Question Answering on RD exam Scenario
80.5Overall AccuracyCogRAG+
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| CogRAG+Model=Qwen3-8B2026.04 | 80.5 | 78.8 | 96.7 | 71.2 | 74.5 | 82.7 | 68.8 | |
| CogRAGModel=Qwen3-8B2026.04 | 74.1 | 72 | 90 | 64.3 | 66.2 | 79.6 | 60 | |
| DenseModel=Qwen3-8B2026.04 | 72.6 | 70.6 | 86.7 | 66.1 | 62.3 | 78 | 60 | |
| HybridModel=Qwen3-8B2026.04 | 67.3 | 65.7 | 80 | 60.7 | 54.6 | 73.3 | 60 | |
| BM25Model=Qwen3-8B2026.04 | 66.2 | 63.1 | 80 | 60.7 | 53.3 | 73.3 | 48 | |
| BaselineModel=Qwen3-8B2026.04 | 63.3 | 60.5 | 73.3 | 55.4 | 45.5 | 72.3 | 56 | |
| CogRAG+Model=Llama3.1-8B2026.04 | 57.8 | 58.9 | 72.9 | 52.7 | 61.8 | 57.8 | 49.5 | |
| CogRAGModel=Llama3.1-8B2026.04 | 56.7 | 56.5 | 70 | 51.8 | 54.6 | 58.1 | 48 | |
| DenseModel=Llama3.1-8B2026.04 | 54.1 | 52.4 | 63.3 | 53.6 | 48.1 | 57.1 | 40 | |
| HybridModel=Llama3.1-8B2026.04 | 51.7 | 51.1 | 63.3 | 48.2 | 45.5 | 54.5 | 44 | |
| BM25Model=Llama3.1-8B2026.04 | 47 | 47.8 | 56.7 | 46.4 | 45.5 | 46.6 | 44 | |
| BaselineModel=Llama3.1-8B2026.04 | 45.1 | 41.1 | 53.3 | 46.4 | 35.1 | 50.8 | 20 |