Question Answering on MusQ (Accuracy)
17.21AccuracyCDKC
Evaluation Results
| Method | Links | |
|---|---|---|
| CDKCBackbone=Qwen2.5-14B-Instruct2026.02 | 17.21 | |
| GRPOBackbone=Qwen2.5-14B-Instruct2026.02 | 16.84 | |
| CDKCBackbone=Qwen2.5-3B-Instruct2026.02 | 10.1 | |
| GRPOBackbone=Qwen2.5-3B-Instruct2026.02 | 10.01 | |
| CoTBackbone=Qwen2.5-14B-Instruct2026.02 | 8.94 | |
| CGKEBackbone=Qwen2.5-14B-Instruct2026.02 | 8.27 | |
| RAGBackbone=Qwen2.5-14B-Instruct2026.02 | 6.7 | |
| CGKEBackbone=Qwen2.5-3B-Instruct2026.02 | 5.34 | |
| Vanilla SFTBackbone=Qwen2.5-14B-Instruct2026.02 | 5.09 | |
| RAGBackbone=Qwen2.5-3B-Instruct2026.02 | 4.43 | |
| Vanilla LLMBackbone=Qwen2.5-14B-Instruct2026.02 | 4.3 | |
| CoTBackbone=Qwen2.5-3B-Instruct2026.02 | 3.93 | |
| Vanilla SFTBackbone=Qwen2.5-3B-Instruct2026.02 | 2.57 | |
| Vanilla LLMBackbone=Qwen2.5-3B-Instruct2026.02 | 1.16 |