Technology ❯ Computer Science ❯ Artificial Intelligence

Model Optimization

Adapter Mechanisms Parameter-Efficient Fine-Tuning Sparse Models Sparse Mixture-of-Experts Supervised Learning Fused Forward-Backward Adapters Memory Efficiency Training Techniques Visual Prompt Frameworks Sparse Computation Diversity Distillation

HI-MoE Proposes Two-Stage Expert Routing for Object Detection

Early COCO tests report small-object gains over a DINO baseline in an unreviewed draft.