LiteLVLM: Training-Free Token Pruning for Efficient Vision-Language Models
Discover LiteLVLM, a training-free token pruning method that boosts pixel grounding efficiency in large vision-language models with faster speed and less m...
Benchmarking Hierarchical Agent Coordination in Industrial Scheduling
Explore DESBench, a benchmark for evaluating hierarchical agent coordination in event-driven industrial scheduling with multi-agent systems.
PanoWorld: Advanced 360° Spatial Supersensing AI Model
Discover PanoWorld, a breakthrough AI model enabling superior 360° spatial reasoning with spherical geometry and pano-native understanding.
EvObj: Unsupervised 3D Instance Segmentation Breakthrough
Discover EvObj, a novel method for unsupervised 3D instance segmentation that adapts object representations without scene supervision for real-world accura...
AcquisitionSynthesis: Boost AI Data with Acquisition Functions
Discover how AcquisitionSynthesis uses acquisition functions to generate high-quality synthetic data, improving AI model performance and resilience.