Generation cs.CV · 2026-05-22 · unverdicted · none · ref 16

    DFSAttn is a training-free framework for dynamic fine-grained sparse attention in video DiTs that achieves up to 2.1x speedup while preserving generation quality via Hilbert reordering, hierarchical scoring, and adaptive caching.

  • Attention Sinks in Diffusion Transformers: A Causal Analysis cs.CV · 2026-05-10 · unverdicted · none · ref 18 · 2 links

    Suppressing attention sinks in diffusion transformers does not degrade CLIP-T alignment at moderate levels but induces sink-specific perceptual shifts six times larger than equal-budget random masking.