Data Leakage Prevention

Split data by entity (company), not by page, so the model can’t memorize company-specific language and appear to generalize. Once you’ve looked at test-set results, that test set can no longer give an honest evaluation.

Core Idea

(To be expanded)

Key Principles

(To be added)

Examples

(To be added)

Connections

(To be added)

Source

Extracted from How To Fine-Tune a Small LLM on Your Own Data (Full Guide)