Long Context Modeling on LongBench-E
23.92S. QA AccuracyDeepSeek V2-Lite (MLA)
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V2-Lite (MLA)Context Length=64K, Hardware=H200 GPUs2026.04 | 23.92 | 11.15 | 17.52 | 63.19 | 3.09 | 58.17 | 29.51 | 3.2 | |
| LCAContext Length=64K, Hardware=H200 GPUs2026.04 | 22.61 | 8.9 | 23.74 | 58.27 | 2.33 | 58.67 | 29.09 | 1.8 | |
| KDAContext Length=64K, Hardware=H200 GPUs2026.04 | 18.81 | 11.68 | 14.22 | 56.97 | 3.25 | 57.97 | 27.15 | 2.47 | |
| MInferenceContext Length=64K, Hardware=H200 GPUs2026.04 | 14.54 | 5.61 | 13.8 | 41.15 | 2.17 | 41.01 | 19.71 | 2.99 | |
| FlexPrefillContext Length=64K, Hardware=H200 GPUs2026.04 | 14.35 | 5.94 | 14.1 | 48 | 2.22 | 41.72 | 21.05 | 2.51 |