Position Rebinding Cache Reuse: Replay-Free Visual Revisiting for Interleaved Multimodal Reasoning
Quick Answer
The proposed Position Rebinding Cache Reuse (PRCR) framework enhances multimodal reasoning by effectively reusing visual key-value caches without token replay.
Quick Take
The proposed Position Rebinding Cache Reuse (PRCR) framework enhances multimodal reasoning by effectively reusing visual key-value caches without token replay. PRCR achieves a 5% accuracy improvement and reduces visual-revisiting computation significantly, demonstrating superior performance across various benchmarks.
Key Points
- PRCR reconstructs visual evidence to match current decoding states.
- Achieves replay-level performance with a 5% accuracy increase.
- Reduces visual-revisiting computation by tens of thousands of times.
- Utilizes raw visual KV cache with original spatial coordinates.
- Addresses issues of stale positional binding in existing methods.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Interleaved multimodal reasoning improves visual grounding by revisiting visual evidence during multi-step generation, yet existing methods typically rely on token replay, repeatedly forwarding selected visual tokens. A natural shortcut is to reuse the historical visual key-value (KV) cache directly. However, we identify a critical failure mode of this strategy: cached visual keys are already bound to their original positional context. Such stale positional binding distorts attention under later decoding contexts and can trigger severe autoregressive decoding collapse. This failure suggests that effective cache reuse requires reconstructing visual evidence under positions compatible with the current decoding state, rather than directly copying position-bound historical cache entries. To this end, we propose Position Rebinding Cache Reuse (PRCR), a cache-level framework for replay-free visual revisiting. PRCR stores raw visual KV cache together with their original spatial coordinates, then reassigns position-compatible coordinates to select entries and rebinds their keys before injecting the reconstructed cache into the active decoder cache. This design reuses historical visual evidence while preserving textual positional continuity and relative visual structure. Experiments across multiple multimodal reasoning benchmarks show that PRCR achieves replay-level or better performance, improving average accuracy by 5 percent and reducing visual-revisiting computation by up to tens of thousands of times.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2606.26631 [cs.CV] |
| (or arXiv:2606.26631v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2606.26631 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yanli Ji [view email]
[v1]
Thu, 25 Jun 2026 05:47:52 UTC (1,405 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →LLM-Guided ANN Index Optimization for Human-Object Interaction Retrieval
A phase-aware LLM agent optimizes human-object interaction retrieval, outperforming Optuna TPE by 33.3% and VDTuner by 34.2% on the HICO-DET benchmark. This method enhances throughput by 15.3x over UniIR and demonstrates strong transferability across vector database management systems.