Mathematical Reasoning on AIME24 Math (Performance %)
50Performance (%)RLKV
Evaluation Results
| Method | Links | |
|---|---|---|
| RLKVKV Cache Budget Sparsity=0.22025.10 | 50 | |
| RLKVKV Cache Budget Sparsity=0.42025.10 | 50 | |
| KVZipKV Cache Budget Sparsity=0.22025.10 | 46.7 | |
| RLKVKV Cache Budget Sparsity=0.22025.10 | 46.7 | |
| KVZipKV Cache Budget Sparsity=0.22025.10 | 43.3 | |
| R-KVKV Cache Budget Sparsity=0.22025.10 | 43.3 | |
| RLKVKV Cache Budget Sparsity=0.4, Backbone=Llama-3.1-8B-R12025.10 | 40 | |
| KVZipKV Cache Budget Sparsity=0.42025.10 | 40 | |
| RLKVKV Cache Budget Sparsity=0.2, Backbone=Llama-3.1-8B-R12025.10 | 36.7 | |
| DuoAttentionKV Cache Budget Sparsity=0.22025.10 | 36.7 | |
| DuoAttentionKV Cache Budget Sparsity=0.2, Backbone=Llama-3.1-8B-R12025.10 | 33.3 | |
| DuoAttentionKV Cache Budget Sparsity=0.42025.10 | 33.3 | |
| KVZipKV Cache Budget Sparsity=0.2, Backbone=Llama-3.1-8B-R12025.10 | 30 | |
| R-KVKV Cache Budget Sparsity=0.22025.10 | 30 | |
| RLKVKV Cache Budget Sparsity=0.62025.10 | 30 | |
| R-KVKV Cache Budget Sparsity=0.2, Backbone=Llama-3.1-8B-R12025.10 | 26.7 | |
| DuoAttentionKV Cache Budget Sparsity=0.22025.10 | 26.7 | |
| KVZipKV Cache Budget Sparsity=0.42025.10 | 26.7 | |
| RLKVKV Cache Budget Sparsity=0.42025.10 | 26.7 | |
| R-KVKV Cache Budget Sparsity=0.42025.10 | 26.7 | |
| DuoAttentionKV Cache Budget Sparsity=0.4, Backbone=Llama-3.1-8B-R12025.10 | 20 | |
| KVZipKV Cache Budget Sparsity=0.4, Backbone=Llama-3.1-8B-R12025.10 | 20 | |
| RLKVKV Cache Budget Sparsity=0.6, Backbone=Llama-3.1-8B-R12025.10 | 20 | |
| R-KVKV Cache Budget Sparsity=0.42025.10 | 20 | |
| KVZipKV Cache Budget Sparsity=0.62025.10 | 20 | |
| R-KVKV Cache Budget Sparsity=0.4, Backbone=Llama-3.1-8B-R12025.10 | 13.3 | |
| KVZipKV Cache Budget Sparsity=0.6, Backbone=Llama-3.1-8B-R12025.10 | 13.3 | |
| DuoAttentionKV Cache Budget Sparsity=0.42025.10 | 13.3 | |
| RLKVKV Cache Budget Sparsity=0.62025.10 | 13.3 | |
| R-KVKV Cache Budget Sparsity=0.6, Backbone=Llama-3.1-8B-R12025.10 | 10 | |
| R-KVKV Cache Budget Sparsity=0.62025.10 | 10 | |
| DuoAttentionKV Cache Budget Sparsity=0.6, Backbone=Llama-3.1-8B-R12025.10 | 6.7 | |
| H2OKV Cache Budget Sparsity=0.22025.10 | 6.7 | |
| R-KVKV Cache Budget Sparsity=0.62025.10 | 6.7 | |
| KVZipKV Cache Budget Sparsity=0.62025.10 | 6.7 | |
| H2OKV Cache Budget Sparsity=0.2, Backbone=Llama-3.1-8B-R12025.10 | 3.3 | |
| H2OKV Cache Budget Sparsity=0.4, Backbone=Llama-3.1-8B-R12025.10 | 3.3 | |
| KVZipKV Cache Budget Sparsity=0.8, Backbone=Llama-3.1-8B-R12025.10 | 3.3 | |
| H2OKV Cache Budget Sparsity=0.22025.10 | 3.3 | |
| DuoAttentionKV Cache Budget Sparsity=0.62025.10 | 3.3 | |
| H2OKV Cache Budget Sparsity=0.6, Backbone=Llama-3.1-8B-R12025.10 | 0 | |
| H2OKV Cache Budget Sparsity=0.8, Backbone=Llama-3.1-8B-R12025.10 | 0 | |
| R-KVKV Cache Budget Sparsity=0.8, Backbone=Llama-3.1-8B-R12025.10 | 0 | |
| DuoAttentionKV Cache Budget Sparsity=0.8, Backbone=Llama-3.1-8B-R12025.10 | 0 | |
| RLKVKV Cache Budget Sparsity=0.8, Backbone=Llama-3.1-8B-R12025.10 | 0 | |
| H2OKV Cache Budget Sparsity=0.42025.10 | 0 | |
| H2OKV Cache Budget Sparsity=0.62025.10 | 0 | |
| H2OKV Cache Budget Sparsity=0.82025.10 | 0 | |
| R-KVKV Cache Budget Sparsity=0.82025.10 | 0 | |
| DuoAttentionKV Cache Budget Sparsity=0.62025.10 | 0 | |
| DuoAttentionKV Cache Budget Sparsity=0.82025.10 | 0 | |
| KVZipKV Cache Budget Sparsity=0.82025.10 | 0 | |
| RLKVKV Cache Budget Sparsity=0.82025.10 | 0 | |
| H2OKV Cache Budget Sparsity=0.42025.10 | 0 | |
| H2OKV Cache Budget Sparsity=0.62025.10 | 0 | |
| H2OKV Cache Budget Sparsity=0.82025.10 | 0 | |
| R-KVKV Cache Budget Sparsity=0.82025.10 | 0 | |
| DuoAttentionKV Cache Budget Sparsity=0.82025.10 | 0 | |
| KVZipKV Cache Budget Sparsity=0.82025.10 | 0 | |
| RLKVKV Cache Budget Sparsity=0.82025.10 | 0 |