Summarization on MultiNews (Rouge-L)
23.66Rouge-LS3-Pure
Evaluation results
| Method | Links | |
|---|---|---|
| S3-PureModel=Llama-3.1-8B-Instruct2026.01 | 23.66 | |
| RAG-DenseModel=Llama-3.1-8B-Instruct2026.01 | 23.66 | |
| FullKVModel=Mistral-7B-Instruct-v0.32026.01 | 23.52 | |
| RAG-RerankModel=Llama-3.1-8B-Instruct2026.01 | 23.42 | |
| HyDE-RerankModel=Llama-3.1-8B-Instruct2026.01 | 23.42 | |
| FullKVModel=Llama-3.1-8B-Instruct2026.01 | 23.4 | |
| HyDE-RerankModel=Mistral-7B-Instruct-v0.32026.01 | 23.38 | |
| S3-HybridModel=Llama-3.1-8B-Instruct2026.01 | 23.38 | |
| BM25Model=Llama-3.1-8B-Instruct2026.01 | 23.38 | |
| S3-HybridModel=Mistral-7B-Instruct-v0.32026.01 | 23.37 | |
| BM25Model=Mistral-7B-Instruct-v0.32026.01 | 23.37 | |
| RAG-RerankModel=Mistral-7B-Instruct-v0.32026.01 | 23.32 | |
| S3-PureModel=Mistral-7B-Instruct-v0.32026.01 | 22.68 | |
| RAG-DenseModel=Mistral-7B-Instruct-v0.32026.01 | 22.68 | |
| FullKVModel=Qwen2-7B-Instruct2026.01 | 21.73 | |
| RAG-RerankModel=Qwen2-7B-Instruct2026.01 | 21.7 | |
| HyDE-RerankModel=Qwen2-7B-Instruct2026.01 | 21.7 | |
| S3-HybridModel=Qwen2-7B-Instruct2026.01 | 21.66 | |
| S3-PureModel=Qwen2-7B-Instruct2026.01 | 21.66 | |
| RAG-DenseModel=Qwen2-7B-Instruct2026.01 | 21.66 | |
| BM25Model=Qwen2-7B-Instruct2026.01 | 21.66 |