16 papers
Messages, Not Tokens: Grounded Coresets for Faithful VLM Compression
Long Qian, Jiaqi Wei, Bingke Zhu +2
Modern vision language models (VLMs) turn high-resolution images into long sequences of visual tokens. Every token traverses the language decoder and persists in its prompt KV cach…
UniVAD v2: Unified Visual Anomaly Detection via Support-Conditioned Boundary Construction
Zhaopeng Gu, Bingke Zhu, Zhaowen Li +5
Unified visual anomaly detection seeks to train a single detector that can be deployed across categories, domains, and application scenarios. In the few-shot transfer regime, the k…
ReST-KV: Robust KV Cache Eviction with Layer-wise Output Reconstruction and Spatial-Temporal Smoothing
Yongqi An, Chang Lu, Kuan Zhu +5
Large language models (LLMs) face growing challenges in efficient generative inference due to the increasing memory demands of Key-Value (KV) caches, especially for long sequences.…
Semantic Noise Reduction via Teacher-Guided Dual-Path Audio-Visual Representation Learning
Linge Wang, Yingying Chen, Bingke Zhu +2
Recent advances in audio-visual representation learning have shown the value of combining contrastive alignment with masked reconstruction. However, jointly optimizing these object…
FiLo++: Zero-/Few-Shot Anomaly Detection by Fused Fine-Grained Descriptions and Deformable Localization
Zhaopeng Gu, Bingke Zhu, Guibo Zhu +3
Anomaly detection methods typically require extensive normal samples from the target class for training, limiting their applicability in scenarios that require rapid adaptation, su…
FiLo: Zero-Shot Anomaly Detection by Fine-Grained Description and High-Quality Localization
Zhaopeng Gu, Bingke Zhu, Guibo Zhu +4
Zero-shot anomaly detection (ZSAD) methods entail detecting anomalies directly without access to any known normal or abnormal samples within the target item categories. Existing ap…