List your company

Open-Domain Question Answering on NaturalQuestions (NQ)

52.88SubEMStable-RAG
20.702429.056237.4145.7638Jan 6, 2026

Evaluation results

MethodLinks
Stable-RAGBackbone=LLaMA3-8B-Instruct, Retriever=DPR, exhaustive decoding=true2026.0152.8851.78
Stable-RAGBackbone=Qwen3-8B, Retriever=DPR, exhaustive decoding=true2026.0152.1252.38
Stable-RAGBackbone=LLaMA3-8B-Instruct, Retriever=DPR, exhaustive decoding=false2026.0152.0250.72
Stable-RAGBackbone=Qwen3-8B, Retriever=DPR, exhaustive decoding=false2026.0151.6951.78
ATMBackbone=Qwen3-8B, Retriever=DPR2026.0150.9451.03
Pos2DistillBackbone=Qwen3-8B, Retriever=DPR2026.0150.7150.93
Vanilla RAGBackbone=Qwen3-8B, Retriever=DPR2026.0150.5550.67
RAATBackbone=Qwen3-8B, Retriever=DPR2026.0150.1250.03
Ms-PoEBackbone=Qwen3-8B, Retriever=DPR2026.0150.0450.08
ATMBackbone=LLaMA3-8B-Instruct, Retriever=DPR2026.0149.7850.19
RetRobustBackbone=Qwen3-8B, Retriever=DPR2026.0149.550.81
Pos2DistillBackbone=LLaMA3-8B-Instruct, Retriever=DPR2026.0149.2548.37
RAATBackbone=LLaMA3-8B-Instruct, Retriever=DPR2026.0149.1249.85
Stable-RAGBackbone=LLaMA3-8B-Instruct, Retriever=Contriever, exhaustive decoding=true2026.0148.7546.58
RetRobustBackbone=LLaMA3-8B-Instruct, Retriever=DPR2026.0148.749.29
Stable-RAGBackbone=LLaMA3-8B-Instruct, Retriever=Contriever, exhaustive decoding=false2026.0148.1445.8
Stable-RAGBackbone=Qwen3-8B, Retriever=Contriever, exhaustive decoding=true2026.0146.9447.13
Vanilla SFTBackbone=LLaMA3-8B-Instruct, Retriever=DPR2026.0146.249.44
Stable-RAGBackbone=Qwen3-8B, Retriever=Contriever, exhaustive decoding=false2026.0146.1246.79
Vanilla RAGBackbone=LLaMA3-8B-Instruct, Retriever=DPR2026.0145.8147.8
Vanilla SFTBackbone=Qwen3-8B, Retriever=DPR2026.0145.649.19
Ms-PoEBackbone=LLaMA3-8B-Instruct, Retriever=DPR2026.0145.5847.53
ATMBackbone=Qwen3-8B, Retriever=Contriever2026.0145.4745.86
RAATBackbone=Qwen3-8B, Retriever=Contriever2026.0145.1345.87
Pos2DistillBackbone=Qwen3-8B, Retriever=Contriever2026.0144.8945.52
Vanilla RAGBackbone=Qwen3-8B, Retriever=Contriever2026.0144.6545.34
Pos2DistillBackbone=LLaMA3-8B-Instruct, Retriever=Contriever2026.0144.5843.12
Ms-PoEBackbone=Qwen3-8B, Retriever=Contriever2026.0144.3945.12
ATMBackbone=LLaMA3-8B-Instruct, Retriever=Contriever2026.0143.7544.88
RetRobustBackbone=Qwen3-8B, Retriever=Contriever2026.0143.144.99
RAATBackbone=LLaMA3-8B-Instruct, Retriever=Contriever2026.0142.3343.85
Vanilla SFTBackbone=LLaMA3-8B-Instruct, Retriever=Contriever2026.0142.144.78
RetRobustBackbone=LLaMA3-8B-Instruct, Retriever=Contriever2026.0141.8244.26
Vanilla SFTBackbone=Qwen3-8B, Retriever=Contriever2026.0141.4145.05
Vanilla RAGBackbone=LLaMA3-8B-Instruct, Retriever=Contriever2026.0140.7542.82
Ms-PoEBackbone=LLaMA3-8B-Instruct, Retriever=Contriever2026.0140.3242.49
Direct GenerationBackbone=LLaMA3-8B-Instruct, Retriever=Contriever2026.0125.1829.11
Direct GenerationBackbone=LLaMA3-8B-Instruct, Retriever=DPR2026.0125.1829.11
Naive GenerationBackbone=Qwen3-8B, Retriever=Contriever2026.0121.9424.07
Naive GenerationBackbone=Qwen3-8B, Retriever=DPR2026.0121.9424.07