CohereLabs/aya_dataset

Dataset Summary The Aya Dataset is a multilingual instruction fine-tuning dataset curated by an open-science community via Aya Annotation Platform from Cohere Labs. The dataset contains a total of 204k human-annotated prompt-completion pairs along with the demographics data of the annotators. This dataset can be used to train, finetune, and evaluate multilingual LLMs. Curated by: Contributors of Aya Open Science Intiative. Language(s): 65 languages (71 including dialects &… See the full description on the dataset page: https://huggingface.co/datasets/CohereLabs/aya_dataset.