Summarization on SAMSum LongBench (ROUGE-L)
43.57ROUGE-LBaseline
Evaluation Results
| Method | Links | |
|---|---|---|
| BaselineModel=Llama 3.1-8B2026.02 | 43.57 | |
| InnerQBaseModel=Llama 3.1-8B2026.02 | 43.38 | |
| InnerQSmallModel=Llama 3.1-8B2026.02 | 43.12 | |
| KIVISinkModel=Llama 3.1-8B2026.02 | 43.09 | |
| InnerQHybridModel=Llama 3.1-8B2026.02 | 42.82 | |
| KIVISinkModel=Llama 2-13B2026.02 | 42.57 | |
| BaselineModel=Llama 2-13B2026.02 | 42.55 | |
| KIVIModel=Llama 3.1-8B2026.02 | 42.33 | |
| InnerQBaseModel=Llama 2-13B2026.02 | 42.16 | |
| KIVIModel=Llama 2-13B2026.02 | 42.12 | |
| InnerQHybridModel=Llama 2-13B2026.02 | 41.77 | |
| BaselineModel=Llama 2-7B2026.02 | 41.36 | |
| InnerQBaseModel=Llama 2-7B2026.02 | 40.62 | |
| InnerQSmallModel=Llama 2-13B2026.02 | 40.54 | |
| KIVISinkModel=Llama 2-7B2026.02 | 40.44 | |
| KIVIModel=Llama 2-7B2026.02 | 40.37 | |
| InnerQHybridModel=Llama 2-7B2026.02 | 40.16 | |
| InnerQSmallModel=Llama 2-7B2026.02 | 14.77 | |
| KIVISinkModel=Llama 3.2-3B2026.02 | 7.48 | |
| KIVIModel=Llama 3.2-3B2026.02 | 7.16 | |
| InnerQSmallModel=Llama 3.2-3B2026.02 | 7.13 | |
| InnerQHybridModel=Llama 3.2-3B2026.02 | 7.06 | |
| InnerQBaseModel=Llama 3.2-3B2026.02 | 7 | |
| KIVIModel=Llama 3.2-1B2026.02 | 6.83 | |
| BaselineModel=Llama 3.2-3B2026.02 | 6.7 | |
| InnerQBaseModel=Llama 3.2-1B2026.02 | 6.57 | |
| InnerQSmallModel=Llama 3.2-1B2026.02 | 6.25 | |
| InnerQHybridModel=Llama 3.2-1B2026.02 | 6.18 | |
| KIVISinkModel=Llama 3.2-1B2026.02 | 6.11 | |
| BaselineModel=Llama 3.2-1B2026.02 | 6.04 |