Classes | |
| struct | Linear |
| Linear activation (identity). More... | |
| struct | ReLU |
| ReLU activation. More... | |
| struct | Sigmoid |
| Sigmoid activation. More... | |
| struct | Tanh |
| Tanh activation. More... | |
| class | Layer |
| Abstract layer interface. More... | |
| class | DenseLayer |
| Fully-connected layer with activation. More... | |
| class | BFGS |
| BFGS (Broyden–Fletcher–Goldfarb–Shanno) minimizer. More... | |
| class | FullBatchMinimizer |
| Base class for Full Batch Minimizers. More... | |
| class | GradientDescent |
| Standard Gradient Descent with optional Line Search. More... | |
| class | LBFGS |
| Limited-memory BFGS (L-BFGS) minimizer. More... | |
| class | Newton |
| Newton minimizer (full Newton) for unconstrained optimization. More... | |
| class | RingBuffer |
| A fixed-capacity ring buffer (circular buffer). More... | |
| class | StochasticGradientDescent |
| Stochastic Gradient Descent (SGD) minimizer. More... | |
| class | SLBFGS |
| Stochastic Limited-memory BFGS (S-LBFGS) minimizer. More... | |
| class | StochasticMinimizer |
| Base class for Stochastic Minimizers. More... | |
| class | Network |
| Simple feedforward network with contiguous parameter storage. More... | |
Typedefs | |
| template<typename M > | |
| using | DefaultSolverT = typename std::conditional< isSparse< M >, Eigen::ConjugateGradient< M >, Eigen::LDLT< M > >::type |
Functions | |
| template<typename V , typename BatchFn > | |
| V | finite_difference_hvp_batch (BatchFn &g, const V &weights, const std::vector< size_t > &indices, const V &v, double epsilon=1e-4) |
| template<typename V > | |
| V | lbfgs_two_loop (const RingBuffer< V > &s_list, const RingBuffer< V > &y_list, const RingBuffer< double > &rho_list, const V &v) |
Variables | |
| template<typename M > | |
| constexpr bool | isSparse = std::is_base_of_v<Eigen::SparseMatrixBase<M>, M> |
Typedef Documentation
◆ DefaultSolverT
template<typename M >
| using cpu_mlp::DefaultSolverT = typedef typename std::conditional<isSparse<M>, Eigen::ConjugateGradient<M>, Eigen::LDLT<M> >::type |
Function Documentation
◆ finite_difference_hvp_batch()
template<typename V , typename BatchFn >
| V cpu_mlp::finite_difference_hvp_batch | ( | BatchFn & | g, |
| const V & | weights, | ||
| const std::vector< size_t > & | indices, | ||
| const V & | v, | ||
| double | epsilon = 1e-4 |
||
| ) |
Here is the caller graph for this function:
◆ lbfgs_two_loop()
template<typename V >
| V cpu_mlp::lbfgs_two_loop | ( | const RingBuffer< V > & | s_list, |
| const RingBuffer< V > & | y_list, | ||
| const RingBuffer< double > & | rho_list, | ||
| const V & | v | ||
| ) |
Here is the call graph for this function:
Here is the caller graph for this function:
Variable Documentation
◆ isSparse
template<typename M >
|
constexpr |
