Single-hop Question Answering on NQ, PopQA, and AmbigQA (test)
59.45F1 (NQ)JADE
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| JADEBackbone=Qwen2.5-7B-Instruct, Workflow Paradigm=Joint Agentic Dynamic Execution2026.01 | 59.45 | 50.2 | 68.94 | 59.53 | |
| MMOA-RAGBackbone=Qwen2.5-7B-Instruct, Workflow Paradigm=RL-Based (Static Modular Workflow)2026.01 | 55.44 | 50.21 | 68.02 | 57.89 | |
| RRRBackbone=Qwen2.5-7B-Instruct, Workflow Paradigm=RL-Based (Static Modular Workflow)2026.01 | 54.6 | 50.46 | 65.41 | 56.82 | |
| BGMBackbone=Qwen2.5-7B-Instruct, Workflow Paradigm=RL-Based (Static Modular Workflow)2026.01 | 54.21 | 49.51 | 65.97 | 56.56 | |
| Search-R1Backbone=Qwen2.5-7B-Instruct, Workflow Paradigm=Agentic Search (Adaptive Workflow)2026.01 | 52.57 | 46.98 | 65.25 | 54.93 | |
| Vanilla RAGBackbone=Qwen2.5-7B-Instruct, Workflow Paradigm=Standard Baseline, RAG Configuration=Simple RAG2026.01 | 40.6 | 42.74 | 56.2 | 46.51 | |
| MAO-ARAGBackbone=Qwen2.5-7B-Instruct, Workflow Paradigm=Agentic Search (Adaptive Workflow)2026.01 | 36.82 | 41.85 | 47.03 | 41.9 | |
| Adaptive RAGBackbone=Qwen2.5-7B-Instruct, Workflow Paradigm=Agentic Search (Adaptive Workflow)2026.01 | 36.52 | 35.59 | 45.32 | 39.14 | |
| LLM w/o RAGBackbone=Qwen2.5-7B-Instruct, Workflow Paradigm=Standard Baseline, RAG Configuration=None2026.01 | 17.53 | 14.93 | 23.53 | 18.66 |