tiny pytorch implementation of neural style transfer.Keep it minimal and one click to run!
- install PyTorch,PIL,numpy
- get the code
git clone https://github.com/ggsonic/tiny-style-transfer
- one click to run
./run_train.sh
or
python -u tiny.py
- use torchvision pretrained models
- use l1_norm loss to make style weights and content weights in same unit