SparVAR: Exploring Sparsity in Visual AutoRegressive Modeling for Training-Free Acceleration

ArXi:2602.04361v2 Announce Type: replace-cross Visual AutoRegressive (VAR) modeling has garnered significant attention for its innovative next-scale prediction paradigm. However, mainstream VAR paradigms attend to all tokens across historical scales at each autoregressive step. As the next scale resolution grows, the computational complexity of attention increases quartically with resolution, causing substantial latency. Prior accelerations often skip high-resolution scales, which speeds up inference but discards high-frequency details and harms image quality.