Model Size
Domain: LLM Concepts
🎯 Motivation
The number of parameters determines model size. Bigger models can learn more but need more memory, compute, and money to run.
📋 Concrete Example
A 7B model has 7 billion parameters, needs ~14GB RAM. A 70B model needs ~140GB. Bigger models are smarter but slower and more expensive to run.
🔗 Analogy
A library - a small library (small model) has fewer books but is quick to search. A massive library (large model) has more knowledge but takes longer to find the right answer and costs more to maintain.