List your company

Question Answering on GPQA (EM, F1)

53.5EMPruneRAG
-1.6212.692741.31Jan 16, 2026Jan 24, 2026Feb 2, 2026Feb 11, 2026Feb 19, 2026Feb 28, 2026Mar 9, 2026

Evaluation results

MethodLinks
PruneRAGBackbone=Qwen3-8B, Time Cost (ms)=88952026.0153.553.5
ConTReGenBackbone=Qwen3-8B, Time Cost (ms)=188302026.015151
Standard RAGBackbone=Qwen3-8B, Time Cost (ms)=36732026.0150.550.5
RAG-StarBackbone=Qwen3-8B, Time Cost (ms)=128222026.0147.947.9
SynPlanResearch-R1Model Backbone=Qwen3-8B2026.0347.5—
VanillaBackbone=Qwen3-8B, Time Cost (ms)=34332026.0147.447.4
SimpleDeepSearcherModel Backbone=Qwen3-8B2026.0346.6—
Search-R1Model Backbone=Qwen3-8B2026.0344.3—
Rejection SamplingModel Backbone=Qwen3-8B2026.0342.5—
ProbTreeBackbone=Qwen3-8B, Time Cost (ms)=93042026.0137.337.3
MemoRAGBackbone=Qwen3-8B, Time Cost (ms)=339662026.0135.835.8
SynPlanResearch-R1Model Backbone=Qwen3-4B2026.0333.4—
Search-o1Model Backbone=Qwen3-8B2026.0332.4—
SimpleDeepSearcherModel Backbone=Qwen3-4B2026.0331.2—
Search-o1Backbone=Qwen3-8B, Time Cost (ms)=95272026.0130.330.3
PruneRAGBackbone=Llama-3.1-8B-Instruct, Time Cost (ms)=8552026.0129.729.7
Search-R1Model Backbone=Qwen3-4B2026.0327.1—
Rejection SamplingModel Backbone=Qwen3-4B2026.0324.9—
VanillaBackbone=Llama-3.1-8B-Instruct, Time Cost (ms)=1532026.0124.724.7
Standard RAGBackbone=Llama-3.1-8B-Instruct, Time Cost (ms)=8132026.0124.224.2
Search-o1Model Backbone=Qwen3-4B2026.0323.9—
MemoRAGBackbone=Llama-3.1-8B-Instruct, Time Cost (ms)=104052026.0120.220.3
Direct InferenceModel Backbone=Qwen3-8B2026.0318.4—
Direct InferenceModel Backbone=Qwen3-4B2026.0318.2—
RAGModel Backbone=Qwen3-8B2026.0316.3—
RAGModel Backbone=Qwen3-4B2026.0314.3—
ConTReGenBackbone=Llama-3.1-8B-Instruct, Time Cost (ms)=60222026.0114.114.5
ProbTreeBackbone=Llama-3.1-8B-Instruct, Time Cost (ms)=40352026.0113.613.7
RAG-StarBackbone=Llama-3.1-8B-Instruct, Time Cost (ms)=312532026.0111.111.5
ReactBackbone=Qwen3-8B, Time Cost (ms)=108722026.017.58.2
Search-o1Backbone=Llama-3.1-8B-Instruct, Time Cost (ms)=40242026.0166.3
Self-RAGBackbone=Llama-3.1-8B-Instruct, Time Cost (ms)=36342026.0133.6
Self-RAGBackbone=Qwen3-8B, Time Cost (ms)=36342026.0133.6
ReactBackbone=Llama-3.1-8B-Instruct, Time Cost (ms)=46372026.010.50.6