Accuracy on Winogrande (Common-sense Reasoning)
85.64AccuracyST-D
Evaluation Results
| Method | Links | |
|---|---|---|
| ST-DBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.06 | 85.64 | |
| ST-SBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.06 | 85.16 | |
| BaseBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.06 | 84.93 | |
| Star1Backbone Model=DeepSeek-R1-Distill-Llama-70B2026.06 | 84.61 | |
| SafePathBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.06 | 84.53 | |
| BaseBackbone Model=Qwen3-8B2026.06 | 78.22 | |
| ST-SBackbone Model=Qwen3-8B2026.06 | 77.82 | |
| ST-DBackbone Model=Qwen3-8B2026.06 | 77.74 | |
| VanillaBackbone=Mixtral-8x7B-v0.1, Sparsity=0%2025.11 | 77.2 | |
| BaseBackbone Model=Qwen3-32B2026.06 | 76.48 | |
| ST-DBackbone Model=Qwen3-32B2026.06 | 76.48 | |
| FP16BIT=16, Evaluation Protocol=Few-shot2026.07 | 76.48 | |
| PuzzleMoEBackbone=Mixtral-8x7B-v0.1, Sparsity=50%2025.11 | 75.8 | |
| PuzzleMoEBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.11 | 75.5 | |
| HC-SMoEBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.11 | 75.4 | |
| ST-SBackbone Model=Qwen3-32B2026.06 | 75.14 | |
| NAEEBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.11 | 75.1 | |
| D2Backbone=Mixtral-8x7B-v0.1, Sparsity=20%2025.11 | 75 | |
| Sub-MoEBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.11 | 75 | |
| WandaBackbone=Mixtral-8x7B-v0.1, Sparsity=2:42025.11 | 74.5 | |
| Star1Backbone Model=Qwen3-32B2026.06 | 74.03 | |
| GSRQBIT=2, Evaluation Protocol=Few-shot2026.07 | 73.16 | |
| NAEEBackbone=Mixtral-8x7B-v0.1, Sparsity=50%2025.11 | 73 | |
| D2Backbone=Mixtral-8x7B-v0.1, Sparsity=40%2025.11 | 73 | |
| STUNBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.11 | 72.7 | |
| SafePathBackbone Model=Qwen3-32B2026.06 | 72.45 | |
| HC-SMoEBackbone=Mixtral-8x7B-v0.1, Sparsity=50%2025.11 | 72.1 | |
| Sub-MoEBackbone=Mixtral-8x7B-v0.1, Sparsity=50%2025.11 | 72 | |
| PuzzleMoEBackbone=Deepseek-MoE-16b, Sparsity=25%2025.11 | 70.6 | |
| WandaBackbone=Qwen3-MoE-30B-A3B, Sparsity=2:42025.11 | 70.5 | |
| VanillaBackbone=Qwen3-MoE-30B-A3B, Sparsity=0%2025.11 | 70.4 | |
| PuzzleMoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=25%2025.11 | 70.4 | |
| PuzzleMoEBackbone=Deepseek-MoE-16b, Sparsity=50%2025.11 | 70.3 | |
| VanillaBackbone=Deepseek-MoE-16b, Sparsity=0%2025.11 | 70.1 | |
| PuzzleMoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=50%2025.11 | 70.1 | |
| Sub-MoEBackbone=Deepseek-MoE-16b, Sparsity=25%2025.11 | 70 | |
| PuzzleMoEBackbone=Qwen1.5-MoE-A2.7B, Sparsity=25%2025.11 | 69.6 | |
| PuzzleMoEBackbone=Qwen1.5-MoE-A2.7B, Sparsity=50%2025.11 | 69.4 | |
| VQLLMBIT=2, Evaluation Protocol=Few-shot2026.07 | 69.3 | |
| WandaBackbone=Deepseek-MoE-16b, Sparsity=2:42025.11 | 69.2 | |
| D2Backbone=Qwen3-MoE-30B-A3B, Sparsity=20%2025.11 | 69.1 | |
| D2Backbone=Deepseek-MoE-16b, Sparsity=20%2025.11 | 69 | |
| GSRQBIT=1, Evaluation Protocol=Few-shot2026.07 | 68.98 | |
| VanillaBackbone=Qwen1.5-MoE-A2.7B, Sparsity=0%2025.11 | 68.9 | |
| ST-SBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.06 | 68.35 | |
| WandaBackbone=Qwen1.5-MoE-A2.7B, Sparsity=2:42025.11 | 67.8 | |
| ST-DBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.06 | 67.32 | |
| D2Backbone=Qwen1.5-MoE-A2.7B, Sparsity=20%2025.11 | 66.8 | |
| BaseBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.06 | 66.38 | |
| HC-SMoEBackbone=Qwen1.5-MoE-A2.7B, Sparsity=25%2025.11 | 66.1 | |
| Sub-MoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=25%2025.11 | 66 | |
| HC-SMoEBackbone=Deepseek-MoE-16b, Sparsity=25%2025.11 | 65.7 | |
| Star1Backbone Model=Qwen3-8B2026.06 | 65.51 | |
| SafePathBackbone Model=Qwen3-8B2026.06 | 65.43 | |
| D2Backbone=Deepseek-MoE-16b, Sparsity=40%2025.11 | 65 | |
| Sub-MoEBackbone=Deepseek-MoE-16b, Sparsity=50%2025.11 | 65 | |
| GSRQBIT=0.75, Evaluation Protocol=Few-shot2026.07 | 63.54 | |
| HC-SMoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=25%2025.11 | 63.2 | |
| Sub-MoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=50%2025.11 | 63 | |
| D2Backbone=Qwen1.5-MoE-A2.7B, Sparsity=40%2025.11 | 60.3 | |
| D2Backbone=Qwen3-MoE-30B-A3B, Sparsity=40%2025.11 | 60.2 | |
| HC-SMoEBackbone=Qwen1.5-MoE-A2.7B, Sparsity=50%2025.11 | 56.1 | |
| HC-SMoEBackbone=Deepseek-MoE-16b, Sparsity=50%2025.11 | 55.3 | |
| HC-SMoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=50%2025.11 | 54 | |
| Star1Backbone Model=DeepSeek-R1-Distill-Llama-8B2026.06 | 53.59 | |
| VQLLMBIT=1, Evaluation Protocol=Few-shot2026.07 | 53.12 | |
| SafePathBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.06 | 51.3 |