Long-context Question Answering on MuSiQue Fixed Chunk 2048
39.46ScoreBaseline
Evaluation Results
| Method | Links | |
|---|---|---|
| BaselineModel=ChatGLM, Setting=Fixed Chunk (2048)2026.03 | 39.46 | |
| Our + ReorderModel=ChatGLM, Setting=Fixed Chunk (2048)2026.03 | 37.86 | |
| BaselineModel=Qwen, Setting=Fixed Chunk (2048)2026.03 | 37.18 | |
| OurModel=ChatGLM, Setting=Fixed Chunk (2048)2026.03 | 36.88 | |
| OurModel=Qwen, Setting=Fixed Chunk (2048)2026.03 | 33.84 | |
| BaselineModel=LLaMA, Setting=Fixed Chunk (2048)2026.03 | 32.85 | |
| OurModel=LLaMA, Setting=Fixed Chunk (2048)2026.03 | 31.04 | |
| Our + ReorderModel=LLaMA, Setting=Fixed Chunk (2048)2026.03 | 30.44 | |
| EPIC (15%)Model=ChatGLM, Setting=Fixed Chunk (2048)2026.03 | 29.02 | |
| Our + ReorderModel=Qwen, Setting=Fixed Chunk (2048)2026.03 | 28.72 | |
| No RecomputeModel=ChatGLM, Setting=Fixed Chunk (2048)2026.03 | 28.33 | |
| CacheBlendModel=LLaMA, Setting=Fixed Chunk (2048)2026.03 | 28.23 | |
| EPIC (15%)Model=LLaMA, Setting=Fixed Chunk (2048)2026.03 | 26.38 | |
| CacheBlendModel=ChatGLM, Setting=Fixed Chunk (2048)2026.03 | 26.24 | |
| CacheBlendModel=Qwen, Setting=Fixed Chunk (2048)2026.03 | 26.11 | |
| No RecomputeModel=LLaMA, Setting=Fixed Chunk (2048)2026.03 | 25.23 | |
| EPIC (15%)Model=Qwen, Setting=Fixed Chunk (2048)2026.03 | 23.68 | |
| No RecomputeModel=Qwen, Setting=Fixed Chunk (2048)2026.03 | 13.42 |