{"ID":2858756,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.08629","arxiv_id":"2510.08629","title":"Dynamic Mixture-of-Experts for Visual Autoregressive Model","abstract":"Visual Autoregressive Models (VAR) offer efficient and high-quality image generation but suffer from computational redundancy due to repeated Transformer calls at increasing resolutions. We introduce a dynamic Mixture-of-Experts router integrated into VAR. The new architecture allows to trade compute for quality through scale-aware thresholding. This thresholding strategy balances expert selection based on token complexity and resolution, without requiring additional training. As a result, we achieve 20% fewer FLOPs, 11% faster inference and match the image quality achieved by the dense baseline.","short_abstract":"Visual Autoregressive Models (VAR) offer efficient and high-quality image generation but suffer from computational redundancy due to repeated Transformer calls at increasing resolutions. We introduce a dynamic Mixture-of-Experts router integrated into VAR. The new architecture allows to trade compute for quality throug...","url_abs":"https://arxiv.org/abs/2510.08629","url_pdf":"https://arxiv.org/pdf/2510.08629v2","authors":"[\"Jort Vincenti\",\"Metod Jazbec\",\"Guoxuan Xia\"]","published":"2025-10-08T12:57:21Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Transformer\"]","has_code":false}