Focused Forcing: Content-Aware Per-Frame KV Selection for Efficient Autoregressive Video Diffusion

• URL: http://arxiv.org/abs/2605.18346v1
• Date: Mon, 18 May 2026 12:58:13 GMT
• Title: Focused Forcing: Content-Aware Per-Frame KV Selection for Efficient Autoregressive Video Diffusion
• Abstract summary: We present textbfFocused Forcing, a training-free KV selection method that focuses cached history along both generated-frame and head dimensions.<n>For each generated frame, Focused Forcing preserves the most relevant and distinctive historical frames.<n>Across multiple autoregressive generation paradigms, Focused Forcing achieves up to $textbf1.48times$ end-to-end acceleration without training.
• Score: 25.555611454522126
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent advances in autoregressive video diffusion have enabled sequential and streaming video generation. However, long-horizon generation requires increasingly large KV caches, making efficient compression without sacrificing quality challenging. Existing methods mostly select historical frames based on attention scores, but their context decisions remain coarse. When multiple frames are generated in the same chunk, these methods often apply a shared history selection to the whole chunk, score historical frames solely by attention, and assign head-wise budgets either uniformly or by attention-pattern heuristics rather than explicit head-importance estimation. We show that frames within the same generated chunk can depend on distinct historical frames, that the same historical frame can receive different attention scores as its relative temporal distance to the current frames changes, and that masking different heads induces unequal generation degradation. Motivated by these findings, we propose \textbf{Focused Forcing}, a training-free KV selection method that focuses cached history along both generated-frame and head dimensions. For each generated frame, Focused Forcing preserves the most relevant and distinctive historical frames by combining attention scores with diversity scores of historical frames, while assigning larger budgets to heads with higher estimated importance. Across multiple autoregressive generation paradigms, Focused Forcing achieves up to $\textbf{1.48}\times$ end-to-end acceleration without training, while \textbf{improving visual quality and text alignment}. \textit{Our code will be released on GitHub.}

Related papers

• RIDGE: Region-Informed Derivative-Guided Evidence Selection for Long Video Understanding [55.75126327155788]
  RIDGE is a frame selection framework that reads the frame-query similarity curve as a temporal signal.<n>It partitions the timeline into structural regions and applies region-specific selection to preserve event cores, transitions, buildup, aftermath, and contextual frames under a fixed budget.<n>It achieves the best performance in most settings and remains competitive in the others.
  arXiv  Detail & Related papers  (2026-08-30T18:40:51Z)
• TetherCache: Stabilizing Autoregressive Long-Form Video Generation with Gated Recall and Trusted Alignment [51.33418612284208]
  We propose TetherCache, a training-free and plug-and-play cache management strategy for drift-resistant long video generation.<n>Gated Recall with Attention-Diversity Balancing selects long-range memory frames using a gated score.<n>TAME lightly edits newly recalled memory tokens by aligning their statistics to a trusted context distribution.
  arXiv  Detail & Related papers  (2026-06-11T08:16:08Z)
• FadeMem: Distance-Aware Memory Consolidation for Autoregressive Video Diffusion [59.207505503284715]
  FadeMem organizes historical KV blocks into a temporal hierarchy under a fixed cache budget.<n>New history is inserted as fine-grained entries, while older adjacent entries are progressively merged.<n>Experiments show improved subject consistency, background stability, and temporal coherence over existing bounded-cache strategies.
  arXiv  Detail & Related papers  (2026-06-09T10:22:18Z)
• DySink: Dynamic Frame Sinks for Autoregressive Long Video Generation [51.58559206569209]
  We propose DySink, a retrieval-based framework that maintains a compact memory bank and selects visually relevant historical frames as dynamic frame sinks.<n> Experiments on minute-long videos show that DySink consistently improves dynamic degree over strong baselines while also achieving higher temporal quality.
  arXiv  Detail & Related papers  (2026-05-20T11:01:01Z)
• Echo-Forcing: A Scene Memory Framework for Interactive Long Video Generation [48.476317015122625]
  Echo-Forcing is a training-free scene memory framework for interactive long video generation.<n>It supports smooth transitions, hard cuts, and long-range scene recall under a bounded cache budget.
  arXiv  Detail & Related papers  (2026-05-15T14:33:09Z)
• Memorize-and-Generate: Towards Long-Term Consistency in Real-Time Video Generation [33.32047364623734]
  Memorize-and-Generate (MAG) is a framework that decouples memory compression and frame generation into distinct tasks.<n>We train a memory model to compress historical information into a compact KV cache, and a separate generator model to synthesize subsequent frames utilizing this compressed representation.<n>Experiments demonstrate that MAG achieves superior historical consistency while maintaining competitive performance on standard video generation benchmarks.
  arXiv  Detail & Related papers  (2025-12-21T14:02:53Z)
• MemFlow: Flowing Adaptive Memory for Consistent and Efficient Long Video Narratives [54.07515675393396]
  Existing solutions maintain the memory by compressing historical frames with predefined strategies.<n>We propose MemFlow to address this problem.<n>MemFlow achieves outstanding long-context consistency with negligible burden.
  arXiv  Detail & Related papers  (2025-12-16T18:59:59Z)
• InfLVG: Reinforce Inference-Time Consistent Long Video Generation with GRPO [73.33751812982342]
  InfLVG is an inference-time framework that enables coherent long video generation without requiring additional long-form video data.<n>We show that InfLVG can extend video length by up to 9$times$, achieving strong consistency and semantic fidelity across scenes.
  arXiv  Detail & Related papers  (2025-05-23T07:33:25Z)
• Frame Context Packing and Drift Prevention in Next-Frame-Prediction Video Diffusion Models [63.99949971803903]
  We present a neural network structure, FramePack, to train next-frame (or next-frame-section) prediction models for video generation.<n>FramePack compresses input frame contexts with frame-wise importance so that more frames can be encoded within a fixed context length.<n>We show that existing video diffusion models can be finetuned with FramePack, and analyze the differences between different packing schedules.
  arXiv  Detail & Related papers  (2025-04-17T04:02:31Z)
• History-Guided Video Diffusion [61.03681839276652]
  We introduce History Guidance, a family of guidance methods uniquely enabled by DFoT.<n>A more advanced method, history guidance across time and frequency further enhances motion dynamics.<n>Project website: https://boyuan.space/history-guidance.
  arXiv  Detail & Related papers  (2025-02-10T18:44:25Z)
• Ada-VE: Training-Free Consistent Video Editing Using Adaptive Motion Prior [13.595032265551184]
  Video-to-video synthesis poses significant challenges in maintaining character consistency, smooth temporal transitions, and preserving visual quality during fast motion. We propose an adaptive motion-guided cross-frame attention mechanism that selectively reduces redundant computations. This enables a greater number of cross-frame attentions over more frames within the same computational budget.
  arXiv  Detail & Related papers  (2024-06-07T12:12:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.