sniklaus / softmax-splatting

an implementation of softmax splatting for differentiable forward warping using PyTorch
468 stars 58 forks source link

Some questions about the supplementary materials and the training details #30

Closed mmSir closed 3 years ago

mmSir commented 3 years ago

Here are two questions:

  1. Where can i get the supplementary materials? I haven't found it in the paper or your website...
  2. The second question is about the training details: It was mentioned in the paper, you trained portion of the Vimeo-90k dataset. I would like to know the approximate amount of data. Is 5000 triplets enough? Training is too slow on my poor GPU.

Looking forward to your reply!~

sniklaus commented 3 years ago

You can find the supplementary here. Reducing the number of samples to a mere 5000 is likely not a good idea. It would probably be better to just train for fever iterations in your case (why train on fewer samples if you can just do fewer iterations/epochs).

mmSir commented 3 years ago

Thanks for your reply!