A synthetic dataset of 24,000 LLM question/instruction prompts distributed across 100 top-level categories and 2,000 subcategories, with 4 difficulty levels. Built on top of:

categories-100-en subcategories-2000-en

	Structure

Dimension Breakdown

Parent categories 100

Subcategories 2,000 (20 per category)

Difficulty levels 4 (0, 1, 2, 3)

Prompts per (subcategory, difficulty) 3

Total prompts 24,000

(This was the instructed… See the full description on the dataset page: https://huggingface.co/datasets/RowRed/prompts-24000-en.