Multi-Hop Question Answering on Synthetic Multi-Hop Benchmark 2-4 hop average
0.98Avg F1Self Consistency (S-C)
Evaluation results
| Method | Links | |
|---|---|---|
| Self Consistency (S-C)Type=Context Average (2-4 hop), Context Length=0.5k, Backbone=Qwen3-8B2025.09 | 0.98 | |
| InfoQAType=Context Average (2-4 hop), Context Length=0.5k, Backbone=Qwen3-8B2025.09 | 0.97 | |
| InfoQAType=Overall Average (2-4 hop), Backbone=Qwen3-8B2025.09 | 0.74 | |
| Chain-of-Thought (CoT)Type=Overall Average (2-4 hop), Backbone=Qwen3-8B2025.09 | 0.65 | |
| ReActType=Context Average (2-4 hop), Context Length=10k, Backbone=Qwen3-8B2025.09 | 0.21 |