The dataset consists of 16,125 rows of input, instruction, and output packed into a train set.
Each schema has its own CSV file… See the full description on the dataset page:
https://huggingface.co/datasets/Highgroundbkk/Pangpuriye-generated_by_typhoon.