metadata
license: unknown
Dataset Card for Flickr 30k Romanian Captions
Dataset Summary
This dataset is a translation in romanian of the flickr 30k captions dataset. This was generated using nllb-200-distilled-1.3B, with Hugging face for both tokenization and translation. Observations:
- the translation keeps the context pretty well.
- there are a few grammatical errors: "Doi tineri sare peste un balustradă"
- some translations are silly: "Un bărbat ţine o jucărie mare de leu împăiat.", "Un bărbat cu barbă care poartă un dulap."
Languages
romanian