Cross-cultural recipe adaptation
Format of a recipe
{
'id': str,
'title': str,
'title_translated': Optional[str],
'ingredients': List[str],
'steps': List[str],
'dish': Optional[str],
}
Format of a matched_recipe
{
'source': recipe,
'targets': List[recipe]
}
Format of a data file: jsonl with one matched_recipe per line
| Set | direction | Scale | Link | Notes |
|---|---|---|---|---|
| Silver | cn2en | 82 | Download | 1:1 |
| Human | cn2en | 25 | Download | 1:1 |
| Silver | en2cn | 52 | Download | 1:1 |
| Human | en2cn | 41 | Download | 1:1 |
Chinese monolingual dataset (cleaned) and English monolingual dataset (raw)
Statistics of CulturalDataset
| Direction | Source | Target | Train | Valid | Test | Silver | Human |
|---|---|---|---|---|---|---|---|
| cn2en | 44,473 | 144,645 | 35,578 | 4,447 | 4,448 | 82 | 25 |
| en2cn | 43,767 | 120,674 | 35,013 | 4,377 | 4,377 | 52 | 41 |