Retrieval-Augmented Generation on WoW
88.87LLM ScoreConsJudge
Evaluation Results
| Method | Links | |
|---|---|---|
| ConsJudgeGenerator=Llama3-8B-Instruct, Reward Model Backbone=Llama3-8B-Instruct2025.02 | 88.87 | |
| Vanilla LLMGenerator=Llama3-8B-Instruct, Reward Model Backbone=Llama3-8B-Instruct2025.02 | 88.31 | |
| SFTGenerator=Llama3-8B-Instruct, Reward Model Backbone=Llama3-8B-Instruct2025.02 | 87.97 | |
| SFTGenerator=MiniCPM-2.4B, Reward Model Backbone=Qwen2.5-14B-Instruct2025.02 | 87.51 | |
| Vanilla LLMGenerator=MiniCPM-2.4B, Reward Model Backbone=Qwen2.5-14B-Instruct2025.02 | 87.49 | |
| ConsJudgeGenerator=MiniCPM-2.4B, Reward Model Backbone=Qwen2.5-14B-Instruct2025.02 | 87.21 | |
| ConsJudgeGenerator=MiniCPM-2.4B, Reward Model Backbone=Llama3-8B-Instruct2025.02 | 86.13 | |
| Vanilla LLMGenerator=MiniCPM-2.4B, Reward Model Backbone=Llama3-8B-Instruct2025.02 | 85.98 | |
| SFTGenerator=MiniCPM-2.4B, Reward Model Backbone=Llama3-8B-Instruct2025.02 | 85.98 | |
| Raw MetricGenerator=MiniCPM-2.4B, Reward Model=Raw Metric2025.02 | 85.48 | |
| Raw MetricGenerator=Llama3-8B-Instruct, Reward Model=Raw Metric2025.02 | 82.66 |