chengfei lv

Dens3R: A Foundation Model for 3D Geometry Prediction

FlowMM: Cross-Modal Information Flow Guided KV Cache Merging for Efficient Multimodal Context Inference

PureKV: Plug-and-Play KV Cache Optimization with Spatial-Temporal Sparse Attention for Vision-Language Large Models

ARC-Decode: Risk-Bounded Acceptance for Sampling-Based Speculative Decoding

FreqExit: Enabling Early-Exit Inference for Visual Autoregressive Models via Frequency-Aware Guidance

SaMer: A Scenario-aware Multi-dimensional Evaluator for Large Language Models

EcoFace: Audio-Visual Emotional Co-Disentanglement Speech-Driven 3D Talking Face Generation

CORE: Reducing UI Exposure in Mobile Agents via Collaboration Between Cloud and Local LLMs

LAST: Latent Structure guided Gaussian Splatting from Monocular Human Videos