ConceptAI & ML
Data-constrained scaling
The principle that when total unique data is limited, models can be trained for multiple epochs over the same data without significant performance loss.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Language model1Training epochs1
Lecturers
1Connections
2The lectures that define it or assume it, and the concepts next to it.