List your company

Question Answering on HotpotQA (Exact Match Accuracy)

42.9Exact Match Accuracy (EM)ParallelSearch
0.46811.48422.533.516Mar 9, 2026

Evaluation results

MethodLinks
ParallelSearchModel=Qwen2.5-7B2026.0342.9
ICRLModel=Qwen2.5-7B2026.0342.6
Search-R1Model=Qwen2.5-7B2026.0337
ICRLModel=Qwen2.5-3B2026.0335.4
ZeroSearchModel=Qwen2.5-7B2026.0334.6
Reject SamplingModel=Qwen2.5-7B2026.0333.1
Search-R1Model=Qwen2.5-3B2026.0332.4
RAGModel=Qwen2.5-7B2026.0329.9
ZeroSearchModel=Qwen2.5-3B2026.0327.4
RAGModel=Qwen2.5-3B2026.0325.5
R1-baseModel=Qwen2.5-7B2026.0324.2
Reject SamplingModel=Qwen2.5-3B2026.0324
R1-instructModel=Qwen2.5-7B2026.0323.7
Search-o1Model=Qwen2.5-3B2026.0322.1
SFTModel=Qwen2.5-7B2026.0321.7
R1-instructModel=Qwen2.5-3B2026.0320.8
Search-o1Model=Qwen2.5-7B2026.0318.7
SFTModel=Qwen2.5-3B2026.0318.6
DirectModel=Qwen2.5-7B2026.0318.3
IRCoTModel=Qwen2.5-3B2026.0316.4
DirectModel=Qwen2.5-3B2026.0314.9
IRCoTModel=Qwen2.5-7B2026.0313.3
CoTModel=Qwen2.5-7B2026.039.2
CoTModel=Qwen2.5-3B2026.032.1