Single-hop Question Answering on TriviaQA
72EMDep-Search
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Dep-SearchBackbone=Qwen2.5-7B-Instruct2026.01 | 72 | — | — | — | — | — | — | — | — | — | — | |
| Llama3.3-70BReasoning Protocol=Direct Reasoning (w/o Retrieval), Parameter Count=70B2025.01 | 68.8 | — | 76.8 | — | — | — | — | — | — | — | — | |
| HierSearchBackbone=Qwen2.5-7B-Instruct2026.01 | 67 | — | — | — | — | — | — | — | — | — | — | |
| SD-Search-InstructTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.05 | 66.8 | — | — | — | — | — | — | — | — | — | — | |
| MR-Search-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.05 | 66.6 | — | — | — | — | — | — | — | — | — | — | |
| SD-Search-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.05 | 66.5 | — | — | — | — | — | — | — | — | — | — | |
| ZEROSEARCH-baseBackbone=Qwen-2.5-7B, Variant=Base2025.05 | 66.4 | — | — | — | — | — | — | — | — | — | — | |
| O2-SearcherBackbone=Qwen2.5-7B-Instruct2026.01 | 66.2 | — | — | — | — | — | — | — | — | — | — | |
| AutoRefineBackbone=Qwen2.5-7B-Instruct2026.03 | 65.9 | — | — | — | — | — | — | — | — | — | — | |
| AutoRefine-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.05 | 65.9 | — | — | — | — | — | — | — | — | — | — | |
| RAG-Qwen2.5-32BReasoning Protocol=Retrieval-augmented Reasoning, Parameter Count=32B2025.01 | 65.8 | — | 79.2 | — | — | — | — | — | — | — | — | |
| RAG-QwQ-32BReasoning Protocol=Retrieval-augmented Reasoning, Parameter Count=32B2025.01 | 65.6 | — | 77.6 | — | — | — | — | — | — | — | — | |
| EvalActBackbone=Qwen2.5-7B-Instruct2026.03 | 65.6 | — | — | — | — | — | — | — | — | — | — | |
| ZEROSEARCH-instBackbone=Qwen-2.5-7B, Variant=Instruct2025.05 | 65.2 | — | — | — | — | — | — | — | — | — | — | |
| Dep-SearchBackbone=Qwen2.5-3B-Instruct2026.01 | 65 | — | — | — | — | — | — | — | — | — | — | |
| R1-SearcherBackbone=Qwen2.5-7B-Instruct2026.01 | 64.8 | — | — | — | — | — | — | — | — | — | — | |
| GiGPOBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.07 | 64.7 | — | — | — | — | — | — | — | — | — | — | |
| IGRPOBackbone=Qwen2.5-7B-Instruct, Type=Tree-based2026.07 | 64.7 | — | — | — | — | — | — | — | — | — | — | |
| Thinker-InstructTraining Pipeline=Synthesize → SFT, External Teacher=Qwen2.5-72B, Backbone=Qwen2.5-7B2026.05 | 64.2 | — | — | — | — | — | — | — | — | — | — | |
| SKILLGRAPHEvaluation Protocol=Zero-shot2026.05 | 63.8 | — | — | — | — | — | — | — | — | — | — | |
| ZEROSEARCH-baseBackbone=LLaMA-3.2-3B, Variant=Base2025.05 | 63.8 | — | — | — | — | — | — | — | — | — | — | |
| Search-o1Reasoning Protocol=Retrieval-augmented Reasoning with Reason-in-Documents, Parameter Count=32B2025.01 | 63.4 | — | 74.1 | — | — | — | — | — | — | — | — | |
| Search-R1Backbone=Qwen2.5-7B-Instruct2026.01 | 63.4 | — | — | — | — | — | — | — | — | — | — | |
| EvolveREvaluation Protocol=Zero-shot2026.05 | 63.4 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1-instBackbone=Qwen-2.5-7B, Variant=Instruct2025.05 | 63.4 | — | — | — | — | — | — | — | — | — | — | |
| SkillRLEvaluation Protocol=Zero-shot2026.05 | 63.3 | — | — | — | — | — | — | — | — | — | — | |
| RAgent-Qwen2.5-32BReasoning Protocol=Retrieval-augmented Reasoning, Parameter Count=32B2025.01 | 63 | — | 72.6 | — | — | — | — | — | — | — | — | |
| MultiSearchBackbone=3B-Base2026.05 | 63 | — | 70.3 | — | — | — | — | — | — | 67.9 | — | |
| Search-R1-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.05 | 62.7 | — | — | — | — | — | — | — | — | — | — | |
| ZeroSearchBackbone=Qwen2.5-7B-Instruct2026.01 | 62.6 | — | — | — | — | — | — | — | — | — | — | |
| Tree-GRPOBackbone=Qwen2.5-7B-Instruct, Type=Tree-based2026.07 | 62.5 | — | — | — | — | — | — | — | — | — | — | |
| EvalActBackbone=Qwen2.5-3B-Instruct2026.03 | 62.1 | — | — | — | — | — | — | — | — | — | — | |
| RAgent-QwQ-32BReasoning Protocol=Retrieval-augmented Reasoning, Parameter Count=32B2025.01 | 62 | — | 74 | — | — | — | — | — | — | — | — | |
| AutoRefineBackbone=Qwen2.5-3B-Instruct2026.03 | 62 | — | — | — | — | — | — | — | — | — | — | |
| AutoRefine2026.05 | 62 | — | 68.9 | — | — | — | — | — | — | 67.4 | — | |
| ZeroSearchEvaluation Protocol=Zero-shot2026.05 | 61.8 | — | — | — | — | — | — | — | — | — | — | |
| ZeroSearchBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.07 | 61.8 | — | — | — | — | — | — | — | — | — | — | |
| ZEROSEARCH-baseBackbone=Qwen-2.5-3B, Variant=Base2025.05 | 61.6 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1-baseBackbone=Qwen-2.5-7B, Variant=Base2025.05 | 61.4 | — | — | — | — | — | — | — | — | — | — | |
| HierSearchBackbone=Qwen2.5-3B-Instruct2026.01 | 61 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1Backbone=Qwen2.5-7B-Instruct2026.03 | 61 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1Evaluation Protocol=Zero-shot2026.05 | 61 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1-InstructTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.05 | 61 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1Backbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.07 | 61 | — | — | — | — | — | — | — | — | — | — | |
| IGRPOBackbone=Qwen2.5-3B-Instruct, Type=Tree-based2026.07 | 60.9 | — | — | — | — | — | — | — | — | — | — | |
| CoDA-BaseBackbone=Qwen2.5-3B, Retrieval Protocol=w/ Multi-Hop Retrieval2025.12 | 60.8 | — | — | — | — | — | — | — | — | — | — | |
| O2-SearcherBackbone=Qwen2.5-3B-Instruct2026.01 | 60.4 | — | — | — | — | — | — | — | — | — | — | |
| Tree-GRPOBackbone=Qwen2.5-3B-Instruct, Type=Tree-based2026.07 | 60.4 | — | — | — | — | — | — | — | — | — | — | |
| IGPOBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.07 | 60.1 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1-baseBackbone=Qwen-2.5-3B, Variant=Base2025.05 | 60 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1-baseBackbone=LLaMA-3.2-3B, Variant=Base2025.05 | 60 | — | — | — | — | — | — | — | — | — | — | |
| ReSearch-BaseBackbone=Qwen2.5-3B, Retrieval Protocol=w/ Multi-Hop Retrieval2025.12 | 59.7 | — | — | — | — | — | — | — | — | — | — | |
| GiGPOBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.07 | 59.5 | — | — | — | — | — | — | — | — | — | — | |
| ARPOModel=Qwen2.5-7B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 59.2 | — | — | — | 1.91 | — | — | — | — | — | — | |
| Search-R1Model=Qwen2.5-7B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 59.18 | — | — | — | 1.55 | — | — | — | — | — | — | |
| SIGHTModel=Qwen2.5-7B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 59 | — | — | — | 1.1 | — | — | — | — | — | — | |
| AutoRefine-BaseBackbone=Qwen2.5-3B, Retrieval Protocol=w/ Multi-Hop Retrieval2025.12 | 58.7 | — | — | — | — | — | — | — | — | — | — | |
| Naïve RAGBackbone=Qwen2.5-7B-Instruct2026.03 | 58.5 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1-BaseBackbone=Qwen2.5-3B, Retrieval Protocol=w/ Multi-Hop Retrieval2025.12 | 58.3 | — | — | — | — | — | — | — | — | — | — | |
| Search-R12026.05 | 58.3 | — | 65 | — | — | — | — | — | — | 64.2 | — | |
| RAGEvaluation Protocol=Zero-shot2026.05 | 58.2 | — | — | — | — | — | — | — | — | — | — | |
| RAGBackbone=Qwen-2.5-7B2025.05 | 58.2 | — | — | — | — | — | — | — | — | — | — | |
| IGPOBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.07 | 58.1 | — | — | — | — | — | — | — | — | — | — | |
| RAGBackbone=Qwen-2.5-3B2025.05 | 58 | — | — | — | — | — | — | — | — | — | — | |
| ZEROSEARCH-instBackbone=LLaMA-3.2-3B, Variant=Instruct2025.05 | 58 | — | — | — | — | — | — | — | — | — | — | |
| Tree-GRPOModel=Qwen2.5-7B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 57.81 | — | — | — | 1.24 | — | — | — | — | — | — | |
| RAGBackbone=LLaMA-3.2-3B2025.05 | 57.6 | — | — | — | — | — | — | — | — | — | — | |
| ZEROSEARCH-instBackbone=Qwen-2.5-3B, Variant=Instruct2025.05 | 57.4 | — | — | — | — | — | — | — | — | — | — | |
| ZeroSearchBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.07 | 57.4 | — | — | — | — | — | — | — | — | — | — | |
| MemPO2026.02 | 57.1 | — | 67.83 | — | — | — | 0.25 | 14 | — | — | — | |
| SIGHTModel=Qwen2.5-3B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 56.84 | — | — | — | 1.26 | — | — | — | — | — | — | |
| Qwen2.5-72BReasoning Protocol=Direct Reasoning (w/o Retrieval), Parameter Count=72B2025.01 | 56.8 | — | 65.8 | — | — | — | — | — | — | — | — | |
| ARPOModel=Qwen2.5-3B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 56.25 | — | — | — | 2.96 | — | — | — | — | — | — | |
| R1-SearcherBackbone=Qwen2.5-3B-Instruct2026.01 | 56.2 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1Backbone=Qwen2.5-3B-Instruct2026.01 | 55.8 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1-instBackbone=Qwen-2.5-3B, Variant=Instruct2025.05 | 55.8 | — | — | — | — | — | — | — | — | — | — | |
| Search-o1Model=Qwen2.5-7B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 54.7 | — | — | — | 0.74 | — | — | — | — | — | — | |
| GRPOmemory=none2026.02 | 54.5 | — | 64.57 | — | — | — | 0.52 | 23 | — | — | — | |
| Search-R1Backbone=Qwen2.5-3B-Instruct2026.03 | 54.5 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1Backbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.07 | 54.5 | — | — | — | — | — | — | — | — | — | — | |
| ZeroSearchBackbone=Qwen2.5-3B-Instruct2026.01 | 54.4 | — | — | — | — | — | — | — | — | — | — | |
| Naive RAGBackbone=Qwen2.5-3B, Retrieval Protocol=w/ Single-Hop Retrieval2025.12 | 54.4 | — | — | — | — | — | — | — | — | — | — | |
| Naïve RAGBackbone=Qwen2.5-3B-Instruct2026.03 | 54.4 | — | — | — | — | — | — | — | — | — | — | |
| QwQ-32BReasoning Protocol=Direct Reasoning (w/o Retrieval), Parameter Count=32B2025.01 | 53.8 | — | 60.7 | — | — | — | — | — | — | — | — | |
| R1-InstructBackbone=Qwen2.5-3B-Instruct, Type=RL Training2026.07 | 53.7 | — | — | — | — | — | — | — | — | — | — | |
| Search-R1-instBackbone=LLaMA-3.2-3B, Variant=Instruct2025.05 | 53.6 | — | — | — | — | — | — | — | — | — | — | |
| qwen2.5Mode=ReAct2026.02 | 52.6 | — | 62.73 | — | — | — | 1.02 | 22 | — | — | — | |
| GRPOModel=Qwen2.5-7B-Instruct, Retrieval Configuration=w/o Retrieval2026.02 | 52.54 | — | — | — | 0 | — | — | — | — | — | — | |
| DeepResearcher2026.02 | 52.3 | — | 61.89 | — | — | — | 0.95 | 23 | — | — | — | |
| R1-instructBackbone=LLaMA-3.2-3B, Variant=Instruct2025.05 | 52.2 | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-32BReasoning Protocol=Direct Reasoning (w/o Retrieval), Parameter Count=32B2025.01 | 52 | — | 60.3 | — | — | — | — | — | — | — | — | |
| Tree-GRPOModel=Qwen2.5-3B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 50.2 | — | — | — | 1.36 | — | — | — | — | — | — | |
| ReSearch2026.02 | 50.1 | — | 62.04 | — | — | — | 0.24 | 17 | — | — | — | |
| Search-R1Model=Qwen2.5-3B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 50 | — | — | — | 1.68 | — | — | — | — | — | — | |
| RA-ISFBackbone=Qwen2.5-7B-Instruct2026.01 | 49.2 | — | — | — | — | — | — | — | — | — | — | |
| Search-o1Backbone=LLaMA-3.2-3B2025.05 | 48.4 | — | — | — | — | — | — | — | — | — | — | |
| IRCoTBackbone=Qwen2.5-7B-Instruct2026.03 | 47.8 | — | — | — | — | — | — | — | — | — | — | |
| Search-o1Model=Qwen2.5-3B-Instruct, Retrieval Configuration=w/ Retrieval2026.02 | 47.66 | — | — | — | 0.85 | — | — | — | — | — | — | |
| IRCoTBackbone=Qwen2.5-7B-Instruct2026.01 | 47.4 | — | — | — | — | — | — | — | — | — | — | |
| R1-baseBackbone=Qwen-2.5-7B, Variant=Base2025.05 | 47.4 | — | — | — | — | — | — | — | — | — | — |