Conditional Text Generation on ROCStories
36.2Grammaticality Win RateDM-LSTM
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DM-LSTMComparison Baseline=PRPN2022.03 | 36.2 | 14.5 | 49.3 | 0.225 | 56.5 | 17.5 | 26 | 0.306 | |
| DM-LSTMComparison Baseline=AWD-LSTM2022.03 | 28 | 14.5 | 57.5 | 0.224 | 43 | 34.5 | 22.5 | 0.214 | |
| DM-GPT-2Comparison Baseline=GPT-22022.03 | 20.4 | 5 | 74.6 | 0.374 | 50.6 | 18.8 | 30.6 | 0.224 | |
| DM-TransformerComparison Baseline=Transformer2022.03 | 18.2 | 5.2 | 76.6 | 0.358 | 50.6 | 18.6 | 30.8 | 0.342 | |
| DM-LSTMComparison Baseline=ON-LSTM2022.03 | 12.8 | 6.4 | 80.8 | 0.238 | 48.4 | 24.4 | 27.2 | 0.409 |