Question Answering on SeaQA
86.03AccuracyCDKC
Evaluation Results
| Method | Links | |
|---|---|---|
| CDKCBackbone=Qwen2.5-14B-Instruct2026.02 | 86.03 | |
| GRPOBackbone=Qwen2.5-14B-Instruct2026.02 | 85 | |
| CDKCBackbone=Qwen2.5-3B-Instruct2026.02 | 71.33 | |
| GRPOBackbone=Qwen2.5-3B-Instruct2026.02 | 69.43 | |
| CoTBackbone=Qwen2.5-14B-Instruct2026.02 | 67.63 | |
| CGKEBackbone=Qwen2.5-14B-Instruct2026.02 | 66.43 | |
| Vanilla SFTBackbone=Qwen2.5-14B-Instruct2026.02 | 65.97 | |
| Vanilla LLMBackbone=Qwen2.5-14B-Instruct2026.02 | 59.67 | |
| RAGBackbone=Qwen2.5-14B-Instruct2026.02 | 57.4 | |
| RAGBackbone=Qwen2.5-3B-Instruct2026.02 | 45.37 | |
| Vanilla SFTBackbone=Qwen2.5-3B-Instruct2026.02 | 44.53 | |
| CGKEBackbone=Qwen2.5-3B-Instruct2026.02 | 43.17 | |
| CoTBackbone=Qwen2.5-3B-Instruct2026.02 | 40.47 | |
| Vanilla LLMBackbone=Qwen2.5-3B-Instruct2026.02 | 35.43 |