Hybrid Retrieval-Augmented Generation on Hybrid RAG
0.24TTFT (s)ContextPilot
Evaluation Results
| Method | Links | |
|---|---|---|
| ContextPilotModel=Qwen3-4B-Instruct-2507, # Sessions=22025.11 | 0.24 | |
| ContextPilotModel=Qwen3-4B-Instruct-2507, # Sessions=42025.11 | 0.29 | |
| ContextPilotModel=Qwen3-4B-Instruct-2507, # Sessions=82025.11 | 0.34 | |
| CacheBlendModel=Qwen3-4B-Instruct-2507, # Sessions=22025.11 | 0.4 | |
| ContextPilotModel=Qwen3-4B-Instruct-2507, # Sessions=162025.11 | 0.41 | |
| CacheBlendModel=Qwen3-4B-Instruct-2507, # Sessions=42025.11 | 0.42 | |
| CacheBlendModel=Qwen3-4B-Instruct-2507, # Sessions=82025.11 | 0.45 | |
| RadixCacheModel=Qwen3-4B-Instruct-2507, # Sessions=22025.11 | 0.46 | |
| RadixCacheModel=Qwen3-4B-Instruct-2507, # Sessions=42025.11 | 0.49 | |
| RadixCacheModel=Qwen3-4B-Instruct-2507, # Sessions=82025.11 | 0.53 | |
| CacheBlendModel=Qwen3-4B-Instruct-2507, # Sessions=162025.11 | 0.54 | |
| ContextPilotModel=Qwen3-4B-Instruct-2507, # Sessions=322025.11 | 0.65 | |
| RadixCacheModel=Qwen3-4B-Instruct-2507, # Sessions=162025.11 | 0.67 | |
| CacheBlendModel=Qwen3-4B-Instruct-2507, # Sessions=322025.11 | 0.78 | |
| LMCacheModel=Qwen3-4B-Instruct-2507, # Sessions=22025.11 | 0.81 | |
| LMCacheModel=Qwen3-4B-Instruct-2507, # Sessions=42025.11 | 0.87 | |
| LMCacheModel=Qwen3-4B-Instruct-2507, # Sessions=82025.11 | 0.94 | |
| RadixCacheModel=Qwen3-4B-Instruct-2507, # Sessions=322025.11 | 0.97 | |
| LMCacheModel=Qwen3-4B-Instruct-2507, # Sessions=162025.11 | 1.19 | |
| LMCacheModel=Qwen3-4B-Instruct-2507, # Sessions=322025.11 | 1.72 |