Question Answering on TriviaQA 500 QA pairs
55.4Exact Match (EM)Naive RAG
Evaluation Results
| Method | Links | |
|---|---|---|
| Naive RAGBackbone=Qwen-2.5-14B-Instruct2025.10 | 55.4 | |
| Ft. AgentBackbone=Qwen-2.5-7B-Instruct2025.10 | 54.9 | |
| Base AgentBackbone=Qwen-2.5-14B-Instruct2025.10 | 54.1 | |
| Base LLMBackbone=Qwen-2.5-14B-Instruct2025.10 | 50.9 | |
| Ft. AgentBackbone=Qwen-2.5-3B-Instruct2025.10 | 49.5 | |
| Naive RAGBackbone=Mistral-NeMo-8B-Instruct2025.10 | 49.5 | |
| Naive RAGBackbone=Qwen-2.5-7B-Instruct2025.10 | 48.8 | |
| Naive RAGBackbone=Qwen-2.5-3B-Instruct2025.10 | 45.9 | |
| Base AgentBackbone=Qwen-2.5-7B-Instruct2025.10 | 45.5 | |
| Base AgentBackbone=Mistral-NeMo-8B-Instruct2025.10 | 43.1 | |
| Base LLMBackbone=Mistral-NeMo-8B-Instruct2025.10 | 41.5 | |
| Base LLMBackbone=Qwen-2.5-7B-Instruct2025.10 | 39.2 | |
| Base AgentBackbone=Qwen-2.5-3B-Instruct2025.10 | 35.7 | |
| Base LLMBackbone=Qwen-2.5-3B-Instruct2025.10 | 31.4 |