The columns available in the datasets typically conform to: Instruction, Targets, Task, Language, Data Source, ID. You are able to view samples of the datasets at the Huggingface links. More details for this datasets (meta data and data sheets) are available at these links as well.

Here are the datasets

• Sentiment Analysis: Train, Test
• AfriXNLI: Train, Test
• Machine translation: Train, Test

Starter notebook: https://github.com/Lelapa-AI/zindi-inkuba-notebook