Skip to main content
Fulltext search
Example 2: Predictions and attention are in spans
ch1
Example 30: Condition number predicts sensitivity
Example 31: Sparse adjacency matvec (pure NumPy COO)
Example 32: Attention block computation (QK^T then AV)
Example 33: Vector space closure + centering as a subspace projection
Example 34: Predictions and attention are in spans
Example 35: One-hot basis vectors and coordinate representations
Example 36: Linear layers and backprop are linear maps + adjoints
Example 37: Dot products, norms, and cosine similarity (retrieval)
Example 38: Least squares residual is orthogonal to column space
Example 39: Null space explains non-identifiability (overparameterized linear model)
Example 3: One-hot basis vectors and coordinate representations
Example 40: Power iteration for dominant eigenvector (PCA direction)
Example 41: PSD in ML: covariance and kernel Gram matrices
Example 42: SVD and conditioning: why normal equations are risky
Example 43: PCA bookkeeping: EVR and reconstruction error
Example 44: Least squares: lstsq vs normal equations
Example 45: Solving systems: Cholesky factor reuse
Example 46: Conditioning: small perturbations â big solution changes
Pagination
First page
Previous page
…
Page
4
Page
5
Page
6
Page
7
Page
8
Page
9
Page
10
Page
11
Page
12
…
Next page
Last page
