Summarization on CNN/DailyMail (ROUGE1, ROUGE2, ROUGE-L, m, FC)
7.49Metric mTopK
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TopKTarget Model=Llama-3.1-8B, Draft Model=Llama-3.2-1B2025.09 | 7.49 | 26.8 | 10.5 | 22 | 62.5 | |
| SelfJudge-FTarget Model=Llama-3.1-8B, Draft Model=Llama-3.2-1B2025.09 | 6.05 | 26.2 | 10.3 | 21.5 | 63.9 | |
| SelfJudge-RTarget Model=Llama-3.1-8B, Draft Model=Llama-3.2-1B2025.09 | 4.92 | 27.5 | 11 | 22.8 | 65.5 | |
| AutoJudge-FTarget Model=Llama-3.1-8B, Draft Model=Llama-3.2-1B2025.09 | 4.74 | 27.1 | 10.9 | 22.5 | 64.5 | |
| AutoJudge-RTarget Model=Llama-3.1-8B, Draft Model=Llama-3.2-1B2025.09 | 4.36 | 27 | 10.8 | 22.4 | 64.5 | |
| SDTarget Model=Llama-3.1-8B, Draft Model=Llama-3.2-1B2025.09 | 4.35 | 27.1 | 10.8 | 22.4 | 64.6 |