mlsanityData to 14 Oct 2026

STaR-KV: Spatio-Temporal Adaptive Re-weighting for KV Cache Compression in GUI Vision-Language Models

arXiv 2606.01790 ↗First posted 1 Jun 2026cs.CV cs.AIcited by 1

On the map: Multimodal LLMs / vision-language models → Visual token pruning and efficiency

also in 1 more
Abstract (arXiv)

Vision-language-model-based graphical user interface (GUI) agents have shown broad automation capabilities, yet deployment is bottlenecked by a key-value (KV) cache that grows linearly with interaction steps. For instance, UI-TARS-1.5-7B consumes 76 GB of GPU memory on merely five screenshots, approaching the capacity of mainstream 80 GB accelerators. Existing KV compression methods share two structural assumptions: aggregating visual-token importance into a single shared saliency map, and applying a fixed top-B cutoff to the fused score distribution. Pilot measurements refute both: spatial specialization lives at the attention-subspace level and migrates across layers, while the score distribution drifts in shape along a trajectory. We propose STaR-KV (Spatio-Temporal Adaptive Re-weighting), a training-free KV cache compression framework that calibrates token importance along three axes: (i) subspace-aware scoring driven by online spatial mutual information; (ii) a temporal stability discount that suppresses redundant cache entries from persistently attended subspaces; and (iii) an entropy-derived temperature that adaptively reshapes the score distribution. Across four GUI benchmarks, STaR-KV achieves the strongest average accuracy among state-of-the-art KV compression methods (e.g., GUIKV, SnapKV) at matched budgets, with no compression-stage FLOPs overhead (-0.07%) and cutting peak GPU memory by nearly 40% at a 20% KV-cache budget. Code is available at https://github.com/kawhiiiileo/STaR-KV.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Visual token pruning and efficiency, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 17 of its 23 references that are in the map. First posted on arXiv 1 Jun 2026.

Builds on

17 of 23 references.

Closest ancestors · 5

Same direction, newest first

  1. 2026Stability Implies Redundancy: Delta Attention Selective Halting for Efficient Long-Context Prefilling20 Apr
  2. 2025MEDA: Dynamic KV Cache Allocation for Efficient Multimodal Long-Context Inference24 Feb
  3. 2024Filter, Correlate, Compress: Training-Free Token Reduction for MLLM Acceleration26 Nov
  4. 2024VL-Cache: Sparsity and Modality-Aware KV Cache Compression for Vision-Language Model Inference Acceleration29 Oct
  5. 2024LOOK-M: Look-Once Optimization in KV Cache for Efficient Multimodal Long-Context Inference26 Jun

Other references · 12

  1. 2017Attention Is All You Needcited by 60,361
  2. 2023Efficient Streaming Language Models with Attention Sinkscited by 2,186
  3. 2023GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpointscited by 1,346
  4. 2024SnapKV: LLM Knows What You are Looking for Before Generationcited by 880
  5. 2024SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agentscited by 567
  6. 2023Model Tells You What to Discard: Adaptive KV Cache Compression for LLMscited by 474
  7. 2025ScreenSpot-Pro: GUI Grounding for Professional High-Resolution Computer Usecited by 254
  8. 2024ShowUI: One Vision-Language-Action Model for GUI Visual Agentcited by 232
  9. 2024On the Effects of Data Scale on UI Control Agentscited by 194
  10. 2025GUI-KV: Efficient GUI Agents via KV Cache with Spatio-Temporal Awarenesscited by 11
  11. 2026Spatio-Temporal Token Pruning for Efficient High-Resolution GUI Agentscited by 7
  12. 2026ST-Lite: Training-Free KV Cache Compression with Spatio-Trajectory Guidance for Long-Horizon GUI Agentscited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026TRACE: Trajectory-robust Admission with Evidence Ordering for Efficient GUI Agentscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025SSD: Spatial-Semantic Head Decoupling for Efficient Autoregressive Image Generation
  2. 2025Head-Aware KV Cache Compression for Efficient Visual Autoregressive Modeling
  3. 2026Dual-Signal Adaptive KV-Cache Optimization for Long-Form Video Understanding in Vision-Language Models
  4. 2026KVCapsule: Efficient Sequential KV Cache Compression for Vision-Language Models with Asymmetric Redundancy
  5. 2025AirCache: Activating Inter-modal Relevancy KV Cache Compression for Efficient Large Vision-Language Model Inference
  6. 2026Compress to Focus: Efficient Coordinate Compression for Policy Optimization in Multi-Turn GUI Agents
  7. 2025Mixing Importance with Diversity: Joint Optimization for KV Cache Compression in Large Vision-Language Models
  8. 2026UniCache: Task- and Type-Aware KV Cache Compression for Unified Multimodal Models