Summarization on PUMA (test)
30.16ROUGE-1 RecallPLASMA
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| PLASMAtuning=prefix-tuning2024.06 | 30.16 | 23.23 | 10.23 | 7.38 | 27.78 | 21.38 | 0.869 | 0.244 | 0.0405 | |
| FLAN-T5Fine-tuning approach=FDProm (Document and Structured Prompt)2024.06 | 25.9 | 21.22 | 8.28 | 6.5 | 23.8 | 20.82 | 0.852 | 0.217 | 0.034 | |
| FLAN-T5Fine-tuning approach=FDPer (Document and Perspective)2024.06 | 25.34 | 22.81 | 7.83 | 6.03 | 23.23 | 18.8 | 0.854 | 0.21 | 0.036 | |
| BARTFine-tuning approach=FDProm (Document and Structured Prompt)2024.06 | 24.17 | 22.69 | 8.74 | 6.66 | 22.09 | 20.65 | 0.867 | 0.201 | 0.038 | |
| GPT2Fine-tuning approach=FDPer (Document and Perspective)2024.06 | 21.49 | 20.54 | 6.44 | 5.72 | 19.87 | 20.55 | 0.855 | 0.134 | 0.03 | |
| PEGASUSFine-tuning approach=FDProm (Document and Structured Prompt)2024.06 | 20.32 | 18.52 | 4.77 | 4.53 | 17.72 | 16.17 | 0.849 | 0.167 | 0.025 | |
| BARTFine-tuning approach=FDPer (Document and Perspective)2024.06 | 20.18 | 21.26 | 6.79 | 6.92 | 18.35 | 19.34 | 0.865 | 0.171 | 0.032 | |
| T5Fine-tuning approach=FDPer (Document and Perspective)2024.06 | 19.88 | 22.73 | 6.04 | 6.1 | 20.2 | 19.49 | 0.86 | 0.172 | 0.03 | |
| GPT2Fine-tuning approach=FDProm (Document and Structured Prompt)2024.06 | 19.81 | 14.86 | 4.86 | 6.6 | 19.19 | 17.79 | 0.834 | 0.148 | 0.023 | |
| PEGASUSFine-tuning approach=FDPer (Document and Perspective)2024.06 | 19.65 | 19.23 | 5.44 | 5.55 | 17.31 | 17.01 | 0.851 | 0.159 | 0.027 | |
| T5Fine-tuning approach=FDProm (Document and Structured Prompt)2024.06 | 11.69 | 13.4 | 4.04 | 3.8 | 11.11 | 12.67 | 0.836 | 0.103 | 0.02 |