Human fluency, entity-relevance, and faithfulness scores of entity-controlled models with the reference entities as the input. Faithful. denotes the percentage of generated summaries that are faithful. The Krippendorf’s α inter-rater agreement for all columns are 0.68, 0.77, and 0.56.
| Method . | Fluency . | Entity-rel. . | Faithful. . |
|---|---|---|---|
| SD2 | 4.83 | 3.63 | 70% |
| D.GPT2 | 4.65 | 3.33 | 68% |
| D.GPT2+CMDP | 4.83 | 3.92 | 71% |
| Method . | Fluency . | Entity-rel. . | Faithful. . |
|---|---|---|---|
| SD2 | 4.83 | 3.63 | 70% |
| D.GPT2 | 4.65 | 3.33 | 68% |
| D.GPT2+CMDP | 4.83 | 3.92 | 71% |