LLMs
Chinchilla Scaling
Compute-optimal training scales parameters and tokens roughly equally - about 20 tokens per parameter, with C ≈ 6ND.
Why interviewers ask about it
It optimises training compute only. Production models deliberately overtrain far past it (Llama-3-8B at ~1,900 tokens/param) because inference dominates lifetime cost.
This term is part of the free AI/ML Engineer interview preparation module - browse the full glossary for every definition.