Human Correlation Analysis on Original human judgment dataset
0.643Generation PerplexityGeneration Perplexity
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Generation PerplexityHuman Judgment Criterion=Interesting/BT2023.05 | 0.643 | — | — | — | — | — | — | |
| Generation PerplexityHuman Judgment Criterion=Sensible/BT2023.05 | 0.738 | — | — | — | — | — | — | |
| Generation PerplexityHuman Judgment Criterion=Human-like/BT2023.05 | 0.81 | — | — | — | — | — | — | |
| Distinct-4Human Judgment Criterion=Human-like/BT2023.05 | — | — | — | 0.738 | — | — | — | |
| Distinct-4Human Judgment Criterion=Interesting/BT2023.05 | — | — | — | 0.524 | — | — | — | |
| Distinct-4Human Judgment Criterion=Sensible/BT2023.05 | — | — | — | 0.595 | — | — | — | |
| MAUVEHuman Judgment Criterion=Human-like/BT2023.05 | — | — | — | — | — | — | 0.952 | |
| MAUVEHuman Judgment Criterion=Interesting/BT2023.05 | — | — | — | — | — | — | 0.81 | |
| MAUVEHuman Judgment Criterion=Sensible/BT2023.05 | — | — | — | — | — | — | 0.857 | |
| RepetitionHuman Judgment Criterion=Human-like/BT2023.05 | — | — | -0.167 | — | — | — | — | |
| RepetitionHuman Judgment Criterion=Interesting/BT2023.05 | — | — | -0.143 | — | — | — | — | |
| RepetitionHuman Judgment Criterion=Sensible/BT2023.05 | — | — | -0.071 | — | — | — | — | |
| Self-BLEUHuman Judgment Criterion=Human-like/BT2023.05 | — | — | — | — | 0.595 | — | — | |
| Self-BLEUHuman Judgment Criterion=Interesting/BT2023.05 | — | — | — | — | 0.405 | — | — | |
| Self-BLEUHuman Judgment Criterion=Sensible/BT2023.05 | — | — | — | — | 0.524 | — | — | |
| SOHuman Judgment Criterion=Human-like/BT2023.05 | — | — | — | — | — | 0.881 | — | |
| SOHuman Judgment Criterion=Interesting/BT2023.05 | — | — | — | — | — | 0.762 | — | |
| SOHuman Judgment Criterion=Sensible/BT2023.05 | — | — | — | — | — | 0.786 | — | |
| Zipf CoefficientHuman Judgment Criterion=Human-like/BT2023.05 | — | 0.833 | — | — | — | — | — | |
| Zipf CoefficientHuman Judgment Criterion=Interesting/BT2023.05 | — | 0.524 | — | — | — | — | — | |
| Zipf CoefficientHuman Judgment Criterion=Sensible/BT2023.05 | — | 0.667 | — | — | — | — | — |