Core Concepts:

Knowledge Generation: Encourages models to gather and synthesize information from their training data before generating responses.
Knowledge Integration: Update your model with recent data beyond its training cut-off. Use input files and copy-pasta text.