Long-form Generation on LongGenBench
80.03CRFull
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FullBackbone=Llama-3.1-8B-Instruct, KV Cache Management Method=Full, KV Budget/Sparsity=Full Cache2025.05 | 80.03 | 26.82 | |
| FullBackbone=Qwen-2.5-7B-Instruct, KV Cache Management Method=Full, KV Budget/Sparsity=Full Cache2025.05 | 79.56 | 31.09 | |
| ShadowKVBackbone=Llama-3.1-8B-Instruct, KV Cache Management Method=ShadowKV, KV Budget/Sparsity=20482025.05 | 79.28 | 30.66 | |
| QuestBackbone=Llama-3.1-8B-Instruct, KV Cache Management Method=Quest, KV Budget/Sparsity=20482025.05 | 78.03 | 27.71 | |
| FreeKVBackbone=Llama-3.1-8B-Instruct, KV Cache Management Method=FreeKV, KV Budget/Sparsity=20482025.05 | 78.03 | 27.62 | |
| RaaSBackbone=Qwen-2.5-7B-Instruct, KV Cache Management Method=RaaS, KV Budget/Sparsity=20482025.05 | 77.65 | 34.4 | |
| FreeKVBackbone=Qwen-2.5-7B-Instruct, KV Cache Management Method=FreeKV, KV Budget/Sparsity=20482025.05 | 76.93 | 32.81 | |
| InfiniGenBackbone=Llama-3.1-8B-Instruct, KV Cache Management Method=InfiniGen, KV Budget/Sparsity=20482025.05 | 76.68 | 26.21 | |
| RaaSBackbone=Llama-3.1-8B-Instruct, KV Cache Management Method=RaaS, KV Budget/Sparsity=20482025.05 | 76.63 | 26 | |
| ArkValeBackbone=Qwen-2.5-7B-Instruct, KV Cache Management Method=ArkVale, KV Budget/Sparsity=20482025.05 | 75.91 | 31.79 | |
| InfiniGenBackbone=Qwen-2.5-7B-Instruct, KV Cache Management Method=InfiniGen, KV Budget/Sparsity=20482025.05 | 72.96 | 27.67 | |
| FullBackbone=Qwen-2.5-14B-Instruct, KV Cache Management Method=Full, KV Budget/Sparsity=Full Cache2025.05 | 65.84 | 29.35 | |
| FreeKVBackbone=Qwen-2.5-14B-Instruct, KV Cache Management Method=FreeKV, KV Budget/Sparsity=20482025.05 | 65.46 | 29.39 | |
| InfiniGenBackbone=Qwen-2.5-14B-Instruct, KV Cache Management Method=InfiniGen, KV Budget/Sparsity=20482025.05 | 63.02 | 26.08 | |
| QuestBackbone=Qwen-2.5-7B-Instruct, KV Cache Management Method=Quest, KV Budget/Sparsity=20482025.05 | 62.89 | 25.96 | |
| RaaSBackbone=Qwen-2.5-14B-Instruct, KV Cache Management Method=RaaS, KV Budget/Sparsity=20482025.05 | 62.29 | 29.79 | |
| QuestBackbone=Qwen-2.5-14B-Instruct, KV Cache Management Method=Quest, KV Budget/Sparsity=20482025.05 | 45.49 | 19.76 | |
| ArkValeBackbone=Qwen-2.5-14B-Instruct, KV Cache Management Method=ArkVale, KV Budget/Sparsity=20482025.05 | 45.31 | 19.65 | |
| RazorBackbone=Qwen-2.5-7B-Instruct, KV Cache Management Method=Razor, KV Budget/Sparsity=0.152025.05 | 42.13 | 21.48 | |
| ArkValeBackbone=Llama-3.1-8B-Instruct, KV Cache Management Method=ArkVale, KV Budget/Sparsity=20482025.05 | 39.36 | 10.36 | |
| RazorBackbone=Llama-3.1-8B-Instruct, KV Cache Management Method=Razor, KV Budget/Sparsity=0.152025.05 | 35.9 | 12.2 | |
| ShadowKVBackbone=Qwen-2.5-7B-Instruct, KV Cache Management Method=ShadowKV, KV Budget/Sparsity=20482025.05 | 35.49 | 11.43 | |
| RazorBackbone=Qwen-2.5-14B-Instruct, KV Cache Management Method=Razor, KV Budget/Sparsity=0.152025.05 | 26.48 | 13.89 | |
| ShadowKVBackbone=Qwen-2.5-14B-Instruct, KV Cache Management Method=ShadowKV, KV Budget/Sparsity=20482025.05 | 21.25 | 8.25 |