Browse Datasets

Aya Dataset

Multilingual Instruction Dataset from Cohere

instruction-following multilingual fine-tuning 204k human annotations, millions translated Apache 2.0
Access Dataset

Description

Massive multilingual instruction fine-tuning dataset with human-annotated and translated instructions for many African languages.

Quick Start

Python 
from datasets import load_dataset

dataset = load_dataset("cohere/aya_dataset", split="train")

Project Ideas

Yoruba Assistant

Fine-tune a model to act as helpful assistant in Yoruba

Intermediate

Ethical Considerations

  • Participatory design mitigates some bias
  • Translation artifacts may remain