Prediction Is All MoE Needs: Expert Load Distribution Goes from Fluctuating to Stabilizing

arXiv:2404.16914 · cs.LG, cs.AI, cs.CL · Submitted 2024-04-25 · Read on arXiv

cs.LG, cs.AI, cs.CL

Submitted: 2024-04-25

Updated: 2026-10-10

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Computational overhead — 1× in this paper · explained in 2 episodes

Sources

• Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer
• GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks