Multiple Choice Question Answering on HellaSwag (normalized accuracy)
78.2Normalized AccuracyOriginal
Evaluation Results
| Method | Links | |
|---|---|---|
| OriginalModel=OLMoE-1B-7B-0125, Reduction ratio=0%2026.05 | 78.2 | |
| OriginalModel=Qwen3-30B-A3B, Reduction ratio=0%2026.05 | 77.6 | |
| FrequencyModel=Qwen3-30B-A3B, Reduction ratio=25%2026.05 | 77.6 | |
| M-SMoEModel=Qwen3-30B-A3B, Reduction ratio=25%2026.05 | 77.5 | |
| ConMoEModel=Qwen3-30B-A3B, Reduction ratio=25%2026.05 | 77.2 | |
| OriginalModel=deepseek-moe-16b-base, Reduction ratio=0%2026.05 | 77.2 | |
| REAPModel=Qwen3-30B-A3B, Reduction ratio=25%2026.05 | 76.8 | |
| REAPModel=deepseek-moe-16b-base, Reduction ratio=25%2026.05 | 76.5 | |
| ConMoEModel=deepseek-moe-16b-base, Reduction ratio=25%2026.05 | 75.7 | |
| FrequencyModel=Qwen3-30B-A3B, Reduction ratio=50%2026.05 | 74.7 | |
| FrequencyModel=deepseek-moe-16b-base, Reduction ratio=25%2026.05 | 74.7 | |
| REAPModel=OLMoE-1B-7B-0125, Reduction ratio=25%2026.05 | 74.7 | |
| ConMoEModel=OLMoE-1B-7B-0125, Reduction ratio=25%2026.05 | 74.1 | |
| HC-SMoEModel=deepseek-moe-16b-base, Reduction ratio=25%2026.05 | 74 | |
| REAPModel=Qwen3-30B-A3B, Reduction ratio=50%2026.05 | 73.6 | |
| M-SMoEModel=Qwen3-30B-A3B, Reduction ratio=50%2026.05 | 73.5 | |
| ConMoEModel=Qwen3-30B-A3B, Reduction ratio=50%2026.05 | 71.5 | |
| HC-SMoEModel=OLMoE-1B-7B-0125, Reduction ratio=25%2026.05 | 71.2 | |
| REAPModel=deepseek-moe-16b-base, Reduction ratio=50%2026.05 | 69.9 | |
| ConMoEModel=deepseek-moe-16b-base, Reduction ratio=50%2026.05 | 69 | |
| M-SMoEModel=deepseek-moe-16b-base, Reduction ratio=25%2026.05 | 68.2 | |
| FrequencyModel=deepseek-moe-16b-base, Reduction ratio=50%2026.05 | 67.8 | |
| FrequencyModel=OLMoE-1B-7B-0125, Reduction ratio=25%2026.05 | 66.8 | |
| HC-SMoEModel=Qwen3-30B-A3B, Reduction ratio=25%2026.05 | 65.3 | |
| HC-SMoEModel=deepseek-moe-16b-base, Reduction ratio=50%2026.05 | 62.9 | |
| M-SMoEModel=OLMoE-1B-7B-0125, Reduction ratio=25%2026.05 | 62.7 | |
| REAPModel=OLMoE-1B-7B-0125, Reduction ratio=50%2026.05 | 56.3 | |
| HC-SMoEModel=OLMoE-1B-7B-0125, Reduction ratio=50%2026.05 | 52.8 | |
| ConMoEModel=OLMoE-1B-7B-0125, Reduction ratio=50%2026.05 | 49.8 | |
| FrequencyModel=OLMoE-1B-7B-0125, Reduction ratio=50%2026.05 | 49.4 | |
| HC-SMoEModel=Qwen3-30B-A3B, Reduction ratio=50%2026.05 | 37 | |
| M-SMoEModel=OLMoE-1B-7B-0125, Reduction ratio=50%2026.05 | 32.7 | |
| M-SMoEModel=deepseek-moe-16b-base, Reduction ratio=50%2026.05 | 26.9 |