Multi-session Retrieval-Augmented Generation on MultihopRAG (test)
64.4F1 ScoreContextPilot
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ContextPilotModel=Qwen3-32B2025.11 | 64.4 | 36,296.1 | |
| LMCacheModel=Llama3.3-70B-Instruct2025.11 | 62.9 | 11,596.4 | |
| Radix CacheModel=Llama3.3-70B-Instruct2025.11 | 62.9 | 14,777.1 | |
| ContextPilotModel=Llama3.3-70B-Instruct2025.11 | 62.9 | 30,046.7 | |
| LMCacheModel=Qwen3-32B2025.11 | 60.4 | 14,708.6 | |
| Radix CacheModel=Qwen3-32B2025.11 | 60.4 | 17,682.6 | |
| CacheBlendModel=Llama3.3-70B-Instruct2025.11 | 54.9 | 14,134.2 | |
| CacheBlendModel=Qwen3-32B2025.11 | 50.1 | 36,128.6 | |
| ContextPilotModel=Qwen3-4B-Instruct-25072025.11 | 36.6 | 106,799.5 | |
| LMCacheModel=Qwen3-4B-Instruct-25072025.11 | 35.2 | 34,710.9 | |
| Radix CacheModel=Qwen3-4B-Instruct-25072025.11 | 35.2 | 58,990.1 | |
| CacheBlendModel=Qwen3-4B-Instruct-25072025.11 | 34.8 | 85,405.3 |