Summarization on CNN/Daily Mail anonymized (test)
40.13ROUGE-1rnn-ext + RL
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| rnn-ext + RLparadigm=extractive, training=Reinforcement Learning2018.05 | 40.13 | 16.58 | 36.47 | |
| Paulus et al. (2018) (RL+ML)paradigm=abstractive, training=RL + ML2018.05 | 39.87 | 15.82 | 36.9 | |
| rnn-ext + abs + RL + rerankparadigm=abstractive, post-processing=rerank2018.05 | 39.66 | 15.85 | 37.34 | |
| Nallapati et al. (2017)paradigm=extractive2018.05 | 39.6 | 16.2 | 35.3 | |
| SUMMARUNNERAnonymized=true2018.07 | 39.6 | 16.2 | 35.3 | |
| ff-extparadigm=extractive, architecture=feed-forward2018.05 | 39.51 | 16.85 | 35.8 | |
| lead-3paradigm=extractive, source=Nallapati et al. (2017)2018.05 | 39.2 | 15.7 | 35.5 | |
| LEAD3Anonymized=true2018.07 | 39.2 | 15.7 | 35.5 | |
| rnn-extparadigm=extractive, architecture=RNN2018.05 | 38.97 | 16.65 | 35.32 | |
| rnn-ext + abs + RLparadigm=abstractive, training=Reinforcement Learning2018.05 | 38.8 | 15.66 | 36.37 | |
| ff-ext + absparadigm=abstractive, extractor=feed-forward2018.05 | 38.73 | 15.7 | 36.33 | |
| Fan et al. (2017) (controlled)paradigm=abstractive2018.05 | 38.68 | 15.4 | 35.47 | |
| Paulus et al. (2018) (ML)paradigm=abstractive, training=Maximum Likelihood2018.05 | 38.3 | 14.81 | 35.49 | |
| rnn-ext + absparadigm=abstractive, extractor=RNN2018.05 | 37.58 | 14.68 | 35.24 | |
| Nallapati et al. (2016)paradigm=abstractive2018.05 | 35.46 | 13.3 | 32.65 |