Factual Recall on TriviaQA
77AccuracyLLM-PeerReview-W
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LLM-PeerReview-WType=LLM Ensemble2025.12 | 77 | 67.8 | |
| LLM-PeerReviewType=LLM Ensemble2025.12 | 76.9 | 67.4 | |
| Llama-3.1-8B-SelectionType=Our variants2025.12 | 76.5 | 66.4 | |
| Mistral-7B-SelectionType=Our variants2025.12 | 75.6 | 64.7 | |
| Qwen2.5-7B-SelectionType=Our variants2025.12 | 75.5 | 65.5 | |
| Llama-3.1-8B-InstructType=Single LLM2025.12 | 75.3 | 53.5 | |
| Qwen2-7B-SelectionType=Our variants2025.12 | 74.2 | 62.1 | |
| Smoothie-LocalType=LLM Ensemble2025.12 | 73.6 | 59.8 | |
| Mistral-7B-InstructType=Single LLM2025.12 | 72.7 | 43.5 | |
| GaCType=LLM Ensemble2025.12 | 71.5 | 60.2 | |
| Agent-ForestType=LLM Ensemble2025.12 | 70.5 | 60.1 | |
| Theoretical averageType=Single LLM2025.12 | 68.4 | 54.1 | |
| RandomType=LLM Ensemble2025.12 | 68.4 | 54.2 | |
| Qwen2-7B-InstructType=Single LLM2025.12 | 63 | 56.6 | |
| Smoothie-GlobalType=LLM Ensemble2025.12 | 63 | 60.5 | |
| Qwen2.5-7B-InstructType=Single LLM2025.12 | 62.5 | 62.7 |