Flexible Feature visualization on PyTorch, for research and art
MIT License
Updates:
AutoImageParam
now has a new arg batch_size
which defaults to 1
). This would lead to higher quality visualizations with a smaller number of iterations.PytorchVersionError
on torch 2.x
. Thanks @matthiasgeihs 🙌Interface change:
The input (layer_outputs
) in custom objective functions is not anymore a list of tensors of shape (c, h, w)
. It's now a list of tensors of shape (n, c, h, w)
where n
= batch size. (The same applies for other shapes of intermediate layer outputs (...) -> (n, ...)
)
This is an example of an old v/s new objective function:
Published by Mayukhdeb almost 2 years ago
Published by Mayukhdeb almost 3 years ago
v1.8
Published by Mayukhdeb over 3 years ago
Changes:
1.9.x
. (Thanks to @seba-eng)Published by Mayukhdeb over 3 years ago
Updates:
torch_dreams.dreamer.set_custom_normalization()
Bug fixes:
self.__array__()
for torch_dreams.masked_image_param
Published by Mayukhdeb over 3 years ago
changes:
Published by Mayukhdeb over 3 years ago
Changes:
custom_image_param
Published by Mayukhdeb over 3 years ago
Changes:
v1.8.x
[1, 3, height, width]
from [1, 3, height, width//2, 2]
Published by Mayukhdeb over 3 years ago
Changes:
torch.fft.irfftn
on torch v1.8.x
Published by Mayukhdeb over 3 years ago
Updates:
auto_image_param
instances can be saved as images with image_param.save('image.jpg')
dreamer.get_snapshot()
now in testingdeepcopy
insteadPublished by Mayukhdeb over 3 years ago
torchvision.transforms
torch.fft
v1.7
and v1,8