List your company

Multi-hop Question Answering on HotpotQA (Tokens and Time)

17,910Output TokensRAG-Critic
3,797.27,461.111,12514,788.9Apr 1, 2026

Evaluation results

MethodLinks
RAG-CriticBaseline=Search-o12026.0417,9104,735
RAG-CriticBaseline=Search-R12026.0415,2035,511
RAG-CriticBaseline=ReAct2026.0414,1965,036
Step-wiseBaseline=ReAct2026.0411,5241,031
RerunBaseline=ReAct2026.048,714713
RerunBaseline=Search-o12026.047,844633
Step-wiseBaseline=Search-o12026.046,861611
RerunBaseline=Search-R12026.046,457652
Step-wiseBaseline=Search-R12026.045,848530
DR-RAGBaseline=Search-R12026.044,594270
DR-RAGBaseline=ReAct2026.044,514279
DR-RAGBaseline=Search-o12026.044,340274