Technology ❯ Computer Science ❯ Artificial Intelligence

Deep Learning

Neural Networks Transformers Model Optimization Model Training Vision Transformers Convolutional Neural Networks Mixture-of-Experts Large Language Models Fine-Tuning Techniques LSTM Models Pre-trained Models Gated Autoencoders Image Classification Model Compression Sparse Autoencoders Feature Learning Loss Functions GNN Architecture Feature Evolution Implicit Neural Representations Training Techniques Scaffold Learning Cascaded Frameworks Neural Network Architectures Parameter-Efficient Fine-Tuning Parameter Efficiency Network Architectures Attention Mechanisms Transfer Learning Model Training Techniques Mixture of Experts Sparse Supervision Progressive Learning Model Architectures Model Evaluation Zero-shot Learning Dynamic Routing ReLU Networks Low-Rank Adaptation Critic Models Sparse Computation Teacher-Student Models Multi-modal Learning Language Models

MetaNet Lets MoE Models Cut Active Experts Per Layer With Tunable Accuracy Trade-Off

A lightweight support-set controller promises per-task, per-layer control of expert activation that could let teams lower compute and cost without retraining large model components.

MoE Research Delivers New Playbooks for Design, Routing, and Real-World Deployment

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

DeepSeek Publishes mHC to Stabilize Wider Residual Streams in LLMs