If the files aren't already built, this will build them as well.
$ make runBuilds the specified file in the build directory.
$ make buildBuilds all files in the build directory.
$ make build_allRemoves all files in the build directory.
$ make clean- Stochastic Gradient Descent (SGD): Implemented for efficient training.
- Momentum: Added to accelerate SGD in the relevant direction and dampen oscillations.
- Learning Rate Decay: Adjusts the learning rate over time to improve convergence.
- HogBatching: Implemented HOGWILD with mini-batching (HogBatching) for efficient parallel updates. For more details, refer to this paper.
#include "gym.h"
typedef double T;
// Function to generate training data
void generateData(nn::DataSet &ds)
{
SetTraceLogLevel(LOG_NONE); // Use this for a cleaner console output
const std::vector<std::vector<T>> t = {
{1, 1, 0},
{1, 0, 1},
{0, 1, 1},
{0, 0, 0},
};
// Iterate through the data and add it to the dataset
for (const auto &x : t)
{
std::vector<T> input{x[0], x[1]}; // Create input vector
std::vector<T> output{x[2]}; // Create output vector
ds.addData(input, output); // Add the input-output pair to the dataset
}
}
int main()
{
// Define the architecture of the neural network
// And set the number of threads for parallel processing
nn::NeuralNetwork n{{2, 2, 1}, 5};
// Set hyperparameters for the neural network or leave them by default
n.setHyperparameters(1, 0.001, 0.9, 1);
// Create a dataset and generate training data
nn::DataSet ds;
generateData(ds);
// Create a Gym instance and train the neural network
nn::Gym gym{n};
gym.train(ds, 50'000);
// Create an input matrix and perform a forward pass
nn::Matrix input{{0, 1}};
nn::Matrix output = n.forward(input);
// Print the output matrix
output.printX();
}transitionVideo.mp4
76 160 166 166 204 191 166 33
117 242 254 254 254 254 206 215 254 247 128
70 248 254 198 64 27 27 6 10 27 168 245 129
6 195 253 128 15 10 236 189
146 254 166 115 212 8
215 246 41 63 252 201
215 232 34 228 254 139
215 246 68 94 227 254 212 10
194 254 248 176 148 148 148 222 251 254 254 135
16 176 254 254 254 254 254 233 140 217 254 51
4 76 75 133 114 44 24 184 254 9
184 254 47
184 254 98
184 254 98
184 254 75
184 254 9
16 245 213 4
60 254 174
108 254 102
108 218
1
1 1 2 2
1 1 1 1 7 13 6
1 1 1 1 1 8 62 164 190 187 192 195 139 22 1
5 124 249 251 241 217 198 207 231 242 240 195 3
74 252 253 216 57 14 7 8 16 65 212 242 77 1
2 226 248 125 2 38 182 67 15 8 4 1
132 244 177 3 1 77 161 42 17 15 10 4
12 238 240 63 2 45 226 150 6 1 1 1 1
30 234 224 37 1 5 22 76 220 249 68
4 194 221 84 12 9 21 53 98 134 186 248 248 8
134 233 198 136 132 155 171 171 162 199 249 229
54 208 213 203 201 196 178 140 106 170 246 145
3 21 76 141 165 150 102 50 33 131 238 49
11 39 38 19 7 9 126 230 15
1 2 1 1 5 162 225 6
6 209 221 2
14 235 216 1
40 244 206
86 247 188
109 246 157
70 238 106
11 192 46
43 9
1 + 3 = 4.00486
6 + 4 = 10.0266
1 + 4 = 5.00045
6 + 6 = 12.0141
2 + 0 = 2.03343
1 + 1 = 2.03916
1 + 3 = 4.00486
1 + 3 = 4.00486
3 + 6 = 9.0138
3 + 4 = 6.9895
7 + 4 = 10.9675
6 + 1 = 7.0752
4 + 6 = 10.0099
5 + 4 = 9.01112
6 + 7 = 13.001
7 + 6 = 12.9788
3 + 2 = 5.03603
5 + 0 = 5.06653
3 + 3 = 5.98779
3 + 6 = 9.0138
1.000000 ^ 1.000000 = 0.007053
1.000000 ^ 0.000000 = 0.993426
0.000000 ^ 1.000000 = 0.993421
0.000000 ^ 0.000000 = 0.007556