List your company

Multi-hop Question Answering on HotpotQA (val)

79Exact MatchSYMPHONY-L
-0.97619.78740.5561.313Jan 30, 2026Feb 6, 2026Feb 13, 2026Feb 20, 2026Feb 27, 2026Mar 6, 2026Mar 14, 2026

Evaluation results

MethodLinks
SYMPHONY-LBackbone=GPT-4, Oracle Feedback=true, Variant=Large2026.0179
MASTERBackbone=GPT-4, Oracle Feedback=true2026.0176
Beam RetrievalBackbone=GPT-4, Oracle Feedback=true2026.0173
LATSBackbone=GPT-4, Oracle Feedback=true2026.0171
RAPBackbone=GPT-4, Oracle Feedback=true2026.0160
SYMPHONY-SBackbone=GPT-4, Oracle Feedback=true, Variant=Small2026.0159
ToTBackbone=GPT-4, Oracle Feedback=true2026.0155
ReflexionBackbone=GPT-4, Oracle Feedback=true2026.0151
Apex-SearcherType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.0340.2
ReActBackbone=GPT-4, Oracle Feedback=true2026.0139
StepSearch-instructType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.0338.6
ReasonRAGType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.0338.4
CoT-SCBackbone=GPT-4, Oracle Feedback=true2026.0138
DecEx-RAGType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.0337.7
Search-R1-InstructType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.0337
Apex-SearcherType=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.0335.6
ZeroSearch-instructType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.0334.6
StepSearch-instructType=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.0334.5
CoTBackbone=GPT-4, Oracle Feedback=true2026.0134
RAGType=Standard RAG, Backbone=Qwen2.5-7B-Instruct2026.0329.9
Search-R1-InstructType=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.0328.4
ZeroSearch-instructType=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.0327.4
RAGType=Standard RAG, Backbone=Qwen2.5-3B-Instruct2026.0325.5
Search-o1Type=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.0322.1
Search-o1Type=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.0318.7
Direct InferenceType=Non-Retrieval, Backbone=Qwen2.5-7B-Instruct2026.0318.3
IRCoTType=Iterative RAG, Backbone=Qwen2.5-3B-Instruct2026.0316.4
Direct InferenceType=Non-Retrieval, Backbone=Qwen2.5-3B-Instruct2026.0314.9
IRCoTType=Iterative RAG, Backbone=Qwen2.5-7B-Instruct2026.0313.3
CoTType=Non-Retrieval, Backbone=Qwen2.5-7B-Instruct2026.039.2
CoTType=Non-Retrieval, Backbone=Qwen2.5-3B-Instruct2026.032.1