Question Answering on TriviaQA (EM, ROUGE, F1)
65.7EMGPT-4o
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPT-4oTraining status=Training-free2026.04 | 65.7 | 64.3 | 78.2 | |
| GRIPTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 57.9 | 55.9 | 67.4 | |
| GRIP w/o RLTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 57 | 55.2 | 66.8 | |
| R1-SearcherTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 56 | 53.3 | 64.9 | |
| GPT-3.5 TurboTraining status=Training-free2026.04 | 55.6 | 57.8 | 69.7 | |
| DRAGINTraining status=Training-free2026.04 | 55.3 | 53.6 | 64.6 | |
| RetRobustTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 53.6 | 50.9 | 61.9 | |
| ETCTraining status=Training-free2026.04 | 52.9 | 52.1 | 63 | |
| InstructRAGTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 51.3 | 52.2 | 62.5 | |
| INFO-RAGTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 50.8 | 47.8 | 58.1 | |
| GainRAGTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 50.3 | 49.1 | 59.2 | |
| Self-RAGTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 50.2 | 47.3 | 57.5 | |
| SFT-RAGTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 50.1 | 24.8 | 57.2 | |
| RobustRAGTraining status=Training-based, Backbone=LLaMA3-8B, Retriever=BM252026.04 | 48.8 | 47.7 | 57.9 | |
| FLARETraining status=Training-free2026.04 | 48.6 | 48.5 | 56.4 | |
| Single RAGTraining status=Training-free2026.04 | 46.4 | 47 | 56.8 | |
| InstructTraining status=Training-free2026.04 | 46.2 | 45.3 | 55.1 |