- I’m sure the implementation could be improved, but it would probably take improvements in TensorFlow itself as well to get it to operate at the same speed as other implementations.
- This may require require a little bit more hyperparameter tuning to get nice results.
- As of now, it seems to be around 3x slower than implementations using Torch.
- With certain images or output sizes, you might need some hyperparameter tuning (especially –content-weight , –style-weight , and –learning-rate ).
- The implementation is a lot simpler than a lot of the other ones out there, thanks to TensorFlow’s really nice API and automatic differentiation .
Read the full article, click here.
@random_forests: “Machine Learning meets art in #TensorFlow. Beautiful! Works out of the box, too.”
neural-style – Neural style in TensorFlow! • http://www.anishathalye.com/2015/12/19/an-ai-that-can-mimic-any-artist/ • https://git.io/style