
Before Training an LLM: From Design Dials to a Frozen Shape
Hyperparameters, vocabulary, and why a 7B model is 7B before day one, how the dials connect, where parameters live, and why the context window is a budgeted choice.

Hyperparameters, vocabulary, and why a 7B model is 7B before day one, how the dials connect, where parameters live, and why the context window is a budgeted choice.

RAG is runtime context construction at query time, not a storage layer you bolt onto an LLM.