Dossier
Mixture of Experts (MoE) models
Coverage of Mixture of Experts (MoE) models in the Nexus archive.
- Rotary GPU: Exploring Local Execution for Large MoE Models Under Limited VRAM
The article introduces Rotary GPU, a method for executing large Mixture of Experts (MoE) models locally on systems with limited VRAM. It focuses on optimizing resource-constrained environments for machine learning model deployment.