List your company

Multi-hop Reasoning on TriviaQA

70.31Exact Match (EM)Prompt-R1
21.107633.881346.65559.4287Nov 2, 2025Nov 24, 2025Dec 17, 2025Jan 9, 2026Jan 31, 2026Feb 23, 2026Mar 18, 2026

Evaluation results

MethodLinks
Prompt-R12025.1170.3176.91—
CoT ReasoningBackbone=GPT-4o-mini2025.1167.1975.81—
GEPACategory=APO (GPT-4o-mini)2025.1165.6374.91—
BaselineBackbone=GPT-4o-mini2025.1163.2871.34—
OPROCategory=APO (GPT-4o-mini)2025.1160.9472.14—
TextGradCategory=APO (GPT-4o-mini)2025.1154.6967.59—
GRPOBackbone=Qwen3-4B2025.1151.5655.98—
CoT ReasoningBackbone=Qwen3-4B2025.1146.8850.26—
BaselineBackbone=Qwen3-4B2025.1145.3148.52—
SFTBackbone=Qwen3-4B2025.1129.6934.21—
Tree-RAG2026.032548.131
RAPTOR2026.0324.245.431.8
MergeRAG-Sym2026.032444.932.5
BM252026.0323.544.931
RECOMP2026.0323.545.531.5
MergeRAG-Asym2026.0323.544.732.5
BGE-reranker2026.032345.631