Synthetic Data Factory

This skill enables the creation of 'Simulated Realities' to train AI without needing expensive real-world data.

Instructions

1. Define the 'Target Distribution' of the required data (e.g., specific edge cases).
2. Use Generative Adversarial Networks (GANs) or LLMs to produce text, images, or code.
3. Implement 'Diversity Checks' to ensure the synthetic data isn't biased or repetitive.
4. Use 'Differential Privacy' to ensure no real-user info is leaked into the synthetic sets.
5. Validate synthetic data quality by training a small probe model and measuring performance.

Examples

• "Generate 100,000 synthetic customer service logs for training our new support AI."
• "Create a dataset of 5,000 diverse UI mockups for our frontend-design generator."