FlexNN 1
Fully connected neural network built from scratch with flexible n-layer design and multiple activations.
Loading...
Searching...
No Matches
detail.hpp
Go to the documentation of this file.
1
10#pragma once
11
13#include "None.hpp"
14#include "ReLU.hpp"
15#include "LeakyReLU.hpp"
16#include "Sigmoid.hpp"
17#include "Tanh.hpp"
18#include "Softmax.hpp"
Per-activation hyperparameters for FlexNN::Activations.
Leaky ReLU with configurable alpha (default 0.01).
Linear (no) activation — identity forward/backward.
ReLU activation — max(0, z) forward, (z>0) backward.
Sigmoid activation — 1/(1+exp(-Z)) with clamping.
Stable column-wise softmax.
Tanh activation — std::tanh forward, (1-A^2) backward.