Verbose AI platform ready to hack and try new algorithms.
Main goal: write a tunable network manager where implementation of new algorithms is easy through writing own lambda functions. Project is writen in C++
- clone the github content to some directory
- install the datasets; currently three datasets are used in some ways in the programs: mnist, cifar-10-binary and stl-10
- create a directory at DATASET_DIR, create mnist, cifar-10-batches-bin and stl10_binary subdirectories. Extract the downloaded file content to these libraries
- in case of mnist, please rename:
- train-images-idx3-ubyte.gz -> train_images
- train-labels-idx1-ubyte.gz -> train_labels
- t10k-images-idx3-ubyte.gz -> test_images
- t10k-labels-idx1-ubyte.gz -> test_labels
- alternatively, edit the include/cmake_variables.hpp.in file, and provide explicitly the names of the files
- run cmake with a switch: cmake -DDATASET_DIR="" .
- run make or build the executables with cmake --build -DDATASET_DIR="" .