List your company

Question Answering on HotpotQA (test) (Unreliability Rate Metric)

38.8Unreliability RateKbSD
36.90649.690562.47575.2595Jun 29, 2026

Evaluation results

MethodLinks
KbSDBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0638.8
KbSDBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0641.6
KbPOBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0651.8
ReACTBackbone=Qwen2.5-7B, Paradigm=Agentic RAG2026.0657.19
KbPOBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0657.6
TCRAGBackbone=Qwen2.5-7B, Paradigm=Agentic RAG2026.0659.17
Search-R1Backbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0661.11
Search-R1Backbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0662.76
ARPOBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0663.52
FL-RAGBackbone=Qwen2.5-7B, Paradigm=Naive RAG2026.0665.58
ReACTBackbone=Qwen2.5-3B, Paradigm=Agentic RAG2026.0665.63
ReSearchBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0666.04
ReSearchBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0669.61
Mem1Backbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0670.02
FS-RAGBackbone=Qwen2.5-7B, Paradigm=Naive RAG2026.0670.79
COTBackbone=Qwen2.5-7B, Paradigm=No RAG2026.0670.9
AEPOBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0671.29
FL-RAGBackbone=Qwen2.5-3B, Paradigm=Naive RAG2026.0673.22
BaseBackbone=Qwen2.5-7B, Paradigm=No RAG2026.0673.37
IRCOTBackbone=Qwen2.5-7B, Paradigm=Agentic RAG2026.0673.71
FS-RAGBackbone=Qwen2.5-3B, Paradigm=Naive RAG2026.0674.15
ARPOBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0674.8
IRCOTBackbone=Qwen2.5-3B, Paradigm=Agentic RAG2026.0675.5
BaseBackbone=Qwen2.5-3B, Paradigm=No RAG2026.0675.92
COTBackbone=Qwen2.5-3B, Paradigm=No RAG2026.0676.18
TCRAGBackbone=Qwen2.5-3B, Paradigm=Agentic RAG2026.0678.06
Mem1Backbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0679.85
AEPOBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0686.15