Reverse Chain-of-Thought Generation on MultiChallenge
45ScoreSSR-D
Evaluation Results
| Method | Links | |
|---|---|---|
| SSR-DBackbone=Qwen3-32B-Think2026.02 | 45 | |
| SSRBackbone=Qwen3-32B-Think2026.02 | 44.6 | |
| NEUBackbone=Qwen3-32B-Think2026.02 | 43.8 | |
| SSR-DBackbone=Qwen3-4B-Think-25072026.02 | 42 | |
| SUPBackbone=Qwen3-32B-Think2026.02 | 41.5 | |
| AUG-SUPBackbone=Qwen3-32B-Think2026.02 | 41.3 | |
| SSR-DBackbone=Qwen3-8B-Think2026.02 | 41.2 | |
| SSRBackbone=Qwen3-4B-Think-25072026.02 | 40.2 | |
| SSRBackbone=Qwen3-8B-Think2026.02 | 39.4 | |
| NEUBackbone=Qwen3-4B-Think-25072026.02 | 38.9 | |
| NEUBackbone=Qwen3-8B-Think2026.02 | 38.2 | |
| SSR-DBackbone=NBG4-3B-Base2026.02 | 37.1 | |
| SUPBackbone=Qwen3-4B-Think-25072026.02 | 36.6 | |
| SSRBackbone=NBG4-3B-Base2026.02 | 35.8 | |
| SUPBackbone=Qwen3-8B-Think2026.02 | 35.5 | |
| AUG-SUPBackbone=Qwen3-4B-Think-25072026.02 | 34.5 | |
| NEUBackbone=NBG4-3B-Base2026.02 | 34.5 | |
| SUPBackbone=NBG4-3B-Base2026.02 | 33.4 | |
| AUG-SUPBackbone=Qwen3-8B-Think2026.02 | 33 | |
| AUG-SUPBackbone=NBG4-3B-Base2026.02 | 33 |