Text Generation on IMDB reviews (test)
78.3GrammaticalityReal samples
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Real samplesOpponent=LM, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 78.3 | 72 | 73.3 | |
| Real samplesOpponent=MaskGAN, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 65.7 | 59.3 | 62.3 | |
| MaskGANOpponent=LM, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 59.7 | 58.3 | 58 | |
| MaskGANOpponent=MaskMLE, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 49.7 | 43.7 | 44.3 | |
| MaskMLEOpponent=LM, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 42.7 | 43.7 | 40.3 | |
| LMOpponent=MaskMLE, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 20 | 28.3 | 21.7 | |
| MaskMLEOpponent=MaskGAN, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 18.7 | 20.3 | 18.3 | |
| MaskGANOpponent=Real samples, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 18 | 20 | 16.7 | |
| LMOpponent=MaskGAN, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 15.3 | 19.7 | 15.7 | |
| LMOpponent=Real samples, Sample length=40 words, Evaluation protocol=Mechanical Turk blind heads-up evaluation2018.01 | 6.7 | 7 | 6.3 |