STaRR: Spatial-Temporal Token-Dynamics-Aware Responsive Remasking for Diffusion Language Models
Diffusion Language Models (DLMs) enable parallel decoding via iterative denoising, where remasking strategies play a critical role in balancing inference speed and output quality. Existing methods predominantly rely on static confidence thresholds, overlooking the spatial-temporal dynamics of token confidence, causing unnecessary remasking. We propose Spatial-Temporal Token-Dynamics-Aware Responsive Remasking (STaRR), a training-free framework that dynamically adapts remasking decisions based on token confidence evolution. STaRR introduces two metrics, temporal variance and spatial deviance, to guide fine-grained, step-wise dynamic thresholding. We further introduce a step-wise dynamic thresholding strategy, further enhanced with responsiveness optimizations for scalability and robustness. Experiments show that STaRR achieves an average speedup of 4.1 and up to 8.9 while maintaining comparable accuracy.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
STARRY: Spatial-Temporal Action-Centric World Modeling for Robotic Manipulation
Robotic manipulation requires reasoning about future spatial-temporal interactions and geometric constraints, yet existing Vision-Language-Action (VLA) policies often leave predictive representation weakly coupled with a…
Heterogeneous Adaptive Policy Optimization: Tailoring Optimization to Every Token's Nature
Using entropy as a measure of heterogeneity to guide optimization has emerged as a crucial research direction in Reinforcement Learning for LLMs. However, existing methods typically treat it as a discrete filter or post-…
Reinforcement LearningMathematical ReasoningToken Dynamics: Towards Efficient and Dynamic Video Token Representation for Video Large Language Models
Token-based video representation has emerged as a promising approach for enabling LLMs to interpret video content. However, existing token reduction, such as token pruning and token merging, often disrupt essential spati…
Computational EfficiencyToken ReductionT2SGrid: Temporal-to-Spatial Gridification for Video Temporal Grounding
Video Temporal Grounding (VTG) aims to localize the video segment that corresponds to a natural language query, which requires a comprehensive understanding of complex temporal dynamics. Existing Vision-LMMs typically pe…
Temporal SequencesFre-Res: Frequency-Residual Video Token Compression for Efficient Video MLLMs
Video MLLMs face a persistent tension between spatial fidelity and temporal coverage: preserving fine-grained visual details requires many spatial tokens, while capturing short-lived events requires dense temporal sampli…