Toy deep learning library written in C++ with support for CUDA.
To get started clone this repository.
git clone https://github.com/adajed/DeepLearningCPP.git
- CUDA 9.0
- g++ 7.3.0
Additional prerequisites for development:
- clang-format 6.0.0
- clang-tidy 6.0.0
To compile just run make from the main directory:
make -j4
This will compile everything: library, tests and samples.
./build/test
Or you can ran a debug version:
./build/test_debug
Source code for samples could be found in samples directory. To run them you have to add compiled library to paths:
export LD_LIBRARY_PATH=/path/to/repository/build:$LD_LIBRARY_PATH
Example for running mnist sample:
./build/sample_mnist
- Adam Jędrych - adajed
This project is licensed under the MIT License - see the LICENSE.md file for details