Sparse Mixture-of-Experts are Domain Generalizable Learners

Open paper
Benchmark Model Rank Results
domain-generalization-on-domainnetHybrid-SF-MoE#17Average Accuracy: 52.0
domain-generalization-on-domainnetGMoE-S/16#21Average Accuracy: 48.7
domain-generalization-on-office-homeGMoE-S/16#15Average Accuracy: 74.2
domain-generalization-on-pacs-2GMoE-S/16#29Average Accuracy: 88.1
domain-generalization-on-terraincognitaGMoE-S/16#24Average Accuracy: 48.5
domain-generalization-on-vlcsGMoE-S/16#16Average Accuracy: 80.2

Rank counts only results with a code link.