PyTorch's implementation of Neural Style Transfer. Learning repository with several experiments and playing around.
Run demo app with: python gradio_app.py
Run jupyter lab with: jupyter lab
and open neural-style-transfer.ipynb
Experiments done:
- Gram matrix values (Normalized vs Not-normalized)
- Number of iterations (Differences in quality)
- Impact of style losses in different layers
- Adam vs LBFG-S optimizer