Human Evaluation of Text Generation on RotoWire-FG (test)
3.63CoherenceBART
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| BART2025.11 | 3.63 | 4.27 | 3.87 | 3.37 | |
| HunterAug-BARTBackbone=BART2025.11 | 3.57 | 3.73 | 4.13 | 4.03 | |
| HunterAug-T5Backbone=T52025.11 | 3.4 | 3.57 | 4.1 | 3.97 | |
| T52025.11 | 3.2 | 4.2 | 3.47 | 3.17 | |
| AuxEncoder2025.11 | 2.97 | 3.67 | 3.3 | 2.93 |