Machine Translation on Translation
39.23LLM-as-Judge ScoreProbMoE
Evaluation Results
| Method | Links | |
|---|---|---|
| ProbMoEBackbone=Qwen1.5-MoE-A2.7B, Top-k=42026.06 | 39.23 | |
| DenseMixerBackbone=Qwen1.5-MoE-A2.7B, Top-k=42026.06 | 33.75 | |
| Frozen RouterBackbone=Qwen1.5-MoE-A2.7B, Top-k=42026.06 | 32.75 | |
| ProbMoEBackbone=OLMoE-1B-7B, Top-k=82026.06 | 31.63 | |
| DenseMixerBackbone=OLMoE-1B-7B, Top-k=82026.06 | 30.32 | |
| ConventionalBackbone=Qwen1.5-MoE-A2.7B, Top-k=42026.06 | 30 | |
| Frozen RouterBackbone=OLMoE-1B-7B, Top-k=82026.06 | 28.29 | |
| ConventionalBackbone=OLMoE-1B-7B, Top-k=82026.06 | 27.56 | |
| DefaultMoEBackbone=Qwen1.5-MoE-A2.7B, Top-k=42026.06 | 24 | |
| ReMoEBackbone=Qwen1.5-MoE-A2.7B, Top-k=42026.06 | 16.99 | |
| Base ModelBackbone=Qwen1.5-MoE-A2.7B, Top-k=42026.06 | 16.53 | |
| Base ModelBackbone=OLMoE-1B-7B, Top-k=82026.06 | 11.09 | |
| SparseMixerBackbone=Qwen1.5-MoE-A2.7B, Top-k=42026.06 | 3.5 |