List your company

Question Answering on Bamboogle (EM)

64.1EMGiGPO-Instruct
-0.06816.59133.2549.909Apr 10, 2026Apr 16, 2026Apr 23, 2026Apr 30, 2026May 7, 2026May 14, 2026May 21, 2026

Evaluation results

MethodLinks
GiGPO-InstructRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Process-supervision RL2026.0564.1
PyRAG-RLBackbone=LLaMA-3.1-8B-Instruct2026.0548.3
Search-E1-InstructRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Process-supervision RL2026.0546.4
PyRAG-RLBackbone=Qwen2.5-7B-Instruct2026.0546.1
M2RModel Family=Mistral-7B-Instruct2026.0446
ReSearchModel Family=Mistral-7B-Instruct2026.0445.5
PyRAG-RLBackbone=Qwen3-4B-Instruct2026.0543.2
ReSearchModel Family=Llama-3.1-8B-Instruct2026.0443.1
M2RModel Family=Llama-3.1-8B-Instruct2026.0442.9
SUREModel Family=Mistral-7B-Instruct2026.0442.8
ReSearchBackbone=Qwen2.5-7B-Instruct2026.0542.4
PiCABackbone=Qwen2.5-7B Instruct2026.0541.9
COFTModel Family=Mistral-7B-Instruct2026.0440.3
StepSearchBackbone=Qwen2.5-7B Instruct2026.0540
StepSearchBackbone=Qwen2.5-7B-Instruct2026.0540
SUREModel Family=Llama-3.1-8B-Instruct2026.0437.8
Search-R1Backbone=Qwen2.5-7B Instruct2026.0536.8
TIPSBackbone=Qwen2.5-7B Instruct2026.0536.8
Search-R1Backbone=Qwen2.5-7B-Instruct2026.0536.8
PiCABackbone=Qwen2.5-3B Instruct2026.0534.7
StepSearchBackbone=Qwen2.5-3B Instruct2026.0534.4
AutoRefine-BaseRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Outcome-reward RL2026.0534.4
StepSearch-InstructRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Process-supervision RL2026.0534.4
AutoRefine-InstructRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Outcome-reward RL2026.0533.6
StepSearch-BaseRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Process-supervision RL2026.0532.8
COFTModel Family=Llama-3.1-8B-Instruct2026.0432.5
Search-o1Retrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Prompting2026.0532
RAG-RLBackbone=LLaMA-3.1-8B-Instruct2026.0531.8
Search-o1Backbone=Qwen2.5-3B Instruct2026.0531.6
Search-o1Backbone=Qwen2.5-7B Instruct2026.0530.2
RAG-RLBackbone=Qwen2.5-7B-Instruct2026.0529.6
RAG-SFTBackbone=LLaMA-3.1-8B-Instruct2026.0529.3
ZEROSEARCHBackbone=Qwen2.5-7B-Instruct2026.0527.7
RAG-RLBackbone=Qwen3-4B-Instruct2026.0527.6
RAG-SFTBackbone=Qwen2.5-7B-Instruct2026.0527.1
ZeroSearchBackbone=Qwen2.5-7B Instruct2026.0526.7
ReSearch-InstructRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Outcome-reward RL2026.0526.6
Search-R1Backbone=Qwen2.5-3B Instruct2026.0526.4
RAG-SFTBackbone=Qwen3-4B-Instruct2026.0525.4
Naive RAGModel Family=Mistral-7B-Instruct2026.0425.4
IRCoTRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Prompting2026.0524
Search-R1-InstructRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Outcome-reward RL2026.0523.2
R1-BaseRetrieval Strategy=None2026.0522.4
RAGBackbone=Qwen2.5-7B Instruct2026.0521.4
Naive RAGModel Family=Llama-3.1-8B-Instruct2026.0420.9
TIPSBackbone=Qwen2.5-3B Instruct2026.0520.8
ZeroSearchBackbone=Qwen2.5-3B Instruct2026.0519.3
R1-InstructRetrieval Strategy=None2026.0519.2
Vanilla RAGBackbone=LLaMA-3.1-8B-Instruct2026.0517.6
Vanilla RAGBackbone=Qwen2.5-7B-Instruct2026.0516
Vanilla RAGBackbone=Qwen3-4B-Instruct2026.0514.8
Search-R1-BaseRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Outcome-reward RL2026.0512.8
ReSearch-BaseRetrieval Strategy=Multi-Hop Retrieval, Supervision Paradigm=Outcome-reward RL2026.0512.8
Naive GenerationModel Family=Mistral-7B-Instruct2026.0412.5
Naive GenerationModel Family=Llama-3.1-8B-Instruct2026.0411.6
MT-PPOBackbone=Qwen2.5-7B Instruct2026.0511.2
SFTRetrieval Strategy=None2026.0511.2
MT-PPOBackbone=Qwen2.5-3B Instruct2026.058
Naive RAGRetrieval Strategy=Single-Hop Retrieval2026.058
RAGBackbone=Qwen2.5-3B Instruct2026.057.6
Direct GenerationRetrieval Strategy=None2026.052.4