Skip to content

Multi-stepsize Mixture-of-Experts (Ms-MoE)

Home / Concepts / Multi-stepsize Mixture-of-Experts (Ms-MoE)

Multi-stepsize Mixture-of-Experts (Ms-MoE)

Auto-generated stub. Edit this file to add more details.

A mixture-of-experts architecture that conditions on temporal stride to maintain stability and accuracy across varying time-steps in forecasting models.

Why It Matters

Provides a novel mechanism for handling multi-scale temporal dynamics in neural operators, which is a common bottleneck in long-horizon forecasting.

Evidence

The model conditions on a requested relative stride and uses a time-step router to activate scale-specific routed experts together with a shared expert.

Metadata & Links