MoE
NJU × Huawei UBEP: Reengineering MoE Expert Parallelism
·2571 words·13 mins
MoE
Expert Parallelism
UBEP
Huawei
Nanjing University
Superpods
NPU
AI Infrastructure
Stanford's Marin 535B AI Model Training Is Live-Streamed Openly
·1407 words·7 mins
AI
Marin
Stanford
Open Science
LLMs
MoE
AI Training
Machine Learning
NVIDIA GB200
FreeToken Lets a Single RTX 5090 Run 284B MoE Models Locally
·2215 words·11 mins
FreeToken
LLM Inference
MoE
Edge AI
RTX 5090
Local AI
AI Agents
GPU Computing
PCIe
NVIDIA AI PCs Can Now Run 30B MoE Models Locally
·1302 words·7 mins
NVIDIA
AI PC
Local AI
Blackwell
MoE
LLM
DGX Spark
RTX
Generative AI
Cursor Composer 2.5 Redefines AI Cost-Performance Economics
·1027 words·5 mins
Cursor
Composer 2.5
AI Models
LLM
Reinforcement-Learning
MoE
Synthetic Data
Developer Tools
IFEC Explained: Memory-Semantic Acceleration Over Ethernet Scale-Up
·663 words·4 mins
AI Infrastructure
Networking
Ethernet
MoE
Data Centers