Question Answering on HotpotQA (Accuracy, Time, Efficiency)
83Accuracyoptimized_multi_hop_prompting
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| optimized_multi_hop_promptingBackbone=Qwen2.5 3B, Prompt Category=Novel Hybrid2026.02 | 83 | — | 0.061 | |
| multi_hop_compact_promptingBackbone=Qwen2.5 3B, Prompt Category=Novel Hybrid2026.02 | 82.5 | 10.144 | 0.081 | |
| multi_hop_decomposition_promptingBackbone=Qwen2.5 3B, Prompt Category=Novel Hybrid2026.02 | 80.6 | 9.951 | 0.081 | |
| multi_hop_chain_promptingBackbone=Qwen2.5 3B, Prompt Category=Novel Hybrid2026.02 | 80.1 | — | — | |
| hybrid_scientific_promptingBackbone=Qwen2.5 3B, Prompt Category=Novel Hybrid2026.02 | 79.9 | — | — | |
| Standard_context_aware (baseline)Backbone=Qwen2.5 3B, Type=Baseline2026.02 | 76.8 | 1.185 | 0.647 | |
| expert_synthesis_promptingBackbone=Qwen2.5 3B, Prompt Category=Novel Hybrid2026.02 | — | 12.733 | — | |
| slm_hotpot_smecBackbone=Qwen2.5 3B2026.02 | — | 9.851 | 0.076 |