Search papers, labs, and topics across Lattice.
Xiamen University
6
0
9
YOLO-PEFT transforms the fine-tuning landscape for real-time detectors by replacing trial-and-error with structured, auditable planning, achieving notable performance gains.
A single ranking can adapt to any frame budget, improving accuracy and reducing latency without retraining the model.
TimePLE redefines video temporal grounding by predicting valid intervals directly, leading to significant performance gains over traditional endpoint-based methods.
OmniView-Space redefines spatial reasoning in MLLMs, achieving unprecedented accuracy by leveraging dynamic, egocentric evidence mapping.
PhySciBench reveals that top LLMs struggle with scientific reasoning, achieving only 33.5% accuracy, while DelveAgent demonstrates a promising 7.5% improvement in performance.
Autoregressive video generation gets a 6x speed boost without sacrificing quality, thanks to a motion-aware caching strategy that finally respects the fact that not all pixels are created equal.
