search

richzhang / colorization-pytorch

PyTorch reimplementation of Interactive Deep Colorization
https://richzhang.github.io/ideepcolor/
MIT License
598 stars 114 forks source link

Model training Loss graph #3

Open Ugness opened 6 years ago

Ugness commented 6 years ago

newplot

Plot.ly Link: https://plot.ly/~Ugness/1/

I am training the model with ILSVRC2012 training set and same options as your implementation and my loss graph looks like above.
I am afraid that my model's loss reduces correctly. Can you check this loss graph or share your loss graph?
Thanks.

richzhang commented 6 years ago

Thanks for the message. I'll look into it in the next week or so.

richzhang commented 6 years ago

My apologies for the delay. I'm caught up with a few things and will have to look into this more after CVPR deadline (mid-Nov).

Ugness commented 6 years ago

Thanks.

Samir55 commented 5 years ago

Hello author, I have the same problem above, what is the solution? Thanks in advance.

richzhang commented 5 years ago

I believe I did not do the initialization properly. I am looking into it. Thanks for your patience!

richzhang commented 5 years ago

The problem was with the initialization. In the original paper, I used k-means initialization, but didn't have the code for pytorch.

I just pushed a change. If you rerun python make_ilsvrc_dataset.py, it will make a directory dataset/ilsvrc2012/train_small/, which contains 10 imagenet classes. Then rerun bash scripts/train_siggraph.sh again. It will first run 100 epochs on this small training set as "initialization", before training on the full training set.

It does still train quite slowly. Better networks/loss functions would be a good update to the project.

Thanks for your patience.