4 ms·
A Neural Network in 10 lines of C++ Code
- rhodysurf 8y agoWell thats a clickbait title... its nowhere near 10 lines
- poster123 8y agoIt compiles and runs fine with g++, but when I try Microsoft's cl.exe version 19.00 I get messages such as xneural.cpp(22): error C2398: Element '1': conversion from 'double' to 'float' requires a narrowing conversion referring to the code vector<float> X { 5.1, 3.5, 1.4, 0.2, 4.9, 3.0, 1.4, 0.2, 6.2, 3.4, 5.4, 2.3, 5.9, 3.0, 5.1, 1.8 }; If I replace <float> with <double> the code compiles and runs with cl.exe.
- AnimalMuppet 8y agoOr transform the numbers to floats, i.e., "5.1" -> "5.1f".
- asdsa5325 8y ago> The core component of the code, the learning algorithm, is only 10 lines AKA gradient descent in 10 lines of C++
- horsawlarway 8y agoA neural network in 10 lines of C++ code... Plus another 200 lines of C++ code. AKA: My "lines of code" count is super tiny if I don't include function definitions!