tom white

Results 38 issues of tom white

TL;DR: custom training is great! is there a good config or way to debug quality of result on small-ish datasets? --- I've managed to train my own custom models using...

This is a more radical followup to #12. This adds an option where all seed images come from disk instead of being computed on the fly. The assumption is that...

Thanks for sharing this code. Was curious if you had considered how to handle a regression task? I thought I might try a few ideas out, perhaps starting with a...

Added a version of decode_predictions specific to vggface2 that returns results in an identical format as the similarily named function in keras.applications.imagenet_utils Sample response: [('n001138', 'Bong_Joon-ho', 0.98794019), ('n009064', 'Yang_Yuanqing', 0.0016884505),...

--smiles can be used to encode smiles strings that are not in the dataset. The dataset is still used to do the charset one-hot encoding correctly. --save_json was also added...

This codebase works well for me and I'm able to replicate the current results. Having worked a good bit with other latent spaces, I'm curious to find out what other...

In looking at manifolds in my own experiments, I have noticed a consistent "dead zone" near the origin of the latent space. Here is an example generated with `faces/train_uncond_dcgan.py` and...

Unless I'm missing something, I think there was a typo in the mnist file names. eg: they didn't match the names in [the Theano tutorial](https://github.com/Newmu/Theano-Tutorials/blob/49de4c52f45ab7b05cea3dbad039270f816be70c/load.py#L16-L28).

FWIW: seems like we could drop the cv2 dependency and let keras do the preprocessing. Similarly, there is a handy utility that will also decode the imagenet predictions. I tried...

Another small tweak for your consideration.