11  Notation

Symbol	Meaning
\(x\)	One input example or token
\(X\)	A batch, design matrix, or sequence representation
\(y\)	Observed target
\(\hat{y}\)	Model prediction
$ heta$	Trainable parameters
\(L( heta)\)	Empirical objective
\(\eta\)	Learning rate
\(Q,K,V\)	Query, key, and value tensors
\(h_t\)	Hidden state at sequence position \(t\)

Each module should extend this table only when new notation is required.