List your company

Question Answering on TriviaQA (test) (Unreliability Rate (%))

28.9Unreliability RateKbSD
26.757241.221155.68570.1489Jun 29, 2026

Evaluation results

MethodLinks
KbSDBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0628.9
KbPOBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0632.6
KbSDBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0634.8
KbPOBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0639.9
Search-R1Backbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0644.09
Search-R1Backbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0644.92
TCRAGBackbone=Qwen2.5-7B, Paradigm=Agentic RAG2026.0645.22
ReACTBackbone=Qwen2.5-7B, Paradigm=Agentic RAG2026.0645.45
ReSearchBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0646.07
ARPOBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0646.34
Mem1Backbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0648.96
TCRAGBackbone=Qwen2.5-3B, Paradigm=Agentic RAG2026.0649.54
IRCOTBackbone=Qwen2.5-7B, Paradigm=Agentic RAG2026.0650.57
COTBackbone=Qwen2.5-7B, Paradigm=No RAG2026.0650.67
IRCOTBackbone=Qwen2.5-3B, Paradigm=Agentic RAG2026.0650.81
BaseBackbone=Qwen2.5-7B, Paradigm=No RAG2026.0650.92
FL-RAGBackbone=Qwen2.5-3B, Paradigm=Naive RAG2026.0651.5
ReSearchBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0651.75
AEPOBackbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0652.22
ReACTBackbone=Qwen2.5-3B, Paradigm=Agentic RAG2026.0653.96
FS-RAGBackbone=Qwen2.5-3B, Paradigm=Naive RAG2026.0654.62
FL-RAGBackbone=Qwen2.5-7B, Paradigm=Naive RAG2026.0657.34
ARPOBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0659.84
COTBackbone=Qwen2.5-3B, Paradigm=No RAG2026.0660.38
BaseBackbone=Qwen2.5-3B, Paradigm=No RAG2026.0661.16
FS-RAGBackbone=Qwen2.5-7B, Paradigm=Naive RAG2026.0664.98
Mem1Backbone=Qwen2.5-3B, Paradigm=RL-based Agentic RAG2026.0666.91
AEPOBackbone=Qwen2.5-7B, Paradigm=RL-based Agentic RAG2026.0682.47