Story Completion on MTG Story Completion (test)
31.2ROUGE-L (ES)tgt-dev
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| tgt-devModel Selection Criterion=Target language development set performance (Upper bound)2023.05 | 31.2 | 30.5 | 35.6 | 28.1 | 0 | |
| cos-simModel Selection Criterion=XLRS-based model selection (Similarity of representations between English and target language)2023.05 | 30.8 | 30.3 | 35.6 | 26.4 | -0.58 | |
| en-devModel Selection Criterion=English development set performance2023.05 | 28.9 | 27.8 | 28.9 | 20.3 | -4.88 |