List your company

Question Answering on MuSiQue (F1, EM)

50F1 ScoreVOI
-0.9612.2725.538.73Sep 7, 2025Oct 20, 2025Dec 2, 2025Jan 15, 2026Feb 27, 2026Apr 11, 2026May 25, 2026

Evaluation results

MethodLinks
VOIBudget=high, Base Model=Qwen3-32B, Tool-call cap=3, Output-token cap=5002026.055040
BATSBudget=high, Base Model=Qwen3-32B, Tool-call cap=3, Output-token cap=5002026.054837
Quest-GNNLLM Generator=Llama-3.3-70B-Instruct2025.1044.9336.65
BATSBudget=upper-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=3002026.054335
HippoRAG2LLM Generator=Llama-3.3-70B-Instruct2025.1042.4435.16
BAVTBudget=high, Base Model=Qwen3-32B, Tool-call cap=3, Output-token cap=5002026.054234
GraphRAGLLM Generator=Llama-3.3-70B-Instruct2025.1041.934.1
GTE-Qwen2-7BLLM Generator=Llama-3.3-70B-Instruct2025.1040.933.6
RAPTORLLM Generator=Llama-3.3-70B-Instruct2025.1040.332.46
ToG2LLM Generator=Llama-3.3-70B-Instruct2025.1040.2832.75
VOIBudget=upper-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=3002026.054033
NV-Embed-v2LLM Generator=Llama-3.3-70B-Instruct2025.1039.2631.66
GFM-RAGLLM Generator=Llama-3.3-70B-Instruct2025.1039.0630.05
G-retrieverLLM Generator=Llama-3.3-70B-Instruct2025.1038.0430.13
KiRAGLLM Generator=Llama-3.3-70B-Instruct2025.1036.3528.66
VOIBudget=lower-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=2002026.053528
Search-o1Budget=high, Base Model=Qwen3-32B, Tool-call cap=3, Output-token cap=5002026.053529
Auto-RAGLLM Generator=Llama-3.3-70B-Instruct2025.1034.8927.36
Self-RAGLLM Generator=Llama-3.3-70B-Instruct2025.1034.7525.83
GNN-RAGLLM Generator=Llama-3.3-70B-Instruct2025.1034.0928.29
BAVTBudget=upper-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=3002026.053426
BAVTBudget=lower-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=2002026.053326
Contriever+IRCoTLLM Generator=Llama-3.3-70B-Instruct2025.1031.6224.29
PRAISEApproach=Ours2026.0430.7319.03
BM25LLM Generator=Llama-3.3-70B-Instruct2025.1028.0420.38
R1-SearcherApproach=Agentic Search2026.0427.7418.16
StepSearchApproach=Process Reward2026.0427.518.16
No RetrievalLLM Generator=Llama-3.3-70B-Instruct2025.1026.2617.29
RAG-GDBackbone=Llama-3.1-8B, Retriever=E5, K=52026.0523.4713.2
RAG-GDBackbone=Llama-3.1-8B, Retriever=BM25, K=52026.0522.6112.54
Search-R1Approach=Agentic Search2026.0422.512.7
Base adapterBackbone=Llama-3.1-8B, Retriever=E52026.0522.0711.83
Base adapterBackbone=Llama-3.1-8B, Retriever=BM252026.0521.9811.46
DecoupleSearchBackbone=Qwen-2.5-14B-Instruct2025.0921.8218.38
Search-o1Backbone=Qwen-2.5-14B-Instruct2025.0920.5416.4
Search-o1Approach=Agentic Search2026.0420.1111.09
RAG-GDBackbone=Qwen-2.5-7B, Retriever=BM25, K=52026.0519.739.26
GreedyAgentBackbone=Qwen-2.5-14B-Instruct2025.0919.3916.06
RAG-GDBackbone=Qwen-2.5-7B, Retriever=E5, K=52026.0519.269.14
Search-o1Budget=upper-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=3002026.051916
RAgentBackbone=Qwen-2.5-14B-Instruct2025.0918.7216.2
TIPSApproach=Process Reward2026.0418.5910.59
DecoupleSearchBackbone=Qwen-2.5-7B-Instruct2025.0917.7317.2
GreedyAgentBackbone=Qwen-2.5-7B-Instruct2025.0917.0112.4
BAVTBudget=low, Base Model=Qwen3-32B, Tool-call cap=1, Output-token cap=1002026.051711
Search-o1Backbone=Qwen-2.5-7B-Instruct2025.0916.6712.4
ReasonRAGApproach=Process Reward2026.0416.378.94
VOIBudget=low, Base Model=Qwen3-32B, Tool-call cap=1, Output-token cap=1002026.051612
AFlowBudget=high, Base Model=Qwen3-32B, Tool-call cap=3, Output-token cap=5002026.051611
Base adapterBackbone=Qwen-2.5-7B, Retriever=E52026.0515.766.95
Base adapterBackbone=Qwen-2.5-7B, Retriever=BM252026.0515.466.41
CoTBackbone=Qwen-2.5-14B-Instruct2025.0915.39.4
RAgentBackbone=Qwen-2.5-7B-Instruct2025.0914.579
CoTBackbone=Qwen-2.5-7B-Instruct2025.0913.697.8
LightRAGLLM Generator=Llama-3.3-70B-Instruct2025.1013.666.18
Vanilla RAGBackbone=Qwen-2.5-7B, Retriever=BM252026.0513.055.87
Gen-RetrieveBackbone=Qwen-2.5-7B-Instruct2025.0912.958.2
StandardBackbone=Qwen-2.5-14B-Instruct2025.0912.919.4
Vanilla RAGBackbone=Qwen-2.5-7B, Retriever=E52026.0512.775.79
DirectBackbone=Qwen-2.5-14B-Instruct2025.0912.636.2
Vanilla RAGBackbone=Llama-3.1-8B, Retriever=BM252026.0512.445.75
Vanilla RAGBackbone=Llama-3.1-8B, Retriever=E52026.0512.256.04
Judge-RetrieveBackbone=Qwen-2.5-7B-Instruct2025.0912.187.6
Gen-RetrieveBackbone=Qwen-2.5-14B-Instruct2025.0911.78.4
StandardBackbone=Qwen-2.5-7B-Instruct2025.0911.577.2
Jud-RetrieveBackbone=Qwen-2.5-14B-Instruct2025.0911.499
DirectBackbone=Qwen-2.5-7B-Instruct2025.0910.965.8
Query OnlyBackbone=Qwen-2.5-7B, Retriever=None2026.0510.573.8
Direct LLMApproach=Standard2026.0410.142.57
Query OnlyBackbone=Llama-3.1-8B, Retriever=None2026.059.513.81
Naive RAGApproach=Standard2026.049.323.52
IterativeBackbone=Qwen-2.5-14B-Instruct2025.098.75.21
BATSBudget=lower-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=2002026.0586
AFlowBudget=upper-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=3002026.0586
IterativeBackbone=Qwen-2.5-7B-Instruct2025.096.393.8
Search-o1Budget=lower-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=2002026.0563
AFlowBudget=lower-mid, Base Model=Qwen3-32B, Tool-call cap=2, Output-token cap=2002026.0543
BATSBudget=low, Base Model=Qwen3-32B, Tool-call cap=1, Output-token cap=1002026.0532
Search-o1Budget=low, Base Model=Qwen3-32B, Tool-call cap=1, Output-token cap=1002026.0520
AFlowBudget=low, Base Model=Qwen3-32B, Tool-call cap=1, Output-token cap=1002026.0511