{"ID":2921894,"CreatedAt":"2026-06-02T02:42:49.606572591Z","UpdatedAt":"2026-06-03T23:52:15.800058688Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2606.01503","arxiv_id":"2606.01503","title":"On the Limits of Token Reduction for Efficient Unified Vision Language Training","abstract":"Unified vision-language models (VLMs) integrate visual understanding and visual generation within a single autoregressive backbone, but their joint training is computationally expensive and largely overlooked from an efficiency perspective. In this work, we study the feasibility and limits of token-reduction-based acceleration for unified VLM training. Through a systematic analysis of layerwise attention allocation, we uncover a fundamental asymmetry: visual understanding exhibits substantial late-layer visual redundancy, whereas visual generation maintains persistent dependence on image tokens across depth. Guided by this observation, we design task-specific accelerators that selectively reduce image-token computation for each objective. While these methods achieve significant efficiency gains in isolated settings, we observe a consistent synergy loss under unified training -- task-specific token dropping necessitates divergent parameter pathways and eliminates the mutual performance gains typically observed in joint optimization. Our findings suggest that efficient unified modeling requires preserving shared cross-task structures, highlighting the need for synergy-aware acceleration strategies. Project page: https://chicychen.github.io/TokenReductionUnifiedVLM/.","short_abstract":"Unified vision-language models (VLMs) integrate visual understanding and visual generation within a single autoregressive backbone, but their joint training is computationally expensive and largely overlooked from an efficiency perspective. In this work, we study the feasibility and limits of token-reduction-based acce...","url_abs":"https://arxiv.org/abs/2606.01503","url_pdf":"https://arxiv.org/pdf/2606.01503v1","authors":"[\"Siyi Chen\",\"Weiming Zhuang\",\"Jingtao Li\",\"Lingjuan Lv\"]","published":"2026-05-31T23:59:12Z","proceeding":"cs.CV","tasks":"[\"cs.CV\",\"cs.AI\",\"cs.CL\"]","methods":"[\"Language Model\"]","has_code":false}