Reverse Chain-of-Thought Generation on EQ-Bench 3
0.896ScoreSSR-D
Evaluation Results
| Method | Links | |
|---|---|---|
| SSR-DBackbone=Qwen3-32B-Think2026.02 | 0.896 | |
| SSRBackbone=Qwen3-32B-Think2026.02 | 0.877 | |
| SSR-DBackbone=NBG4-3B-Base2026.02 | 0.865 | |
| AUG-SUPBackbone=Qwen3-32B-Think2026.02 | 0.864 | |
| SSR-DBackbone=Qwen3-8B-Think2026.02 | 0.861 | |
| SUPBackbone=Qwen3-32B-Think2026.02 | 0.856 | |
| NEUBackbone=Qwen3-32B-Think2026.02 | 0.855 | |
| SSRBackbone=NBG4-3B-Base2026.02 | 0.844 | |
| SSR-DBackbone=Qwen3-4B-Think-25072026.02 | 0.831 | |
| AUG-SUPBackbone=NBG4-3B-Base2026.02 | 0.828 | |
| SSRBackbone=Qwen3-8B-Think2026.02 | 0.826 | |
| SSRBackbone=Qwen3-4B-Think-25072026.02 | 0.819 | |
| SUPBackbone=NBG4-3B-Base2026.02 | 0.815 | |
| NEUBackbone=NBG4-3B-Base2026.02 | 0.796 | |
| NEUBackbone=Qwen3-4B-Think-25072026.02 | 0.793 | |
| SUPBackbone=Qwen3-4B-Think-25072026.02 | 0.792 | |
| AUG-SUPBackbone=Qwen3-4B-Think-25072026.02 | 0.79 | |
| AUG-SUPBackbone=Qwen3-8B-Think2026.02 | 0.783 | |
| SUPBackbone=Qwen3-8B-Think2026.02 | 0.775 | |
| NEUBackbone=Qwen3-8B-Think2026.02 | 0.764 |