Critique and Refinement Evaluation Dataset Summarization
89.6CURCO
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RCOBackbone=UltraCM-13B2025.06 | 89.6 | 7.66 | |
| Base modelBackbone=LLaMA-3-70B-Instruct2025.06 | 87.8 | 7.68 | |
| RCOBackbone=LLaMA-3-8B-Instruct2025.06 | 86 | 7.74 | |
| Base modelBackbone=Auto-J-13B2025.06 | 85.6 | 7.6 | |
| RCOBackbone=LLaMA-2-13B-Chat2025.06 | 85.3 | 7.6 | |
| DPCOBackbone=UltraCM-13B2025.06 | 85.1 | 7.61 | |
| RCOBackbone=Auto-J-13B2025.06 | 83.7 | 7.54 | |
| DPCOBackbone=Auto-J-13B2025.06 | 80.3 | 7.47 | |
| DPCOBackbone=LLaMA-2-13B-Chat2025.06 | 79.9 | 7.56 | |
| DPCOBackbone=LLaMA-3-8B-Instruct2025.06 | 79 | 7.3 | |
| Base modelBackbone=LLaMA-3-8B-Instruct2025.06 | 78.4 | 7.6 | |
| Self-refinement2025.06 | 77.6 | 7.46 | |
| RCOBackbone=LLaMA-2-7B-Chat2025.06 | 77.4 | 7.59 | |
| Base modelBackbone=LLaMA-2-13B-Chat2025.06 | 74.7 | 7.56 | |
| Base modelBackbone=UltraCM-13B2025.06 | 72.6 | 7.24 | |
| DPCOBackbone=LLaMA-2-7B-Chat2025.06 | 70.2 | 7.37 | |
| Base modelBackbone=LLaMA-2-70B-Chat2025.06 | 68.3 | 7.44 | |
| Base modelBackbone=LLaMA-2-7B-Chat2025.06 | 63.4 | 7.43 | |
| Aligner2025.06 | 50.2 | 6.64 | |
| Initial Answer2025.06 | 6.52 | — |